National Academies Press: OpenBook

Advanced Ground Vehicle Technologies for Airside Operations (2020)

Chapter: Chapter 6 - Detailed Evaluation Results

« Previous: Chapter 5 - Evaluation Process
Page 68
Suggested Citation:"Chapter 6 - Detailed Evaluation Results." National Academies of Sciences, Engineering, and Medicine. 2020. Advanced Ground Vehicle Technologies for Airside Operations. Washington, DC: The National Academies Press. doi: 10.17226/26017.
×
Page 68
Page 69
Suggested Citation:"Chapter 6 - Detailed Evaluation Results." National Academies of Sciences, Engineering, and Medicine. 2020. Advanced Ground Vehicle Technologies for Airside Operations. Washington, DC: The National Academies Press. doi: 10.17226/26017.
×
Page 69
Page 70
Suggested Citation:"Chapter 6 - Detailed Evaluation Results." National Academies of Sciences, Engineering, and Medicine. 2020. Advanced Ground Vehicle Technologies for Airside Operations. Washington, DC: The National Academies Press. doi: 10.17226/26017.
×
Page 70
Page 71
Suggested Citation:"Chapter 6 - Detailed Evaluation Results." National Academies of Sciences, Engineering, and Medicine. 2020. Advanced Ground Vehicle Technologies for Airside Operations. Washington, DC: The National Academies Press. doi: 10.17226/26017.
×
Page 71
Page 72
Suggested Citation:"Chapter 6 - Detailed Evaluation Results." National Academies of Sciences, Engineering, and Medicine. 2020. Advanced Ground Vehicle Technologies for Airside Operations. Washington, DC: The National Academies Press. doi: 10.17226/26017.
×
Page 72
Page 73
Suggested Citation:"Chapter 6 - Detailed Evaluation Results." National Academies of Sciences, Engineering, and Medicine. 2020. Advanced Ground Vehicle Technologies for Airside Operations. Washington, DC: The National Academies Press. doi: 10.17226/26017.
×
Page 73
Page 74
Suggested Citation:"Chapter 6 - Detailed Evaluation Results." National Academies of Sciences, Engineering, and Medicine. 2020. Advanced Ground Vehicle Technologies for Airside Operations. Washington, DC: The National Academies Press. doi: 10.17226/26017.
×
Page 74
Page 75
Suggested Citation:"Chapter 6 - Detailed Evaluation Results." National Academies of Sciences, Engineering, and Medicine. 2020. Advanced Ground Vehicle Technologies for Airside Operations. Washington, DC: The National Academies Press. doi: 10.17226/26017.
×
Page 75
Page 76
Suggested Citation:"Chapter 6 - Detailed Evaluation Results." National Academies of Sciences, Engineering, and Medicine. 2020. Advanced Ground Vehicle Technologies for Airside Operations. Washington, DC: The National Academies Press. doi: 10.17226/26017.
×
Page 76
Page 77
Suggested Citation:"Chapter 6 - Detailed Evaluation Results." National Academies of Sciences, Engineering, and Medicine. 2020. Advanced Ground Vehicle Technologies for Airside Operations. Washington, DC: The National Academies Press. doi: 10.17226/26017.
×
Page 77
Page 78
Suggested Citation:"Chapter 6 - Detailed Evaluation Results." National Academies of Sciences, Engineering, and Medicine. 2020. Advanced Ground Vehicle Technologies for Airside Operations. Washington, DC: The National Academies Press. doi: 10.17226/26017.
×
Page 78
Page 79
Suggested Citation:"Chapter 6 - Detailed Evaluation Results." National Academies of Sciences, Engineering, and Medicine. 2020. Advanced Ground Vehicle Technologies for Airside Operations. Washington, DC: The National Academies Press. doi: 10.17226/26017.
×
Page 79
Page 80
Suggested Citation:"Chapter 6 - Detailed Evaluation Results." National Academies of Sciences, Engineering, and Medicine. 2020. Advanced Ground Vehicle Technologies for Airside Operations. Washington, DC: The National Academies Press. doi: 10.17226/26017.
×
Page 80
Page 81
Suggested Citation:"Chapter 6 - Detailed Evaluation Results." National Academies of Sciences, Engineering, and Medicine. 2020. Advanced Ground Vehicle Technologies for Airside Operations. Washington, DC: The National Academies Press. doi: 10.17226/26017.
×
Page 81
Page 82
Suggested Citation:"Chapter 6 - Detailed Evaluation Results." National Academies of Sciences, Engineering, and Medicine. 2020. Advanced Ground Vehicle Technologies for Airside Operations. Washington, DC: The National Academies Press. doi: 10.17226/26017.
×
Page 82
Page 83
Suggested Citation:"Chapter 6 - Detailed Evaluation Results." National Academies of Sciences, Engineering, and Medicine. 2020. Advanced Ground Vehicle Technologies for Airside Operations. Washington, DC: The National Academies Press. doi: 10.17226/26017.
×
Page 83
Page 84
Suggested Citation:"Chapter 6 - Detailed Evaluation Results." National Academies of Sciences, Engineering, and Medicine. 2020. Advanced Ground Vehicle Technologies for Airside Operations. Washington, DC: The National Academies Press. doi: 10.17226/26017.
×
Page 84
Page 85
Suggested Citation:"Chapter 6 - Detailed Evaluation Results." National Academies of Sciences, Engineering, and Medicine. 2020. Advanced Ground Vehicle Technologies for Airside Operations. Washington, DC: The National Academies Press. doi: 10.17226/26017.
×
Page 85
Page 86
Suggested Citation:"Chapter 6 - Detailed Evaluation Results." National Academies of Sciences, Engineering, and Medicine. 2020. Advanced Ground Vehicle Technologies for Airside Operations. Washington, DC: The National Academies Press. doi: 10.17226/26017.
×
Page 86
Page 87
Suggested Citation:"Chapter 6 - Detailed Evaluation Results." National Academies of Sciences, Engineering, and Medicine. 2020. Advanced Ground Vehicle Technologies for Airside Operations. Washington, DC: The National Academies Press. doi: 10.17226/26017.
×
Page 87
Page 88
Suggested Citation:"Chapter 6 - Detailed Evaluation Results." National Academies of Sciences, Engineering, and Medicine. 2020. Advanced Ground Vehicle Technologies for Airside Operations. Washington, DC: The National Academies Press. doi: 10.17226/26017.
×
Page 88
Page 89
Suggested Citation:"Chapter 6 - Detailed Evaluation Results." National Academies of Sciences, Engineering, and Medicine. 2020. Advanced Ground Vehicle Technologies for Airside Operations. Washington, DC: The National Academies Press. doi: 10.17226/26017.
×
Page 89
Page 90
Suggested Citation:"Chapter 6 - Detailed Evaluation Results." National Academies of Sciences, Engineering, and Medicine. 2020. Advanced Ground Vehicle Technologies for Airside Operations. Washington, DC: The National Academies Press. doi: 10.17226/26017.
×
Page 90
Page 91
Suggested Citation:"Chapter 6 - Detailed Evaluation Results." National Academies of Sciences, Engineering, and Medicine. 2020. Advanced Ground Vehicle Technologies for Airside Operations. Washington, DC: The National Academies Press. doi: 10.17226/26017.
×
Page 91
Page 92
Suggested Citation:"Chapter 6 - Detailed Evaluation Results." National Academies of Sciences, Engineering, and Medicine. 2020. Advanced Ground Vehicle Technologies for Airside Operations. Washington, DC: The National Academies Press. doi: 10.17226/26017.
×
Page 92
Page 93
Suggested Citation:"Chapter 6 - Detailed Evaluation Results." National Academies of Sciences, Engineering, and Medicine. 2020. Advanced Ground Vehicle Technologies for Airside Operations. Washington, DC: The National Academies Press. doi: 10.17226/26017.
×
Page 93
Page 94
Suggested Citation:"Chapter 6 - Detailed Evaluation Results." National Academies of Sciences, Engineering, and Medicine. 2020. Advanced Ground Vehicle Technologies for Airside Operations. Washington, DC: The National Academies Press. doi: 10.17226/26017.
×
Page 94
Page 95
Suggested Citation:"Chapter 6 - Detailed Evaluation Results." National Academies of Sciences, Engineering, and Medicine. 2020. Advanced Ground Vehicle Technologies for Airside Operations. Washington, DC: The National Academies Press. doi: 10.17226/26017.
×
Page 95
Page 96
Suggested Citation:"Chapter 6 - Detailed Evaluation Results." National Academies of Sciences, Engineering, and Medicine. 2020. Advanced Ground Vehicle Technologies for Airside Operations. Washington, DC: The National Academies Press. doi: 10.17226/26017.
×
Page 96
Page 97
Suggested Citation:"Chapter 6 - Detailed Evaluation Results." National Academies of Sciences, Engineering, and Medicine. 2020. Advanced Ground Vehicle Technologies for Airside Operations. Washington, DC: The National Academies Press. doi: 10.17226/26017.
×
Page 97
Page 98
Suggested Citation:"Chapter 6 - Detailed Evaluation Results." National Academies of Sciences, Engineering, and Medicine. 2020. Advanced Ground Vehicle Technologies for Airside Operations. Washington, DC: The National Academies Press. doi: 10.17226/26017.
×
Page 98
Page 99
Suggested Citation:"Chapter 6 - Detailed Evaluation Results." National Academies of Sciences, Engineering, and Medicine. 2020. Advanced Ground Vehicle Technologies for Airside Operations. Washington, DC: The National Academies Press. doi: 10.17226/26017.
×
Page 99
Page 100
Suggested Citation:"Chapter 6 - Detailed Evaluation Results." National Academies of Sciences, Engineering, and Medicine. 2020. Advanced Ground Vehicle Technologies for Airside Operations. Washington, DC: The National Academies Press. doi: 10.17226/26017.
×
Page 100
Page 101
Suggested Citation:"Chapter 6 - Detailed Evaluation Results." National Academies of Sciences, Engineering, and Medicine. 2020. Advanced Ground Vehicle Technologies for Airside Operations. Washington, DC: The National Academies Press. doi: 10.17226/26017.
×
Page 101
Page 102
Suggested Citation:"Chapter 6 - Detailed Evaluation Results." National Academies of Sciences, Engineering, and Medicine. 2020. Advanced Ground Vehicle Technologies for Airside Operations. Washington, DC: The National Academies Press. doi: 10.17226/26017.
×
Page 102
Page 103
Suggested Citation:"Chapter 6 - Detailed Evaluation Results." National Academies of Sciences, Engineering, and Medicine. 2020. Advanced Ground Vehicle Technologies for Airside Operations. Washington, DC: The National Academies Press. doi: 10.17226/26017.
×
Page 103
Page 104
Suggested Citation:"Chapter 6 - Detailed Evaluation Results." National Academies of Sciences, Engineering, and Medicine. 2020. Advanced Ground Vehicle Technologies for Airside Operations. Washington, DC: The National Academies Press. doi: 10.17226/26017.
×
Page 104
Page 105
Suggested Citation:"Chapter 6 - Detailed Evaluation Results." National Academies of Sciences, Engineering, and Medicine. 2020. Advanced Ground Vehicle Technologies for Airside Operations. Washington, DC: The National Academies Press. doi: 10.17226/26017.
×
Page 105
Page 106
Suggested Citation:"Chapter 6 - Detailed Evaluation Results." National Academies of Sciences, Engineering, and Medicine. 2020. Advanced Ground Vehicle Technologies for Airside Operations. Washington, DC: The National Academies Press. doi: 10.17226/26017.
×
Page 106
Page 107
Suggested Citation:"Chapter 6 - Detailed Evaluation Results." National Academies of Sciences, Engineering, and Medicine. 2020. Advanced Ground Vehicle Technologies for Airside Operations. Washington, DC: The National Academies Press. doi: 10.17226/26017.
×
Page 107
Page 108
Suggested Citation:"Chapter 6 - Detailed Evaluation Results." National Academies of Sciences, Engineering, and Medicine. 2020. Advanced Ground Vehicle Technologies for Airside Operations. Washington, DC: The National Academies Press. doi: 10.17226/26017.
×
Page 108
Page 109
Suggested Citation:"Chapter 6 - Detailed Evaluation Results." National Academies of Sciences, Engineering, and Medicine. 2020. Advanced Ground Vehicle Technologies for Airside Operations. Washington, DC: The National Academies Press. doi: 10.17226/26017.
×
Page 109
Page 110
Suggested Citation:"Chapter 6 - Detailed Evaluation Results." National Academies of Sciences, Engineering, and Medicine. 2020. Advanced Ground Vehicle Technologies for Airside Operations. Washington, DC: The National Academies Press. doi: 10.17226/26017.
×
Page 110
Page 111
Suggested Citation:"Chapter 6 - Detailed Evaluation Results." National Academies of Sciences, Engineering, and Medicine. 2020. Advanced Ground Vehicle Technologies for Airside Operations. Washington, DC: The National Academies Press. doi: 10.17226/26017.
×
Page 111
Page 112
Suggested Citation:"Chapter 6 - Detailed Evaluation Results." National Academies of Sciences, Engineering, and Medicine. 2020. Advanced Ground Vehicle Technologies for Airside Operations. Washington, DC: The National Academies Press. doi: 10.17226/26017.
×
Page 112
Page 113
Suggested Citation:"Chapter 6 - Detailed Evaluation Results." National Academies of Sciences, Engineering, and Medicine. 2020. Advanced Ground Vehicle Technologies for Airside Operations. Washington, DC: The National Academies Press. doi: 10.17226/26017.
×
Page 113
Page 114
Suggested Citation:"Chapter 6 - Detailed Evaluation Results." National Academies of Sciences, Engineering, and Medicine. 2020. Advanced Ground Vehicle Technologies for Airside Operations. Washington, DC: The National Academies Press. doi: 10.17226/26017.
×
Page 114
Page 115
Suggested Citation:"Chapter 6 - Detailed Evaluation Results." National Academies of Sciences, Engineering, and Medicine. 2020. Advanced Ground Vehicle Technologies for Airside Operations. Washington, DC: The National Academies Press. doi: 10.17226/26017.
×
Page 115
Page 116
Suggested Citation:"Chapter 6 - Detailed Evaluation Results." National Academies of Sciences, Engineering, and Medicine. 2020. Advanced Ground Vehicle Technologies for Airside Operations. Washington, DC: The National Academies Press. doi: 10.17226/26017.
×
Page 116
Page 117
Suggested Citation:"Chapter 6 - Detailed Evaluation Results." National Academies of Sciences, Engineering, and Medicine. 2020. Advanced Ground Vehicle Technologies for Airside Operations. Washington, DC: The National Academies Press. doi: 10.17226/26017.
×
Page 117
Page 118
Suggested Citation:"Chapter 6 - Detailed Evaluation Results." National Academies of Sciences, Engineering, and Medicine. 2020. Advanced Ground Vehicle Technologies for Airside Operations. Washington, DC: The National Academies Press. doi: 10.17226/26017.
×
Page 118
Page 119
Suggested Citation:"Chapter 6 - Detailed Evaluation Results." National Academies of Sciences, Engineering, and Medicine. 2020. Advanced Ground Vehicle Technologies for Airside Operations. Washington, DC: The National Academies Press. doi: 10.17226/26017.
×
Page 119
Page 120
Suggested Citation:"Chapter 6 - Detailed Evaluation Results." National Academies of Sciences, Engineering, and Medicine. 2020. Advanced Ground Vehicle Technologies for Airside Operations. Washington, DC: The National Academies Press. doi: 10.17226/26017.
×
Page 120
Page 121
Suggested Citation:"Chapter 6 - Detailed Evaluation Results." National Academies of Sciences, Engineering, and Medicine. 2020. Advanced Ground Vehicle Technologies for Airside Operations. Washington, DC: The National Academies Press. doi: 10.17226/26017.
×
Page 121

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

68 The previous chapter presented a general evaluation process that can be used by a specific airport to evaluate alternative AGVT applications. This chapter presents the evaluation results for the prioritized technology applications in the airside environment, with respect to the evalu- ation criteria ease of adoption and stakeholder acceptance, technical feasibility, infrastructure impacts, operational impacts, benefits and human factors. The evaluations are conducted at a high level, which reflects the fact that many of the AGVT applications are conceptual and the enabling technologies are still being developed, as illustrated by the TRL shown in Table 22. A more detailed evaluation will be possible as the enabling technologies advance (e.g., as LiDAR capabilities are proven in the airport environment) and when the application is considered for a specific airport. The evaluation results in this chapter include both ramp and airport operations activities. Selected one-page summaries for the most appropriate applications are provided in Chapter 7. Supporting information for the evaluations is provided in the appendices (Appendix M through Appendix R). The following sections present the evaluation of AGVT to support the following airside activities: Airport Operations: • FOD detection and removal: safety assist (including ADS-B for fleet management and to prevent runway incursions) and automated with driver • Mowing: automated with no driver and central control • Snow and ice control: platoon with driver in lead and remote operation (GA airport) • Perimeter inspection: automated with no driver Ramp and Aircraft Activities: • Aircraft pushback and aircraft tug/taxi to runway – Aircraft pushback: remote from ramp and automated with driver – Aircraft tow or tug to and from runway: remote from cockpit and automated with driver • Baggage carts: safety assist Automated AGVT that operate near aircraft (e.g., in the movement area or non-movement area) all have a driver (or operator) who can respond immediately during an emergency or to an unexpected situation. This includes FOD detection and removal, snowplow platoon, aircraft pushback and aircraft tug to runway. Safety assist AGVT (FOD detection and removal and baggage carts) have a driver that retains responsibility for vehicle operations. Automated AGVT that operate in more remote areas of the airport, such as mowers or a perimeter inspection vehicle, do not have a driver (or operator) who would be expected to respond immediately, however, these vehicles can be controlled remotely in the event of an alert. C H A P T E R 6 Detailed Evaluation Results

Detailed Evaluation Results 69 The proposed implementation for each application depends on the permanence of the project, the maturity of the technology and the impact on airport operations. Applications that will utilize mature technologies are recommended for full deployment on a permanent basis. Applications that utilize AGVT that have not been proven in the airside or are otherwise still maturing are recommended for implementation as a demonstration. A demonstration is implemented on a trial basis for a limited time with a scope designed to have minimal disruption to ongoing airside operations. Applications that utilize AGVT with a lower TRL are recommended as an investi- gation on a temporary basis with the intent to collect data and learn more about the capabilities and limitations in support of future activities. AGVT that operate in areas where there are no aircraft (e.g., mowing and perimeter security) and AGVT that provide safety assist rather than automated operation can be implemented with greater confidence. Implementation of all AGVT will provide benefits such as data to support future AGVT activities, including disengagement data, sensor limitations, and human factors considerations. Three technology applications could be implemented as an investigation but are not currently technically feasible. Automated tug to departure runway is not currently feasible because aircraft braking relies on pilot action in the cockpit. Automated pushback is not currently feasible due to the complexity of activities on the ramp. Current sensors and supporting software technologies are not adequately developed to safely function in this complex environment. Both automated tug to departure runway and automated pushback have potential benefits in the long term for safety, efficiency, and sustainability. The best way to work toward this long-term goal is to implement and learn from the implementation of remote from the cockpit tug and remote pushback, both of which are currently technically feasible. Similarly, central control mowing Application Technology Estimated TRL Comments Baggage carts Safety assist 7 Confirmation in airside environment required (TRL 8); issues with respect to chain of trailing carts jackknifing require special algorithms (TRL 3). Aircraft pushback Remote from ramp 9 Commercial products are readily available. Automated with driver 5 or 6 Constrained by obstacle avoidance and complex operating environment. Aircraft tow or tug Remote from cockpit 8 or 9 Demonstrated at other airports. Automated with driver 4 or 5 Sensors have a high level of readiness (TRL ranges from 7 to 9) but positions of other GSE and ramp personnel, and collaborative mapping and deployment has not been applied in real world (TRL 6), constrained by need for pilot to apply brakes if needed since there is no mechanism for remote application of aircraft brakes (TRL 2). FOD detection and removal Safety assist 8 or 9 Confirmation of operation in airside environment required, including issues with vertical spatial features for aircraft recognition. Automated with driver 7 or 8 Confirmation of operation in airside environment required, including issues with vertical spatial features for aircraft recognition. Mowing Automated no driver 8 or 9 Proposed operation in near term does not include automated crossing of runways or taxiways. Central control 7 Has been adopted in relative civil and military applications (e.g., fleet of drones). Centralized control of fleet of mowers needs to be implemented from remote operation of single mower/ground vehicle. Fleet management functionalities need to be developed and integrated. Snow and ice control Platoon with driver in lead 7 or 8 Inter-vehicle issues including collision avoidance for following vehicles and possible issues with latency of 4G network need to be addressed. Remote operation 7 Technology has been used in related fields (remote operation of drones, etc.). Fleet management functionalities need to be developed and integrated. Perimeter inspection Automated no driver 8 Investigation currently underway at Edmonton Airport. Indianapolis Motor Speedway uses system currently (non-airport environment). The elements for the system have high TRL (TRL 8 or 9), and the system is deployable. However, more experiments need to be done to confirm long-term reliability and response to unusual circumstances. Table 22. Overall estimated TRL for airside AGVT applications.

70 Advanced Ground Vehicle Technologies for Airside Operations has potential benefits in the long term for sustainability and safety, but it would be logical to first implement and learn from individual automated mowers, which are currently techni- cally feasible. FOD Detection and Removal Description FOD detection and removal are important components of an FOD management program; the other important components are prevention (which includes personnel training) and evalu- ation. FOD inspections are required under Part 139 as regularly scheduled, continuous, and special inspections. Regularly scheduled inspections must occur at least daily and FOD checks are components of the regular inspections of the pavement area, for snow and ice inspection, and for construction inspections. Larger airports may conduct FOD inspections multiple times a day (e.g., three times in a 24-hour period or more). Continuous surveillance inspections require that FOD be continuously checked for and removed in the movement area, aircraft parking areas, loading ramps and in construction areas by personnel as they go about their regular duties. Special conditions inspections for snow and ice must also ensure that all foreign objects have been picked up after snow and ice removal operations. Under Part 139, FOD must be removed promptly and as completely as practicable. Airport personnel may also conduct “FOD walks” one or more times a week with airport tenants who are involved in activities that are likely to produce FOD. The proposed AGVT applications for evaluation include: • Safety assist • Automation with driver Both stationary and mobile FOD detection has been developed and certified by FAA. While stationary systems provide continuous surveillance, mobile units may be less expensive and allow FOD removal upon detection. Mobile units have been developed and certified by FAA. The proposed evaluation of AGVT reflects the automation of the vehicle itself and not the equipment for FOD detection and removal. There are a variety of equipment for FOD removal, ranging from mechanical (e.g., sweeper and vacuum systems) to non-mechanical (e.g., mats and magnets), as illustrated in Figure 23. The most appropriate FOD removal system varies depending on airport characteristics, including the airport operating characteristics and the kind of FOD usually collected at the airport. Other considerations include airport activities (a) (b) Figure 23. FOD management equipment: (a) multiple units used for a wide sweep (24 ft sweep width clears 3 million sq ft per hour and (b) sweeper truck for FOD removal. Photo: Team Eagle, 2017c.

Detailed Evaluation Results 71 (aircraft operations, maintenance activities, and construction), airport layout (e.g., proximity of construction areas, maintenance and aircraft operation areas to the taxiways and runways), and environmental conditions (e.g., dust, rain, and likely weather contaminants), which are affected by local terrain, weather, and wildlife. Information to support FOD management activities is provided in the following advisory circulars: • AC 150/5210-24, Airport Foreign Object Debris (FOD) Management, 9/30/2010; and • AC 150/5220-24, Airport Foreign Object Debris (FOD) Detection Equipment, 9/30/2009. The use of an ADS-B transponder operating on the 978 MHz frequency to support safety assist for the airport ops vehicle would ensure compatibility with ASDE-X or ASSC airport surface surveillance systems and enable information about the ground vehicle to be transmitted to nearby aircraft to enhance situational awareness. Information to support ADS-B for ground vehicles can be found in the following advisory circulars: • AC 150/5220-26, Airport Ground Vehicle Automatic Dependent Surveillance—Broadcast (ADS-B) Out Squitter Equipment, 11/14/2011. The proposed AGVT can be used to support existing FOD detection and removal technologies that have been approved by the FAA. For this evaluation, a mobile detection and removal system is used to illustrate the concept since it supports all components of a FOD management system. This equipment includes a radar and camera based system that is mounted on an airport opera- tions truck to detect FOD in all weather and lighting conditions, and it is one of four auto- mated FOD detection systems that has been evaluated by FAA (Weller, 2014). This equipment includes a vacuum for FOD collection and provides geotagging and imaging of the FOD collected, which is important for the evaluation phase of a FOD management program. The geotagging component is based on a digital airfield management system that integrates and supports Part 139 airfield inspection requirements. The system provides the capability to scan, detect, retrieve and contain FOD without the driver exiting the vehicle. The FOD identi fication and retrieval components are an FAA certified FOD system with 100% detection capability in all weather and lighting conditions (Trex Aviation, 2016). In this case, there are relatively mature technologies for FOD detection and retrieval that have been approved by FAA; when combined with AGVT, the potential for a fully automated system is significant. It would also be practical to use AGVT technology in conjunction with sweeping units or a sweeper truck. These systems may be very effective at removing FOD, but do not provide as much information about the specific FOD collected since there is no automation to document the location where FOD is collected. Safety assist components include a runway incursion prevention system (RIPS), lane-keeping assist, and collision avoidance system. These systems provide warnings of upcoming runway thresholds, ILS boundaries, and restricted areas, as well as obstacles in the vehicle path. Safety assist also includes warnings and information from sensors and onboard systems to increase operator situational awareness and reduce workload. The use of an ADS-B transponder also allows the airport operations vehicle to be visible to nearby aircraft to enhance situational awareness. Automation with a safety driver would allow the vehicle to automatically traverse the designated route for a regularly scheduled daily inspection; the driver would be required to be in the vehicle and take over if needed. As with most FOD detection and removal systems, in some cases, there may be FOD that is too large for the automated system to collect, in which case manual inter- vention by the driver may be needed. As long as the safety driver is present in the vehicle, this would not result in significant additional delay.

72 Advanced Ground Vehicle Technologies for Airside Operations Implementation Scenario The proposed implementation of the safety assist features would be full deployment which is consistent with the current TRL. The proposed implementation of the automated with a safety driver would be a demonstration project. This would allow close supervision to reduce risk as well as the opportunity to collect data and learn more about airside automation, a significant benefit since automated operation of airside vehicles has not been documented in the United States, and there is no operational data to support full deployment. Operational Area The operational area for evaluation includes the runways, taxiways, and aprons for which the airport is responsible for regular inspections under Part 139. The equipment could also be used in other areas (e.g., tenant areas or construction areas); however, use in these other areas is not a consideration in this evaluation. FOD management programs typically address all of the airfield and even adjacent areas since wind, rain, people, and vehicles can transfer FOD from remote areas where FOD creates no hazard to critical airfield areas where FOD can present significant hazards. Although the airport is responsible for FOD management on the entire airport, airlines, maintenance contractors, construction contractors and other airside tenants also have responsibility for FOD manage- ment on the property where their activities occur. Airport Characteristics Safety assist with ADS-B transponders for FOD management would be most appropriate at airports where runway incursion mitigation activities are required due to a high incidence of runway incursions in the past and/or risk factors that make runway incursions more likely or more dangerous. Risk factors may include airfield geometry with numerous runway inter- sections, airfield hot spots, and/or large pavement areas, or other risk factors such as frequent employee turnover, or a significant number of employees without extensive experience driving. This could occur if the airport is in a metropolitan area where transit rather than personal vehicles is used for mobility. Safety assist may also be appropriate for an airport that has concerns related to ground vehicle accidents and/or obstacle avoidance. The use of ADS-B transponders in ground vehicles increases the situational awareness for pilots near the vehicles. Large airports may be more likely to leverage a safety assist system that utilizes ADS-B technology since AIP and PFC funds can be used for this expenditure. Smaller airports may wish to use GPS transponders that are not ADS-B since other transponders may be less expensive and incorporate technology advancements that have occurred since the ADS-B standard was developed. It is also technologically possible for information from multiple ground vehicle GPS transponders to be communicated with a single ADS-B unit, which would reduce costs. Automation with a safety driver for FOD management would be appropriate for an airport with activities that make FOD more likely (e.g., construction, aircraft maintenance or MRO tenants, significant cargo operations, extensive commercial aircraft activity) and a high level of aircraft activities that make runway closures for FOD detection and removal costly in terms of aircraft delay. A system with automated FOD detection and removal would be most appro- priate at a large-hub or medium-hub airport where capacity constraints justify the cost of the system. These operational considerations that suggest AGVT for larger airports also align

Detailed Evaluation Results 73 with research that suggests that the size of an organization contributes to innovation adoption decisions (Kennedy, 1983). Although the use of FOD detection AGVT will not be readily apparent to passengers, it may be apparent and attractive to prospective and current airport tenants, and may be attractive to airports that wish to convey a strong strategic posture with regards to innovation. At a smaller airport with less aeronautical activity, it may be more appropriate to consider an automated ops truck with a safety driver pulling an airport runway sweeper. In this case, the vehicle automation may allow the airport ops worker to give more attention to FOD detec- tion or other visual inspection duties. In any case, the vehicle automation does not replace the need for airport workers, contractors and other people to be vigilant as part of their continuous surveillance activities. The geographical area and weather-related concerns would not be an issue in terms of system capabilities (the radar can detect objects in snow), however, they may affect the kind of FOD equipment that is most appropriate, since some environments may increase the likelihood of FOD (e.g., windblown FOD and rain swept FOD). Project Partners Large cargo operators, MRO tenants, signatory airlines, or other airport partners that poten- tially contribute to FOD, and/or are vulnerable to FOD damage due to their aeronautical activities may be appropriate partners for the proposed deployment. Technology vendors for FOD equipment and/or AGVT equipment may also be appropriate partners. Key Considerations Key considerations for FOD detection and removal include the following: • Benefits. Benefits for safety assist with ADS-B transponders include increased airfield safety due to reduced runway incursions and reduced potential for damage due to collisions. Benefits for automation with a safety driver include increased safety and the availability of useful data to support future airside AGVT, including full deployment of automated FOD detection and retrieval. The integration of AGVT may also provide a more predictable inspection time, which may allow better utilization of the available airfield capacity. To the extent that the proposed automation supports improved FOD management, the project would result in significant benefits due to the high cost associated with FOD damage to aircraft and FOD injury for personnel. The benefits of improved FOD management include cleaner surfaces (runways, taxiways, and ramps), safer operation of aircraft at the airport, and better Part 139 inspection data collection and analysis. Safety assist may also support increased FOD detection, to the extent that the driver can focus more attention on visual FOD management. • Technical feasibility. Current FOD detection methods involve combined use of millimeter wave radar, LiDAR, and high-resolution camera mounted on mobile platforms such as pickup trucks or stationary stands next to the runway. This technology has been readily avail- able (TRL 9) and has been integrated with a digital representation of the airport, which can guide the driver toward detected FOD and provide a summary of all FOD that has been detected. Mobile systems have integrated the FOD detection system on a pickup truck so a single detection system can perform FOD detection over the whole airport. LiDAR and infrared cameras are also used in combination, but they are often mounted on the roof of airfield vehicles because of their lower working range (Weller, 2014). These remote- sensing technologies have been readily available (TRL 9) and have been operating in

74 Advanced Ground Vehicle Technologies for Airside Operations various airports. The evaluation methodology and criteria ensures these detection methods to locate golf-ball sized objects within 100 ft (30 m) range, with accuracy higher than 15 ft (5 m) (AC 150-5220-24, 2009). Other options for FOD removal involve mechanical systems such as power sweepers, vacuum systems, and jet air blowers; non-mechanical systems include friction mat sweepers, magnetic bars, and rumble strips; both work with supporting storage systems. These systems are of high technology readiness (TRL 9), and have been successfully deployed at airports (AC 150-5220-24, 2009; and AC 150-5210-24, 2010). Safety assist technologies have been used on commercial passenger vehicles, which means they are more readily available (TRL 8 or 9). To achieve L2 autonomous driving, lane keeping can be achieved through front-facing cameras and collision avoidance using millimeter wave radars. This technology has been implemented on commercial vehicles such as Toyota’s lane-keeping and pre-collision warning and braking with pedestrian detection systems (Toyota Inc., 2018), which have been tested in operational environments (TRL 8). On a runway, lane keeping is still feasible, given the center and edge markers of the runway are clearly identifiable. FAA has established standards for airport GIS systems (AC 150/5300-17C, 2011), and airport information is continuously contributed by airports into the FAA data- base. This would suggest a TRL 8 or 9 for the GIS component of the system. Among the additional sensors added to achieve L3 automation, GPS and inertial measure- ment unit (IMU) are combined for vehicle localization. The IMU and GPS sensors are readily available (TRL 9) to industries such as aerospace. Motion planning for the vehicle path in the designated area utilizes a set of automatically generated points (waypoint following) to reach the target position labeled on the map, i.e., the positions of detected FOD, which may be generated by the automated detection scan or assigned by a human operator based on visual detection. Waypoint following is a function achieved by software with mature theory and application in relevant fields (e.g., roombas, industrial robots on assembly lines), which can be rated as at least TRL 6 for airport operations. For localization of airport autono- mous ground vehicles using high-resolution maps, a high-resolution 2D map of the airport can be obtained by aerial surveys or satellite images, which reflects a TRL 9. Alternatively, a 3D map can be obtained with an airport survey using the same hardware application (radar and camera, with LiDAR as an alternative), and a map stitching algorithm based on simul- taneous localization and mapping (SLAM), which is readily available for street mapping. To apply these methods to airport applications, it would be necessary to address the poten- tial lack of vertical spatial features and aircraft motion, which suggests the mapping process would be a TRL 7. • Operational impacts. For safety assist, the operational impacts may be a reduction in runway incursions and vehicle collisions. There may be a slight increase in the capability to detect FOD since the driver may be able to focus more attention on the FOD inspection. Safety assist would not be expected to increase capacity or efficiency during regular inspections. Operational impacts for automated vehicle with a safety driver could potentially be more significant, and may include faster and more reliable time for FOD detection and retrieval, which would have a positive impact on efficiency and capacity. Another benefit is the incre- mental progress toward automated without a driver (for systems that do not rely on human visual observation), which would potentially increase the efficiency of airport personnel. • Infrastructure impacts. Both safety assist and automated with a safety driver require recogni- tion of ground markings to identify the functional zone and corresponding actions (e.g., taxi lanes, runway threshold, and runways, which should be clearly marked). This will provide confirmation of the information provided by an airport GIS map (e.g., movement area, non- movement area and functional zones), in which the coordinates of the associated borders will support software geofencing to ensure safe operation.

Detailed Evaluation Results 75 Automation with a safety driver requires a higher definition (3 ft or 1 m) map of the airport, along with a layer of high-resolution imagery content, which is used primarily for vehicle localization by matching landmarks. As a baseline requirement, the map could be stored in the onboard computer and updated manually during vehicle maintenance, since maps do not usually change frequently (unless airfield construction is underway). To obtain this map, a high-definition satellite image would be appropriate. It could also be obtained by downloading the logged data from the vehicle radar and cameras during maintenance, and performing an off-line SLAM on a server computer to generate a 3D high-resolution map of the airport. Additional features for implementation, such as an update of the airport map with frequency higher than once per week, would require additional infrastructure construc- tion, such as networking facilities (base and relay stations for cellular network), to enable over-the-air updates. • Stakeholder acceptance and ease of adoption. Safety assist is likely to be well accepted and easy to adopt since many of the features have become common in the roadside environment, the benefits have been well documented in this context, and many workers may be familiar with these driver assistance features, which are fairly intuitive. Safety assist has minimal impacts on airfield rules and regulations, and has been approved by FAA for runway incur- sion warning systems, which provide warnings but typically not active control. Automated FOD detection with a safety driver would likely require regulatory approval before implementation, which would reduce the ease of adoption and increase the time required for implementation. Automated with a safety driver may be less attractive to airports that are risk averse and airports that have had previous negative experiences with technology deployments. Airports in locations where AGVT testing has been deployed on city streets (e.g., in states such as California, Arizona, Ohio and Michigan, and in cities such as Phoenix and Pittsburgh) and airports that have participated in previous research projects may have greater acceptance of new technologies. Regulatory approval includes not only FAA but also the Federal Communications Commis- sion (FCC), the federal agency responsible for communications (including radio, wire, wireless, satellite and cable) in the United States, as well as for certification of products that utilize radio frequency (FCC, n.d.). The FOD Finder system received a waiver from the FCC in 2013, demonstrating the potential for approval (Trex Aviation, n.d.). Resistance from workers would be reduced as there is no reduction in workforce and because the AGVT would improve worker safety and reduce workload. This is a positive with regards to individual worker acceptance but could be a negative for airports looking to reduce labor expenditures through AGVT. Worker and union resistance would increase if the AGVT is considered a precursor to full automation and a workforce reduction, which would be more likely for automation with a safety driver. Stakeholder acceptance also reflects perceived usefulness and ease of use for the individual (Davis, 1989). The proposed safety assist is expected to have high perceived usefulness and ease of use. Previous research suggests the use of voice warnings raises the likelihood of compliance (Wogalter and Young, 1991). They are also preferred because they reduce visual workload and interference with non-driving tasks (Naujoks et al., 2016), in this case FOD inspections. Provided false alarms are not common and the “cry wolf” effect (Breznitz, 2013) is avoided, safety assist will be trusted by individuals, which bodes well for its acceptance (Ghazizadeh, Lee, and Boyle, 2012). Automated FOD detection with a safety driver stands to have lower levels of perceived usefulness (worker might not feel it is necessary) and perceived ease of use (automation could be more complex than status quo). Since trust in automation is a function of experience, trust levels may start out low and increase with positive experiences.

76 Advanced Ground Vehicle Technologies for Airside Operations The need for a person for both AGVT applications may reduce the potential for misuse or abuse of the technology since there is no direct threat to employment; however, anecdotally, there are reports of worker resistance and even vandalism to new technologies in the airside envi- ronment, particularly on the ramp. This may be less likely since the proposed application is in airport operations, and the vehicle would be checked out to a single person rather than accessible to a wide variety of workers. Moreover, the use of cameras as part of the AGVT may also reduce the likelihood of vandalism since cameras may increase the perception of accountability. • Human factors. For safety assist, the human factors considerations are positive since the driver assist functions support driver safety but maintain operator responsibility for driving, communications with ATCT and other standard procedures. The successful deployment in the roadside sector supports the value of a reduced workload and the safety benefits of driver assistance technologies. As safety assist features become increasingly common in personal vehicles used in the roadway sector, it may become increasingly important to include them in airside vehicles, due to driver reliance and expectation. Additional research may be needed to identify the most appropriate warning system for the airside environment (e.g., optimal combination of visual, auditory and haptic feedback). This is because workload and environment differ in the airfield and roadway environment. For automated vehicle with a safety driver, there are a number of human factors consider- ations. The most important are the situational awareness of the safety driver, the ability of the safety driver to take over control when needed, and confirmation that the system does reduce human error in this application. Most of these topics have been addressed to some extent in other sectors (e.g., roadway) and have been addressed by theoretical research, but would require confirmation in the airside environment. The proposed demonstration project would support increased understanding about functional design considerations that differ in the airside environment, including tasks and procedures, and interaction with aircraft and ATC. A better understanding of the compatibility of future AGVT operations with existing airside procedures, primarily coordination with ATCT and ramp control, is also important. Since FOD inspection does not require significant interactions with other vehicles or equipment or personnel (other than approval from ATCT), the human factors considerations are simplified. Adequate training for the safety driver is part of human factors and would support ease of acceptance. Potential Challenges There are no expected challenges for the deployment of safety assist technology for FOD inspections. The greatest potential challenges for the deployment of automated vehicle with a safety driver are the uncertainty regarding the automation capabilities in the airside environment. In the roadside environment, vehicles operate in lanes that are approximately 12 ft wide and have pavement markings to delineate lanes; potential obstacles include other vehicles, roadway infrastructure, pedestrians and cyclists. In the airside environment, runways may be 150 ft wide and aprons may be large expanses of pavement with few pavement markings; potential obstacles include aircraft, airport infrastructure, vehicles, and pedestrians. Although airfield pavement signs and markings are standard, the vehicle will have fewer environmental cues in terms of lateral placement due to the broad and often undifferentiated landscape. The airport represents a limited geographic area for operation and ATC provides permission to access the runway and taxiways; this provides structure for operation and limits the likelihood of unexpected obstacles. Nonetheless, aircraft are high value assets and a “minor” collision could cause aircraft damage and delay, both of which are costly; a collision with an aircraft on takeoff or landing could be catastrophic in terms of the loss of life and associated cost. Since there is limited experience with AGVT in the airside environment in the United States, it is challenging to anticipate all the potential challenges with confidence.

Detailed Evaluation Results 77 Mowing Description Airfield mowing is an important activity and is typically conducted as part of the airport wildlife hazard management plan. Airfield mowing can consume significant resources. It has been estimated that grassland represents about 40% to 50% of airport property (DeVault et al., 2012), although not all of this is airside and requires regular mowing. In addition to the wild- life management considerations, mowing is also important for aesthetic value, which adds to the economic value of the airport environment, and ensures an attractive and welcoming view from the terminal, and when approaching from both the ground and the air (Washburn and Seamans, 2013). In developed urban areas, airfield grass may be the largest grassland habitat in the ecosystem (DeVault et al., 2012). Grass also serves a number of important functional requirements. Well-managed turfgrass can reduce aircraft damage due to jet blast and FOD (Washburn and Seamans, 2013), must stand up to aircraft excursions and ARFF vehicles, and helps prevent soil erosion and reduce stormwater runoff. Ideally, grass will provide minimal food and habitat for wildlife, which is why mowing becomes a critical component of the Wildlife Hazard Management Plan. Airport mowing plans may require constant mowing as weather allows, mowing once or twice a week, or less frequent mowing with a greater reliance on chemicals to stunt vegetation growth (Washburn and Seamans, 2013). The proposed AGVT applications for evaluation include: • Automated with no driver • Central control There are no advisory circulars that are specifically targeted to mowing; however, a number of FAA documents reference grass and mowing. • AC 150/5200-33B, Hazardous Wildlife Attractants on or Near Airports, 8/28/2007, acknowl- edges that turf grass areas can attract hazardous wildlife and a management plan should be developed in cooperation with a wildlife biologist. • Draft AC 150/5200-33C, Hazardous Wildlife Attractants on or Near Airports, 1/18/2019, provides the same guidance on turf grass as AC 150/5200-33B [text moved from Section 2-7(b) in 33B to 2.8.22 in 33C]. • AC 150/5210-20A, Ground Vehicle Operations to include Taxiing or Towing an Aircraft on Airports, 9/1/2015, notes the Runway Safety Area (RSA) must be clear at all times during air carrier and aircraft operations, although there may be circumstances such as mowing in which vehicles or equipment are in the RSA for a limited time. • CertAlert 16-07, Personnel and Equipment in the Runway Safety Area (RSA), 10/14/2016, clarifies that if moving or stationary vehicles or equipment operate in the RSA during aircraft landing and departure operations, a letter of agreement must detail the specific activities. The airport operator is responsible for any mowing or maintenance in the RSA. The use of small mowers in the RSA may provide safety advantages relative to the larger mowers currently used that require a human operator. Small mowers may be frangible since they would yield to aircraft; this would allow mowing operations in the RSA during active air- craft operations. Information about frangible objects can be found in the following advisory circulars: • AC 150/5220-23, Frangible Connections, 4/27/2009, defines frangibility and frangible objects in airfield safety areas. Frangibility is the ability of an object to yield when impacted by another object. A frangible object is designed to have minimal mass and absorb a minimal amount of energy during impact. The goal is to minimize the potential for aircraft damage without impeding the motion or altering the path of an aircraft.

78 Advanced Ground Vehicle Technologies for Airside Operations There have been a number of mowers that utilized AGVT, and models have been used for athletic fields, golf courses, and roadway right-of-way. Examples of AGVT mowers are shown in Figure 24. Automated mowers have been in existence for decades; a patent for a random path self- propelled unattended lawnmower was filed 50 years ago in 1969 by MowBot (U.S. Patent 3698523A). MowBot still exists as a service provider and uses electric mowers that use a buried guide wire and GPS. MowBot has been joined by a number of other companies that provide a variety of automated, autonomous and remote mower technologies. Many of the mowers are lightweight, with multiple blades, which increases safety in the event of a collision with a person (Figure 24a). The Echo Robotic mowers automatically dock at charging sta- tions and can cut up to 5 acres with five floating heads configured to cut a 43 in wide path. Husqvarna has developed models that can use battery power or energy from a solar panel on the mower (Trimarchi, 2009). Husqvarna has an automower for athletic fields that uses GPS and assisted navigation or boundary guidewire (Husqvarna, 2019). Other mowers are larger. Turflynx has a mower targeted for golf courses that is sized like a conventional riding mower with a broader mowing path and precise path following capabilities (Turflynx, 2019). Spider mowers have a low center of gravity to accommodate steeper slopes and remove humans from dangerous situations (Figure 24b). There are also current technologies to retrofit existing equip- ment (deck sizes up to 72 in) with GPS positioning, LiDAR, and bumpers for proximity sensing (Botix Automation, 2018). Airports such as Atlanta Airport already use remote mowers on steep slopes. Solar-powered electric robotic mowers have been successfully deployed at Stavager Airport Sola in Norway, where 17 Bigmow robot mowers cut 70 acres of airfield turfs; Bigmow robot mowers are a product of Belgian firm Belrobotics, a sister company to the U.S. company Echo Robotics (both are part of the Yamabiko Corporation). Bigmow is a meter-wide autonomous mower that can cut an area of about 5 acres per charge (Belrobotics, n.d.). Using sonar and bump sensors, the robot avoids obstacles (similar to that of a Roomba) but can be programed to follow set paths. This mower is comparable to the Echo Robotics TM-2000, which is currently used for athletic fields in the United States. The mowers are fully autonomous and utilize an electromagnetic field (buried wire perimeter) for boundary designation to keep the mower in a certain area. When the battery gets low, the mower is programmed to return to a field charger (120 volts), eliminating the need to return to an operations hangar for charging. The airport in Sola uses (a) (b) Figure 24. AGVT Mowers: (a) electric mower attached to in-ground charging station with remote monitoring and control and (b) remote control mowers can handle slopes up to 40 degrees or up to 55 degrees with stabilizing wench. Photo: Echo Robotics, 2019; Spider, 2019.

Detailed Evaluation Results 79 photovoltaic solar panels for charging the mowers and for powering the inductance in the boundary wire, which supports sustainability. At the airport in Sola, the mowers are used throughout the airside. Mowers cut in a random pattern in each 5-acre section, and boundary wires are buried up to 4 inches away from the runway edge. Redundant protection to ensure the mowers stay off the runways include a physical boundary for the mower that is safe for aircraft. Another protection to ensure boundary integrity near the runway is that the mower will shut off automatically if a boundary line is crossed. Autonomous lawn mowers are increasingly available for personal use. In addition to the products using technologies described above, the Terra™ from iRobot (Ackerman, 2019) uses a radio beacon for navigation and geofencing, and is able to automatically plan and follow mowing patterns. Edge wires buried along the boundary emit a high frequency signal and are used for geofencing on many commercial products. Residential mowers are also connected with popular technologies such as Amazon’s Alexa, which is consistent with control for other home appliances ranging from HVAC to garage doors and home security systems. These examples illustrate the progress of automated mowing in recent years, and the poten- tial for successful implementation at airports. For this evaluation, low-profile mowers with a 36 inch (1 meter) cutting path and solar charging are considered. One of the expected benefits of small mowers is the increased likelihood that they could safely operate in the RSA coincident with aircraft operations, and a reduced likelihood of damage to airfield lights. Although not evaluated, it would also be possible for larger AGVT mowers to operate at night or other times when restrictions on aircraft activity would have a reduced impact. Automated mowers with no driver would include multiple mowers that each operate inde- pendently. The mowers would operate in constrained conditions including a limited geo- graphic area, specific hours of the day, and during certain weather conditions. Based on current technology, the mower can operate in automated mode in some domains (e.g., within a fixed geographic area) but in some conditions may occasionally need to turn over control to a person or operator. Instead of having a “safety driver” in the vehicle, the control would be provided remotely, such as from a computer in the operations center, or from a cell phone, tablet or other portable remote device. Central control mowers would include multiple mowers controlled by a central computer with the possibility of remote intervention by a human, if needed. The mowers could each be assigned different paths or could operate in formation to cover a designated area or given path. In both cases, the operating conditions are such that the remote safety operator would not be required or expected to take over quickly in an emergency situation. Many of the oper- ating characteristics, benefits, and obstacles would be similar for automated with no driver and for centralized control. Central control mowers would reflect greater fleet management capabilities. Implementation Scenario The proposed implementation of automated mowers with no driver would be full deployment. The proposed implementation of central control would be an investigation. Operational Area The operational area for automated mowers with no driver and central control mowers is the turf areas on the airfield maintained by airport operations personnel. In both cases, AGVT mowers would first be implemented in more remote areas of the airfield, and upon confirmation

80 Advanced Ground Vehicle Technologies for Airside Operations of operating characteristics, they can be implemented closer to the runway, and eventually in the RSA. Initial mowing areas should have relatively low grade, and should not have NAVAIDs or other critical infrastructure that could potentially be damaged and would impact aircraft operations. AGVT mowers should not maneuver on runways or taxiways to reach mowing areas without human intervention (e.g., an operations worker or someone providing remote oversight); the person will be required to provide coordination with ATC using conventional protocol. Airport Characteristics AGVT mowers may be utilized at any size airport, commercial or GA, which mows turf regularly. The potential for solar power AGVT mowing may be particularly attractive to airports that prioritize sustainability. AGVT mowing may be more attractive to airports that have staffing limitations, currently use contract mowing, or can reassign current mowing personnel to other duties. Airports that utilize the same personnel for mowing in the summer and snow removal in the winter would need to assure their year round staffing needs are met and that personnel currently mowing can be assigned to other duties. On steeper slopes, current technology utilizes remote rather than autonomous or central control, so airfield areas with steep terrain may not be good candidates for the proposed AGVT in the near term. Project Partners Companies currently in or wishing to enter the robotic mowing sector (e.g., Echo Robotics, Husqvarna, John Deere) may be good potential partners for AGVT mowing. Airport tenants that have responsibility for mowing airside turf (e.g., FBOs) may also be appropriate project partners. It may be possible to fund an AGVT mower deployment that uses electric mowers and/or solar power through the FAA VALE Program. The VALE Program is available to commercial airports in non-attainment areas, and eligible projects can be funded with PFCs or AIP grants, both entitlements and discretionary (FAA, 2017b). Key Considerations Key considerations for mowing include the following: • Benefits. The benefits of AGVT for mowing include reduced personnel requirements, envi- ronmental impacts, energy costs, operating costs, and personnel in potentially dangerous environments. AGVT mowers reportedly use 80% to 90% less energy (Turflynx, 2019; Belrobotics, 2019), and as a result significantly reduce energy costs. The reduced need for manpower and reduced energy use significantly reduce operating costs. Some mower models also report reduced maintenance costs due to fewer wearable parts (e.g., 40% reduction in main- tenance cost per Turflynx, 2019). The AGVT mower’s capability for safe nighttime operation may reduce disruption to air- craft operations, not only since it eliminates the need for runway closures for mowing in the RSA but also because mowing even outside the RSA can cause increased wildlife activity that may pose a hazard (see the Technical Memorandum in Appendix N for additional infor- mation). Use of AGVT mowing may reduce the incidence of unauthorized changing in the mower deck height and if mowers are smaller and lighter, mowing equipment may be oper- ated on softer ground than traditional equipment, which may allow more consistent mowing to meet height restrictions and create fewer ruts on the airfield.

Detailed Evaluation Results 81 Both automated with no driver and central control mowers would provide valuable information regarding the use of AGVT at airports with minimal risk in terms of potential disruption to aircraft operations. • Technical feasibility. The technical feasibility of automated mowers with no driver is high given the deployment in other sectors and at Sola Airport in Norway. The proposed deploy- ment utilizes low-profile mowers with GPS for pathfinding and tracking and radar for obstacle avoidance. Each mower would have an onboard computer with a GIS map and zone data. Next to the runway and taxiway, small PVC pipe stakes can be used for redundancy to ensure that mowers do not encroach on the runway. Each mower can be observed or controlled through a cell phone app or a computer. The mower technology is similar for central control, although central control would enhance coordination and reduce the number of mowers needed (due to increased efficiencies). With central control, the central computer provides fleet management and serves as a coordi- nation tool when the fleet is operating normally. After receiving direction from the central computer, each mower’s onboard computer will take over vehicle control, interaction between vehicles within the fleet (collision avoidance and right-of-way consultation), localization, and path-planning functionalities. This would involve significantly less latency on control and reduce the requirement for the communication bandwidth between the central computer and the fleet. This would result in a lower deployment cost and would reserve the communi- cation bandwidth for use in case of an unprecedented incident. The central computer providing control would reside in the APOC, airport operations maintenance office, or other existing space. To transmit information to and from the remotely operated vehicles, a wireless local area network (LAN) can be established via one or more physical interfaces such as WiFi, 4G or 5G, enabled by wireless mesh networking (WMN) technology. For all AGVT mowers, GPS transponders in each vehicle will be used to provide vehicle loca- tion information in conjunction with a GIS airport map that will include airport features and zones, and will be housed on the mower’s computer. Because GPS cannot determine the heading of a vehicle, IMU sensors can be used to provide remote operators with a sense of orientation. Although not evaluated, a car-like mower with a minimum steering radius (aka Ackermann steering) may be able to infer orientation based on previous location this would not be feasible for smaller mowers. Smaller mowers typically use differential drive technology and can steer in place, and change heading without moving, which makes inference of current orientation based on previous location and path infeasible (as a comparison, smart phones typically detect position and orientation with a GPS, an IMU and a compass combined). For the small mowers evaluated for automated control, an edge wire system would be placed at the perimeter of each mowing area in the RSA, and a corresponding edge wire sensor on the mower would serve as a hardware backup for the GPS-based software geo- fencing measure. The edge wire can be powered by renewable energy such as solar power, as used by Echo Robotics (2019) in Norway. For safety, the solar panel powering the edge wire would need to be outside the RSA. Radar is recommended for active collision avoidance, which will draw a higher priority than human input, but lower priority than geofencing. Geofencing will stop the mowers at the boundary. This means a warning from pre-collision detection sensors (radar/ultrasonic sensors) will always override human control input, and a trigger from the geofencing soft- ware or edge wire sensors will always override both human input and collision warnings, and make the vehicle come to a complete stop. This will decrease the likelihood of mower collisions due to human errors, and prevent intrusion of mowers onto the runway. Although not evaluated, ultrasonic sensors could be used for collision avoidance for small profile mowers; similarly LiDAR could be used for obstacle avoidance but is not recommended due to its higher price and the sparsity of information in the vertical direction, which makes it unsuit- able for remote control.

82 Advanced Ground Vehicle Technologies for Airside Operations Although not evaluated, video for a remote view of operations would be recommended for mowers with dimensions larger than a yard (meter) across, since larger mowers provide a better field of view for video and potentially pose a greater threat if geofencing fails. If video is used, visual markers may be appropriate to provide reference points for both remote operators and the camera sensor onboard the robot. Planar angle reflectors could be placed at the edge of designated areas; these are passive reflectors that shine the light back to emitters, which are mounted on the mowers, and provide visual references. While radar is effective for obstacle avoidance, it only provides blurry depth information, which is hard for human operators to visualize. Furthermore, video may be useful if the mower encounters wildlife or another unexpected obstacle. In terms of technology adaptability, optical perception sensors such as cameras and LiDARs were not considered for evaluation since their performance may be compromised in dirty outdoor environments. Therefore, the IMU is used for motion detection, and radar is used for proximity sensor. Alternately, an ultrasonic sensor could be added, although it would require confirmation of effectiveness since airfield turf height (6 to 14 inches) is much higher than athletic field turf height (1 to 2.5 inches) or golf course fairway turf height (less than 0.375 to 0.625 inches depending on the season, type of grass and location). The estimated TRL for automated mowing with no driver is 8 to 9 and the TRL for central control is 7 to 8. • Operational impacts. Operational impacts of autonomous mowing are greater efficiency, reduced staffing needs, increased safety, and an expectation for improved wildlife mitigation. AGVT mowing will allow mowing at night when it will have a reduced impact on aircraft operations and wildlife. Although the mowers are theoretically frangible and can operate near an active runway without being affected by jet blast, this would require confirmation. Confirmation of mowing safety during aircraft operations would provide additional flex- ibility for mowing operations, and may reduce the number of mowers required and allow improved optimization of mowing patterns. Another operational impact of automated and central mowing is that the removal of people from mowing activities may reduce wildlife observations in the field, which may reduce identification of high-risk species and the associated timely response. Similarly, airfield operations personnel may observe a wildlife carcass while mowing, in which case it can be removed and reduce the attraction to scavenger species. Since mowing is a fairly isolated activity and does not require significant interaction between personnel or organizations, the operational impacts of AGVT are otherwise minimal. The proposed deployment does not reflect AGVT mowers crossing active taxiways or runways autonomously; future deployments that integrate coordination with ATC would have signifi- cantly more operational impacts. • Infrastructure impacts. Automated and central mowing would require charging stations for the mowers, boundary wires for mowing areas adjacent to runways and taxiways, and frangible PVC boundaries adjacent to the runways. Power for the charging stations and boundary wires can be supplied by photovoltaic solar panels (or alternately, conventional power supply); this may simplify power requirements in remote areas of the airfield since each solar panel and charging station can be independent. FAA approval will be required for the solar panels, but precedence for airport solar panels does exist, although the solar panels are not necessarily on the airfield (e.g., Indianapolis Airport, Denver Airport, Tucson Airport, Honolulu, Minneapolis-St. Paul, Chattanooga and Tampa all have solar power) (Pickerel, 2016). For central control mowers, additional infrastructure required includes a server for communi- cation and data storage, and base and relay stations, with buried communication cables for the base station. A power supply would be required for the relay stations and the mowers; it may be feasible to use renewable energy such as solar power for the relay stations and mowers, depending on the power requirements.

Detailed Evaluation Results 83 • Stakeholder acceptance and ease of adoption. Mowing is a fairly independent activity; it does not share space with aircraft or ground vehicle operations and there is minimal coordination with other stakeholders. This independence would simplify stakeholder acceptance and ease of adoption. Implementation of AGVT mowing should be coordinated with a qualified airport wildlife biologist and the implementation plan should be documented in the airport’s FAA approved Wildlife Hazard Management Plan. Ease of adoption will be reduced by the need for FAA approval with additional approvals for operation in the RSA when the runway is active. The proposed deployment requires a person to coordinate with ATC when the mower crosses a runway or taxiway, which will improve stakeholder acceptance. Similarly, initiating automated mowing in remote areas and progressing to areas closer to the runway will provide data to substantiate the operating characteristics of the AGVT mowers and increase confidence, which will increase stakeholder acceptance. Resistance from labor will be reduced if the mowers replace mowing contracts or temporary workers, and/or result in worker reassignment to other tasks, instead of employee reduc- tions. Employees and management will support improved working conditions due to a reduction in mowing-related safety concerns and as workers transition to supervising AGVT mowers rather than working in the field. Stakeholder acceptance and ease of adoption will also be supported by the sustainability benefits, including reduced fuel consumption and reduced emissions, as well as the use of solar power. The electrification of mowing not only reduces the airport carbon footprint, but also decreases maintenance costs compared to conventional mowing procedures. Stake- holder acceptance will also be supported by increased safety due to reduced wildlife near the movement areas, based on reports from Sola Airport. Mowing can also be planned to occur when it will have the least impact on wildlife (e.g., nocturnal versus diurnal). Stakeholder acceptance is also supported by the relatively modest infrastructure needs. The installation of solar-powered charging stations will require less labor and eliminate the need for power lines to charging stations. Procedural compatibility concerns focus on main- taining communications with operations personnel and assuring that the mowers stay out of movement areas. Since automated mowers function on their own, their complexity of use is relatively low. Central control mowing AGVT might have slightly higher levels of complexity that are rooted in their control interface and the need to ensure situational aware- ness for the remote operator in the event that intervention is needed. Mowing AGVT are highly trialable and could be phased in gradually, which strongly supports stakeholder acceptance and ease of implementation. Deploying automated mowers initially in remote areas and documenting their operating characteristics will ensure confi- dence as mowing moves closer to the movement area. The results of mowing AGVT are highly observable, with well-maintained grassy areas easy to see, and the operational benefits due to reassigning workers to other tasks highly visible to management. Lastly, the uncertainty associated with the use of mowing AGVT appears to be relatively low; the most uncertain aspect of their operation is how best to integrate the high level of automation into airport operations. With smaller scale automated mowing systems being in existence for almost a decade (e.g., Trimarchi, 2009) and successful use at Sola Airport in Norway, this AGVT appears to be a strong candidate. Individual-level considerations include acceptance by airport operations employees who supervise the mowers, as well as other airport users. The existing systems are easy to use and training is minimal, which combined with removal of workers from field work, would likely increase acceptance by airport operations workers. Incremental deployment in remote areas of the airfield will increase confidence and provide data to facilitate acceptance by other airport users (including ATC and pilots for future deployment), whose main concern will likely be confidence that the mowers will not encroach on the runway or otherwise interfere with

84 Advanced Ground Vehicle Technologies for Airside Operations safe aircraft operations. Demonstration of the frangibility of the mowers (e.g., they would be pushed out of the way by an aircraft) and quadruple redundancy for operation next to the runway (GPS geofencing, boundary wire geofencing, software that shuts the mower off if it crosses the boundary wire geofence and PVC stakes) may also increase confidence and facilitate acceptance. Budgetary considerations may reduce ease of adoption since the mowers do require capital resources for the mowers as well as for the charging infrastructure and power to support the charging infrastructure. • Human factors. Human factors considerations include whether the AGVT mowers reflect appropriate allocation of function, good computer-human interaction, procedural compat- ibility, communications, and compatibility with staffing needs. Given the repetitious and monotonous nature of mowing and the lack of interaction with aircraft or other vehicles, it seems to be an excellent task for automation in terms of allocation of function. The existing use not only for mowing but also for other consumer apps suggests that the computer-human interaction should be adequate. The capability of a person to take over the system appro- priately when needed will need to be demonstrated for both automated control and central control. Additional human factors considerations may become more relevant in the future if the AGVT mowing advances and direct interaction between ATC and the mowers is needed. Potential Challenges The greatest challenges for automated mowing may be the capital investment and the lack of automated mowing in other sectors in the United States. The greatest challenge for central control is that fleet management functionalities need to be developed and integrated in the airside environment. The proposed deployment does not encompass automation for crossing runways or taxiways, which may be necessary for some mowing areas, which would also be a potential challenge. FAA approval for operation in the RSA is another potential challenge long term, although it is expected small profile mowers can be demonstrated to be frangible objects, and initial deployments in remote areas can demonstrate the reliability and safety of mower operation. Snow and Ice Control Description Snow and ice control is critical to ensure safe operations for airports that experience winter weather. Although GA airports without commercial service have lower regulatory require- ments, Part 139 certificated airports must follow well-defined procedures for compliance with FAA requirements. Each airport must have an approved Snow and Ice Control Plan (SICP), which describes equipment preparation, personnel training, criteria or conditions that trigger a response, and operational procedures for snow clearing activities (AC 150/5200-30D, 2016). As stated in AC 150/5220-20A, the goal of winter operations is to maintain the runways in a “no worse than wet” condition during snow events. After clearing the runway with snow removal equipment (SRE), the airport must report the runway condition to pilots using RCAM in accor- dance with AC 150/5200-30D. The RCAM score is based on observed conditions but can be downgraded based on the measured friction, pilot reports, or the braking deceleration. The proposed AGVT applications for evaluation include: • Platoon with driver in lead • Remote operation at a GA airport There is a variety of SRE and different equipment and techniques are used for different pave- ment and weather conditions. Removal strategies consider a variety of factors including the

Detailed Evaluation Results 85 amount of contaminant (snow, slush, or ice) on the pavement, rate of snowfall, moisture content of the snow, pavement temperature, air temperature, and wind, as well as predicted conditions based on the weather forecast. Successful snow removal relies on an accurate assess- ment of current conditions, and the experience and capabilities of the operator and snow removal team to execute the airport’s well-defined plan for snow removal. In many cases, a platoon of SRE comprised of different vehicles with different tasks may be used to clear the pavement. The first vehicle may be a snowplow that removes the first half inch of snow. The second snowplow in the line may move the second half inch of snow. The third machine may be a rotary plow that clears the windrow of snow created by the first two snow- plows. Airports with significant snow may also purchase multi-function equipment to reduce the need for drivers. Examples of SRE are shown in Figure 25. Equipment identified by FAA includes sweepers, brooms, displacement plows (aka snowplow), high-speed rotary plows (which blow the snow), material spreaders and carrier vehicles. A large airport such as Denver International Airport has a trained crew of 500 people that make up nine teams (serving airside and landside) and 250 pieces of airside SRE equipment including blowers, brooms, blades, plows, runway sanders, snow melters, chemical trucks, loaders with box plows, bobcats and bobcats with box plows, and multi-function equipment that can plow, sweep and blow snow with a single piece of equipment (Denver Department of Aviation, n.d.). At the other end of the spectrum, a GA airport may mount blades, brooms and spreaders on existing trucks and tractors, and may rely on the airport manager or a contract employee to drive the snowplow. (a) (b) (c) (d) Figure 25. Snow removal equipment: (a) snow blower, (b) high-speed rotary plow with blower, (c) broom, and (d) multi-function snow equipment. Photo: Denver Department of Aviation, n.d.; Salt Lake City International Airport, 2018.

86 Advanced Ground Vehicle Technologies for Airside Operations Information related to winter operations is provided in the following advisory circulars: • AC 150/5200-30D, Airport Field Condition Assessments and Winter Operations Safety, 7/29/2016. This provides information about winter operations, including the Snow and Ice Control Plan, supporting committee, activities, and RCAM. Information to support the acquisition of SRE and ensure the adequate storage for SRE is provided in the following advisory circulars: • AC 150/5200-20A, Airport Snow and Ice Control Equipment, 9/24/2014, provides an overview of Snow Removal Equipment (SRE) and information regarding SRE acquisition. • AC 150/5220-18A, Buildings for Storage and Maintenance of Airport Snow and Ice Control Equipment and Materials, 9/4/2007, provides information for buildings to provide storage of SRE and consumable materials (e.g., deicing chemicals), as well as space for winter storm personnel. The proposed project would leverage AGVT to enhance existing strategies for snow removal as described in the airport’s Snow and Ice Control Plan (SICP) (FAA, AC 150/5200-30D, 2016; FAA, 2016b). A conventional platoon of SRE equipment is a common strategy for snow removal at airports. Integration of AGVT to support a SRE platoon has been demonstrated in concept with operation during clear conditions, as shown in Figure 26a for a demonstration by Daimler at the former Pferdsfeld Airbase in Bad Sobernheim in 2017. A platoon of two self-driving snowplows (with no driver in either cab) was demonstrated at Fagernes Airport in Leirin, Norway, in March 2018. Figure 26c illustrates the cab of a Yeti self-driving snowplow, which was developed to handle the worst conditions possible and utilizes 4G technology and real-time kinematic (RTK) GPS for positioning (Klingenberg, 2018). To identify the path the vehicle should follow (i.e., to “train the vehicle”), an operator will drive the vehicle in the intended domain and the computer will plot points on a map that it will follow during normal operation (Klingenberg, 2018), providing path precision of about 2 inches (4 to 5 centimeters). Building on this demonstration, two self-driving snowplows were deployed with six manually driven snowplows for winter operations at Oslo Airport Gardermoen in the 2018/2019 winter season (Yeti Snow Technology, 2019). Yeti Snow Technology is jointly owned by technology company Semcon and SRE manufacturer Øveraasen; their partnership with Norwegian airport operator Avinor ensures expertise in all relevant domains. There are other partnerships in this domain, such as Northstar Robotics, Airport Technologies, and Winnipeg Airports Authority (Kozak, 2018). The Daimler snowplow demonstration also uses GPS pre-mapped routes (De La Bastide, 2017). The use of automated technology to follow a prescribed path represents a more mature but less advanced technology than autonomous operation, which would require a more sophis- ticated operational response to the environmental conditions (Klingenberg, 2018). A long-term goal for AGVT to support a SRE platoon may include a driver in the lead with multiple vehicles in formation or a fully automated platoon with a human providing remote oversight and intervention when needed. The formation would be able to avoid internal collisions as well as complete snow removal in a designated coverage area following a defined path. Ideally, the snowplow path reflects the minimum overlap with the previous plow path to ensure that the plow does not create a windrow of snow on the path that has already been cleared. In the long term, AGVT for winter operations may also be supported by technologies such as optical-based runway surface condition assessment, as shown in Figure 26b. This system, called RCAMPro, uses cameras and optical-based devices mounted on the roof of a snow- plow (or other airfield vehicle) combined with machine learning and AI to determine the presence, type and depth of ice or snow on the runway (Team Eagle, 2017b). This technology

Detailed Evaluation Results 87 can be used to report the percent of runway contaminant for RCAM reporting, and in the future, could be combined with AGVT to enhance the autonomy and/or effectiveness of platoon, and support strategic and efficient placement of deicing material on the pavement. The current technology is in its third generation and utilizes optical devices incorporating gated aperture technologies for improved situational awareness capabilities in low visibility conditions (Team Eagle, 2017b). Interest in AGVT for snowplows has been broad and continuous. Figure 26d shows one of the contenders in the ninth annual ION Autonomous Snowplow Competition which was held in 2019 in Ontario, Canada. This contest provides an opportunity for fully autonomous snowplows designed and built by collegiate teams to compete to remove snow from a designated path while also promoting navigation-based technologies (ION, 2019). Descriptions and evaluation of the proposed AGVT deployments for snow and ice control include platoon with a driver in the lead and remote operation, as described below. Platoon with a driver in lead includes one lead vehicle and one platooned vehicle, which would be programmed to safety follow at an appropriate distance and offset. The driver in the (a) (b) (c) (d) Figure 26. AGVT for winter operations: (a) automated plow follows lead truck with driver during test, (b) optical-based runway surface condition assessment system, (c) Yeti Snow Technology’s snowplow cabin with automated function equipment, and (d) autonomous snowplow competition. Photo: Esterdahl, 2018; Team Eagle, 2017b; David Pike, ION Autonomous Snowplow Competition, 2019.

88 Advanced Ground Vehicle Technologies for Airside Operations lead vehicle ensures safety since they can monitor the progress of the platoon, as well as interpret and respond to any unexpected circumstances. This may be especially valuable in early imple- mentation since snow removal has been described an art as well as a science, and an experi- enced snowplow driver adapts their path to reflect the depth and density of the snow, as well as the wind conditions, which can change quickly. Winter operations may also need to contend with multiple contaminants (e.g., wet snow and slush) and/or layered contaminants (e.g., water over compacted snow, dry or wet snow over ice). The presence of a lead driver would allow the strategy and platoon parameters to be modified when needed and reduce the reliance on the automation. Remote Operation reflects a functional concept in which a manager or worker at a GA air- port could initiate the operation of a single snowplow on a predefined path from an office, or from another location using a smart phone. This would significantly reduce the labor requirements for snow removal, and increase the availability of the runway during the winter. Implementation Scenario The proposed implementation of platoon with driver in lead and remote operation would each be a demonstration project. A demonstration project is appropriate since the technology is still advancing and has not been proven for regular operations. Operational Area The operational area for evaluation may include runways, taxiways, ramps, and/or access roads. The operational area will exclude areas with aircraft (e.g., ramps with parked aircraft or active runways). The operational area may include runways and taxiways closed to aircraft activity (e.g., closed for snow removal with associated NOTAM), and may be in the movement or non-movement area; all required radio communications with the ATC (platoon) or on the CTAF (remote operation) will be by a person. Airport Characteristics AGVT snow removal would be of interest to airports that experience moderate or greater levels of snowfall, and more frequent snowfall will likely increase the demand and benefits of AGVT. Airports that have labor shortages or other challenges that make it difficult to staff winter operations would be compatible with AGVT for snow removal, as would airports in which the environment is harsh enough to present safety concerns for the snowplow operator. In the near term, the proposed deployment of platoon with a driver in lead is recommended for a non-hub airport or small-hub airport. Implementation at non-hub or small-hub airports would allow the system to be tested when there are no scheduled air carrier operations, which would minimize the likelihood of disruption to commercial air service. It would also be appro- priate to test a platoon at a medium- or large-hub airport on a crosswind or secondary runway or taxiway that is not being used and where platoon operations will not interfere with scheduled air carrier operations. In the near term, the proposed deployment of remote snowplow operation is recommended for GA airports that serve corporate aircraft and have a moderate amount of traffic but do not have control towers. Corporate aircraft may prefer to land at GA airports but are not currently able to do so when the runways are closed due to snow. These GA airports are also a good proving ground for development since any technical problems that result in delayed snow removal will not present problems for scheduled commercial aircraft. The expense of AGVT may not be justified for a very low volume GA airport.

Detailed Evaluation Results 89 Long-term, remote operation of SRE may be appropriate for any airport that performs snow removal, and a platoon with a driver in the lead may be appropriate for any airport that uses a platoon of SRE vehicles. For airports that use the same personnel for winter operations tasks and summer mowing tasks, it might make sense in the long term to pair implementation of AGVT for winter operations with AGVT for mowing activities to balance labor requirements throughout the year. Project Partners Snow truck manufacturers (e.g., OshKosh snow trucks) may be appropriate partners for the implementation of remote operation and platoon with a driver in the lead. Automotive manufacturers and supporting vendors (e.g., Daimler, Mercedes-Benz and Semcon) and truck platoon vendors (e.g., Peloton Technology and TomTom) may be good partners for platoon with a driver in lead. Key Considerations Key considerations for snow and ice control include the following: • Benefits. The expected benefits of platoon with a driver in lead include increased efficiency and increased safety. Increased efficiency is due to reduced labor requirements to operate two pieces of equipment, and because the vehicles can be separated an optimal distance due to coordinated control. Increased safety may be realized since the platooned vehicle will have automated braking based on the braking in the lead vehicle, and since it removes a person from working in harsh, winter conditions, and reduces the potential for driver fatigue related to overtime, which can contribute to fatigue related errors and accidents. The potential benefits of the remote operation of SRE include increased efficiency, service, and safety. Efficiency is increased because labor requirements are reduced. Greater service is provided because more efficient snow removal increases the accessibility of GA airports in winter weather. Increased safety would be realized because remote operation removes people from harsh, winter environments and reduces issues related to driver fatigue and the need for snowplow operators to be on-call. The benefits of reduced labor requirements for remote operation will be realized only if the system is proven reliable, and runway and taxiway closure via a NOTAM is considered an adequate safety protocol to ensure that the snowplow will not interact with the aircraft. Long term, hard cost savings may be realized if there is a reduced need to contract out snow removal activities or if there is reduced need for labor and labor overtime. Depending on the employment framework for winter operations, soft cost savings may be more likely, and would reflect the ability to shift personnel time away from snow removal activities to other airport duties. AGVT for snow removal will also provide benefits to the extent that sensors may provide increased safety and allow operation in whiteout conditions that prevent humans from safely operating SRE. A final benefit of both projects is the provision of useful data to support additional airside AGVT. • Technical feasibility. Successful demonstrations of platooning technology in the roadway sector and agriculture sector (e.g., tractors and combines) as well as demonstrations of platooning technologies and self-driving snowplows in the airport sector suggest the technology is advancing and demonstration at an airport in the United States is feasible. The greatest challenges may be inter-vehicle issues including collision avoidance for following vehicles and possible issues with the latency of the 4G network. Other issues include deploy- ment in the airport environment in the United States, which has not been attempted. Platoon with a driver in the lead would have an estimated TRL of 7 or 8.

90 Advanced Ground Vehicle Technologies for Airside Operations For the platooning operation, the driver in the lead will drive the vehicle while one (and in the future more than one) vehicle will follow it in a formulation with predefined or real-time adaptive following distance and lateral offsets. Proposed technologies include radar for collision avoidance, GPS for positioning and software geofencing, and DSRC framework for inter-vehicle and vehicle-to-ground control communication. The following vehicle can obtain motion information (position, speed, pedal and steering information) from the lead vehicle and other following vehicles via the DSRC framework. This information will be the basis for any required adjustments to motion controls. The driver of the lead vehicle can obtain the motion and position of the whole fleet and the information can be displayed to the driver through Driver’s Enhanced Visual System (DEVS) for enhanced situational awareness. Addi- tional information about DEVS is provided in AC 150/5210-19A. The proposed communi- cation leverages the cellular network (3G/4G/5G) for long range notification, such that if a vehicle leaves the platoon, notification will be sent to the APOC. For remote snowplow operation, analogous technologies have been used in related fields, such as the remote and automated operation of UAS. The proposed remote operation provides the benefits of automated operation with the capability for a person to respond remotely. Long term, it may allow more efficient operation since the equipment no longer needs to safely convey a person. Remote operation may also allow a single person to monitor multiple pieces of equipment, increasing efficiency. Remote operation does not align with the SAE frame- work for automation levels, and the most analogous level would depend on the degree of automation and the role of the remote driver or operation. In this case, remote operation aligns with L3, since the system would operate autonomously without human intervention and does not require a person to continuously monitor the system during operation. A remote person would intervene, when necessary, but the operational framework is such that it would not be an emergency intervention since the runway would be closed for snow removal. Automated snowplows that are initiated remotely would have to be compliant with an L3 automation level, with a remote operator available as a backup measure in the event of a system malfunction. Operation would allow a snowplow to be initiated and observed from a remote location, with remote intervention when required. GPS would be used for soft- ware geofencing, and radar or sonar could be used for obstacle avoidance. A central server is required for the remote initiation of the mission, and cellular networking is required as well under the remote scenario. Long term, DSRC would be used to coordinate multiple SRE vehicles that are initiated remotely, and additional sensors would be required for independent autonomous operation. For both cases, a schematic layer of the airport GIS map including area designation (e.g., movement area) and coordinates of the associated borders is used to support software geofencing. In the long term, AGVT to support snow clearance around NAVAIDs would require a high level of accuracy and equipment would require confidence to ensure preser- vation not only of the visibility of the NAVAIDs but also that the equipment would not inadvertently damage the NAVAIDs during snow removal activities. In terms of technology readiness, a vehicle platoon with a driver in the lead reflects CV technologies, which have been tested during demonstration projects on roadways, at airports (e.g., snowplows in Norway and Canada), and in agriculture. Daimler has developed and tested autonomous snowplow trucks (Etherington, 2017). These industrial vehicles are platooned or have leader-follower routines to move along GPS pre-mapped routes. As described, the Arocs fleet is networked via radio signals, with a control panel in the leader vehicle, but the stan- dards of the communication are unknown. The control panel involves engine start/stop, parking brake, steering, engine control, steering, service brake, transmission, differential locks, lighting, as well as the mounted sweeper blower. All trucks are identically equipped, which means any vehicle can be the leader truck. This is being done in collaboration with Frankfurt Airport and is called Automated Airfield Ground Maintenance (AAGM) (Tomas, 2017).

Detailed Evaluation Results 91 The solution from Yeti provides a pre-programmed approach (Mogg, 2018). The truck only utilizes GPS signals to obtain coordination, and does not rely on any visual localization or plan- ning methods. Such simplicity of the system makes it robust in snow-covered surroundings; however, the driver is responsible for collision avoidance and the strategies to ensure collision avoidance of following vehicles have not been reported. Therefore, the technology is still in experimental phase (TRL 7) and further integration is needed for on-site applications. Millimeter wave radar and sonar have been tested in operational environments (TRL 8) on passenger vehicles for collision avoidance. GPS sensors are readily available (TRL 9) in industries such as aerospace. Inter-vehicle communication has been tested in experimental environments and therefore considered as TRL 7. FAA has established standards for airport GIS (AC 150/5300-16A, 2007; AC 150/5300-18A, 2009; and AC 150/5220-26, 2011) and airport information is continuously contributed by airports into the FAA database. This would suggest a TRL 8 to 9 for the GIS system, although for each airport it would be necessary to confirm that the data in the GIS maps is adequate to support the requirements of the proposed AGVT. Cellular network with 4G technology is already widely used in telecommu- nication, which is of TRL 9. The latency of 4G networking is typically 50 milliseconds (Enbuske, 2016) and should be considered when developing the safety margin. 5G (TRL 7 or 8) would have significantly lower latency (about 1 millisecond) and therefore in the long term will be more suitable for real-time remote operations. In terms of technology adaptability, it is expected that sensor technology can be generally adapted for this application. GPS and IMU are preferred for localization of vehicle because they are not affected by the surroundings compared with perception-based odometry such as SLAM algorithms, which could be brittle under whiteout conditions. One concern related to the surrounding perception sensor is the reliability of the sensor in freezing conditions and its effectiveness in white out conditions in which sighting would be obscured. To perform platooning, it is feasible to track the distance between the leader and follower vehicles and ensure a constant and safe distance. Another method has been proposed by Taylor et al. (2019) whereby the control commands of the leading vehicle are shared and executed among the entire platoon fleet, coordinated according to the order of following to maintain the formation when turning. As sensor data and the associated AI develops, stopping distance algorithms may be adapted to reflect current conditions. Team Eagle is investigating adap- tive stopping distances that utilize real-time pavement friction data calculated based on vehicle performance characteristics (e.g., sensor input that considers the distance traveled when brakes are applied, and/or compare tire rotation for powered versus non-powered wheels). Snow removal is also complicated by the fact that different machines in the platoon have different functions. The first snowplow in the line may move the first half inch of snow. The second snowplow in the line may move the displaced snow from the first snowplow as well as the half inch of fallen snow already lying in front of the second plow. The third machine may be a rotary plow that clears the windrow of snow created by the first two snowplows. To achieve precision control of snowplows, force sensors can be applied to the joint of the snowplow to measure how much force is applied to remove the snow; this can be compared with theoretical or numerical simulation models, and used to adjust the equipment positions. The GIS map for software geofencing could be stored onboard since the airport layout is not likely to change over one winter, and it can be updated during vehicle maintenance. To enhance driver’s situational awareness under whiteout conditions, it is feasible to implement DEVS, which can project a simulation of the environment (including airfield markers and the position of other vehicles) on the lead driver’s windshield, or by using an augmented reality headset such as Hololens. It may be appropriate to use the technology in the APOC for platoon oversight or remote operation. In terms of technology safety and system redundancy, platooning with a lead driver can use DSRC technology for inter-vehicle communication to coordinate vehicle movements.

92 Advanced Ground Vehicle Technologies for Airside Operations These links can share control status of the leading car and sensor data, to facilitate planning of coordinated trajectories. The intercommunication can be established only as a local network, and physically isolated from the external internet, which will reduce the risk of malicious hacking. The cellular network (3G/4G/5G) is only used for long range notification when necessary, such as when a vehicle leaves the platoon, in which case notification will be sent to the operations center. The difference in the angular velocity of the powered axle and following axle of the snow removal vehicle can be used to determine the friction coefficient of the pavement (lower friction will result in a greater difference in angular velocity). The surface friction coefficient will then be used to determine the distance for following vehicles, reflecting the real-time braking distance and differences in braking distances for different vehicles as well as any delay associated with braking control. Vehicles without a human driver, such as the following vehicles in a platoon or the remotely initiated snowplow, need to have redundant mechanisms for safety-critical sensors exposed to the exterior (e.g., triple modular redundancy) (Lyons and Vanderkulk, 1962) for radar transceivers, so the vehicle is not totally incapacitated in the event of a sensor failure. Fall- back options should be set up for collision-related and geofencing sensors so the vehicle is able to safely exit the safety-critical area and stop in case of a major malfunction. For example, as a safety fallback option for collision avoidance, near to mid-range distance sensors such as ultrasonic (approximately 30 ft or 10 m range) and/or laser range finders (approximately 300 ft or 100 m) could be adopted for imminent collision avoidance. • Operational impacts. The operational impact of either of the proposed deployments is rela- tively low since they are demonstration projects rather than full deployment. For the platoon, coordination and communication with ATC would be with the lead driver (or airport opera- tions lead, depending on the existing protocol) to minimize operational impacts. Further- more, the time and location for the demonstration projects can be selected to have a minimal impact on airport operations. Since runways are closed for winter operations and since SRE has no direct coordination with aircraft or other airfield users, the operational impacts are further reduced. In the future, if a platoon with one or more follow vehicles is integrated into a larger fleet that includes conventional manned vehicles (the current situation in Norway), there would be significantly more operational impacts. The operational impact of remotely operated SRE is low since deployment is proposed for a GA airport. A NOTAM would need to be issued and a person would need to broadcast on the CTAF to ensure pilots are aware that SRE are active on the runway, protocol that is consistent with the current framework for snow removal. At a larger certificated airport, the Snow Control Center issues NOTAMs and coordinates with ATCT, air carriers, and other tenants. In the long term, it may be possible to integrate some or all of these activities into the remote deployment or automation (e.g., automation would issue a required NOTAM associated with runway closure for SRE activity). For remote snow removal, the primary operational impact of remote operation is expe- dited commencement of snow removal activities, even when the snowplow operator is not available on-site (e.g., after hours). This capability could facilitate snow removal at night or during early morning hours, and would have a positive operational impact since the airport could open sooner and have increased hours of operation during the winter. In both cases, there is no current provision for the AGVT to conduct an assessment of the runway condition using the RCAM, which is necessary to provide information to pilots for aircraft operations. Based on the current work underway (e.g., by Team Eagle as shown in Figure 26b), this may be a function that could be automated in the long term. • Infrastructure impacts. Infrastructure impacts for the platoon with a driver in the lead are minimal since the lead vehicle can communicate with the platoon vehicle using DSRC. Both the platoon and the remote snowplow will require a current GIS airport map on the in-vehicle

Detailed Evaluation Results 93 computer. GPS transponders in each vehicle will be used to provide vehicle location infor- mation, which will be considered in conjunction with features and zones identified in the GIS airport map on the in-vehicle computer. This will support software geofencing as well as provide convenience for airport management, who can use it for remote oversight. Real-time transmission of control commands can be enabled from central control (or a cell phone) in the event of an operational anomaly (e.g., an obstacle is encountered and the vehicle cannot resolve the situation by itself). This relies on adequate cellular service (or WiFi) on the airfield wherever the SRE will operate. For the proposed demonstration project, the location can be strategically selected to ensure adequate communications coverage. In the long term, if some areas do not have adequate coverage, it may be necessary to provide signal relay stations to ensure communications capabilities. Central facilities for remote initiation and monitoring are relatively modest, and would require space and facilities analogous to a personal computer. • Stakeholder acceptance and ease of adoption. The proposed deployment of platoon with a driver would have generally positive implications for stakeholder acceptance and ease of adoption. The presence of a driver in the lead vehicle simplifies communications with ATC, other airfield vehicles, and the APOC, as well as ensuring safety for unexpected conditions. Functionally, platoon with a driver in lead is analogous to the operation of multi-function equipment, which is commonly used and accepted; this analogy may facilitate implemen- tation and reduce potential resistance. In the long term, a full deployment with additional following vehicles in the platoon may meet greater resistance if it would displace existing personnel, but given the progress of technology it is likely that platoon length and AGVT capabilities would increase incrementally over time. Changes could potentially be accom- modated with attrition rather than layoffs, which would increase acceptance to some extent. Demonstration of remote SRE operation at a GA airport would be expected to have high stakeholder acceptance and ease of adoption. The benefit of increased capability to clear the runway in winter weather would be significant for both airport management and airport users. Since snow removal is not a full-time position at most GA airports, it is likely to reduce the personnel headcount. Both systems will have increased acceptance to the extent that they reduce exposure to harsh environmental conditions and the need for excessive overtime, which would reduce employee fatigue. Similarly, demonstration of system reliability will be critical for acceptance by operators and other airport users. Additional organizational and individual considerations for stakeholder acceptance and ease of adoption are discussed below. In terms of organizational level considerations, in the long term, the need for regulatory approval and potential labor opposition represent possible impediments to snow removal AGVT. These are not expected to be obstacles for the proposed demonstrations, or for deployment of a remote snowplow at a GA airport in the long term. In the roadway sector there is evidence that labor unions oppose platooning (Teamsters for a Democratic Union, 2019), but this has not been substantiated in an airport context. Perhaps equally likely, at some city airports there may be political pressure from city council members or aldermen to maintain existing levels of airport employment for local constituents. Labor considerations would be less of a constraint at airports that currently contract for snow removal or utilize part-time or seasonal help. Airports interested in adopting platooned SRE with a lead driver might see some labor or political resistance if personnel are displaced by increasingly longer SRE platoons. Similar concerns would arise as remote SRE advances and it becomes feasible to remotely initiate a fleet of automated SRE. Labor and political resistance would likely be minimized if personnel are reassigned to other airport duties instead of laid off. Airport management and sponsors would be expected to support AGVT for winter opera- tions when the technology advances and it can reliably support snow removal and provide

94 Advanced Ground Vehicle Technologies for Airside Operations increased efficiency, safety, and cost savings. Increased safety associated with a reduced need for overtime may be a particularly compelling component of these technologies, since worker fatigue can cause costly errors or accidents, and few workers embrace the long hours that may extend for days during a severe winter storm (illustrated by the need for and provision of sleeping rooms for winter operations personnel at some airports). AGVT SRE might additionally be able to operate in conditions when human workers might have difficulty performing it (e.g., whiteout conditions or during extreme cold). The proposed demonstration projects of AGVT for snow removal support trialability (i.e., ease of trying an innovation without fully implementing it). Long term, with full deploy- ment, the capability for snow clearance and the benefits from reallocated or reduced labor are highly observable, which bode well for organizational acceptance. Finally, despite both snow removal AGVT applications being technically feasible, there is some uncertainty involved in their use to remove snow and ice in an airport environment. Human snowplow operators are able to adapt their path dynamically, based on their experi- ence and intuition to deal with changing environmental conditions. Until this flexibility can be incorporated, AGVT might have trouble duplicating the snow removal performance of human workers. Both snow removal AGVT applications will need to be shown to meet performance standards through demonstrations, and should provide useful data on how best to deploy automated platoons in airport movement areas during this process. In terms of individual characteristics that might affect individual-level acceptance of snow removal AGVT, longer-tenured employees used to status quo procedures for snow removal might be resistant if there is a perception that AGVT complicates the snow removal process. Platooned snow removal AGVT might be misused or abused if it is mandated or seen as a threat to jobs (Parasuraman and Riley, 1997). As AGVT are designed specifically for the task of snow removal, they should have a high level of task-technology compatibility. The snow removal performance of either snow removal AGVT application will determine the level of trust the individual worker places in them. Trust declines if automation fails to meet the performance expectations for tasks that can be easily performed manually by the operator (e.g., Madhavan, Wiegmann, and Lacson, 2006), so trust in AGVT for snow removal will suffer if people need to touch up or completely redo any of the snow removal tasks following the AGVT snow removal. Snow removal AGVT reduces the individual worker’s exposure to harsh winter climates, something that could increase their perceived usefulness. The capability for remote AGVT operation could also enhance perceived usefulness. The ability to reassign workers to other tasks during busy periods or at understaffed airports, or the simplification of SRE platoon operation could also support a high perceived usefulness. Proper system design and adequate training will be needed to assure perceived ease of use. With remote operated snow removal AGVT, training should focus on how to intervene when necessary. With platooned snow removal, training will need to focus on how the lead driver needs to interact with the auto- mated system to assure adequate snow clearing, system integrity and safety, including how the following vehicle path should be adjusted to adapt to the environment, activities that were previously accounted for by experienced human operators. • Human factors. Human factors considerations for AGVT include allocation of function, compatibility with procedures, situational awareness, communications, teamwork, operational suitability, compatibility with workload and existing procedures, and reduction in human error. For a platoon, key issues include whether the lead driver can manage the workload and maintain situational awareness. Situational awareness and response is enhanced by the use of a driver in the lead vehicle, and also by existing technologies such as DEVS. As the AGVT advances and if it becomes integrated with conventional human-driven equipment, issues such as situational awareness, teamwork and operational suitability may

Detailed Evaluation Results 95 become increasingly important. In the long term, these issues may become less relevant if winter operations become fully automated; however, compatibility with existing procedures such as communication with ATC and training and situational awareness for human operators who may need to take over operation could become increasingly important. As automation advances, consideration will need to be given to whether equipment should be capable of operation by automation or by a human, to allow manual operations, if necessary, since an automation failure could otherwise be catastrophic for a large airport. The potential for AGVT to support winter operations is significant since AGVT can support situational awareness through technologies that enhance vision, provide augmented reality and increase operational safety through information collected by sensors about weather, pavement friction and even driver fatigue. Advanced sensors such as radar support snow removal in blizzard conditions and at night. Advancements in gated aperture applications for cameras and LiDAR may also reduce the negative impact of glare from snow and ice and enhance visibility during winter storms, which will reduce accidents and incidents caused by human error in low viability conditions. AGVT allow reduced reliance on human operators, which also may reduce fatigue as human drivers spend less time in the vehicle cab and can instead provide remote oversight from an operations center. Fatigue is a significant human factors concern and a leading cause of accidents and injuries. Long-term, integration of weather information, data from sensors in the infrastructure (e.g., temperature sensors in the pavement) and data from AGVT sensors (e.g., force sensors on the snowplow blade) may collectively be used to support improved decisions and response during the execution of snow removal activities. Potential Challenges The greatest challenges may reflect that execution of the snow removal plan varies depending on the kind of snow, wind conditions, air temperature, pavement temperature, weather forecast, and other environmental conditions. Sensor technologies have not been developed (nor has research and development been initiated) to differentiate the variabilities in snow type, density, and viscosity, which have a significant effect on snow removal, including the displacement characteristics of the snow window after plowing and the casting characteristics from the rotary plow. To illustrate the variable characteristics of snow, consider the following information pro- vided by an SRE equipment developer (S. McKeown, personal correspondence, June 28, 2019). The viscosity of water at 20° C (68° F) has a nominal value of 1 mPa, but the viscosity increases to 1.8 mPa as the temperature decreases to 4° C (40° F). The variability increases by orders of magnitude for different kinds of slush and snow. For example, slush may have a viscosity of 10 mPa; wet snow may have a viscosity of 10 billion mPa and dry snow up to 1018 mPa. Decisions such as the overlap path relative to the previous plow may reflect driver intuition and experience as well as quantitative data such as temperature, wind speed, and wind direction, as well as the aforementioned snow characteristics that cannot currently be ascertained with sensor data. Increased integration of AGVT will need to rely on people until advanced sensing and decision-making systems are able to perform as well as humans can, using experience and intuition. Similarly, few technologies have a high TRL for more than one or two vehicles working together, despite significant research, development and deployment in other sectors such as mining and agriculture. This limits the likelihood of a successful automated SRE platoon in the near term beyond the two vehicles proposed in this evaluation. In the short term, lower levels of AGVT may provide significant benefits for increased situational awareness and collision avoidance, and the resulting data may support the integration of more advanced technologies in the future.

96 Advanced Ground Vehicle Technologies for Airside Operations Perimeter Inspection Description Robust airport perimeter security is needed to ensure public safety and airport security. Airport perimeter security can help prevent events such as the 15-year old boy who climbed a fence at San Jose and flew to Hawaii in the wheel well of an aircraft in 2016 (U.S. GAO, 2016b), and more recently, the 19-year-old man who scaled a 12-ft fence at Atlanta’s Hartsfield- Jackson Airport and ran across a runway before jumping onto the wing of a Delta airplane (Kvetenadze, 2018). At many airports, access to the air operations area is controlled by fencing to ensure the safety and security of aircraft operations, public protection and safety, and compliance with Trans- portation Security Regulation 1542 requirements to prevent inadvertent access to the movement area for airports (i.e., commercial service airports certificated under Part 139). Many GA airports also implement security measures found at certificated airports, including perimeter fencing and inspections, which is consistent with guidance issued by TSA (2017). Airport perimeter fences prevent people and vehicles from entering the air operations area, and reduce the entrance of wildlife that may present a hazard to aircraft operations. The fence and access gates must be monitored and controlled as well as inspected on a regular basis to ensure integrity and compliance. During each regularly scheduled inspection, the fence, gates, locks and other safeguards are scrutinized to ensure they are functioning properly. In addition to looking for holes in fences and partially open or unlocked gates, the inspector can look for people and wildlife as well as evidence that inappropriate or unexpected activity has taken place. For example, trash, debris, damaged fencing or barbed wire could all suggest unauthorized access or attempted access. At most airports, perimeter inspections are conducted by operations personnel, although in some cases, security personnel may conduct these inspections. The proposed AGVT application for evaluation is: • Automated without safety driver Airport perimeter inspection is an important component of an airport’s regularly scheduled inspection and continuous surveillance inspections, and is a primary component of the public protection and wildlife hazard management components of the self-inspection program, which are overseen by FAA as part of certification under Part 139. Additional requirements of the perimeter inspection may be required by Part 1542 and reflected in the Airport Security Plan, which is approved by the TSA. Airport perimeter security includes protection of the fence (or other perimeter barrier), the gates for vehicles, pedestrians, maintenance and construction, roadways and GA areas (U.S. GAO, 2016b). Information related to perimeter security checks as a part of the airport self-inspection program is provided in the following advisory circular. Additional security related provisions of perimeter inspection are considered sensitive security information and are not discussed. • AC 150/5200-18C, Airport Safety Self-Inspection, 4/23/2004. Because of the significant size of the airport and perimeter, inspections are often conducted by vehicle. At some airports, there may be segments of perimeter that are difficult to access due to terrain or other obstacles; accessibility may also be affected by seasonal events such as snow or flooding. Automated perimeter surveillance has been successfully deployed, and includes UAS as well as AGVT. Heathrow, Stansted, Luton and Gatwick Airports in England, and Ben Guerion Airport in Israel, have all used UAS for perimeter security. Although AGVT reduce authorization issues and associated risks with respect to airspace, UAS are more advanced in terms of off-the-shelf availability and allow greater access for areas with rugged terrain.

Detailed Evaluation Results 97 In North America, the Indianapolis Motor Speedway (IMS) has two AVs for security and Edmonton International Airport in Canada is developing an automated security vehicle, as shown in Figure 27. Automated security vehicles at IMS perform security checks on predefined routes following the perimeter or in a defined area, or can serve as a stationary sentry with video feed to the IMS camera system. Each vehicle travels at walking speed, 3 to 4 miles per hour (Guskey, 2018) and serves as a deterrent as well as actively gathers information about its surroundings (Waitt, 2018). The vehicles can traverse all terrains at the speedway, can function in harsh environments such as snow, and can raise a boom with a camera and sensors to look inside vehicles or structures, identify potential problems that may exist, and serve in semi-autonomous mode for incident response (Guskey, 2018). An infrared sensor enhances operation and security at night. When something unexpected is encountered, the vehicle can send an alert to a security center. Audio feed, video feed, and two-way audio communications capabilities allow a responder from the operations or security center to talk with someone near the security vehicle; the resulting information can be used to support an appropriate response. In addition to providing enhanced security capabilities, the robot is consistent with the IMS philosophy for innovation (Guskey, 2018). Edmonton International Airport is developing an automated security vehicle to patrol the 12 mile (20 km) fence line searching for security breaches and unauthorized people. The modified ATV is under development by the Alberta Centre for Advanced Microprocessor and Nanotechnology Products (ACAMP, 2018), with partner firms from the oil and gas and defense industries. The ATV is equipped with five cameras, LiDAR, and speaker phones to provide information to both security and airport operations personnel to ensure security concerns are rapidly addressed and any fence repairs can be made in a timely manner (Sarkonak, 2018). The LiDAR and cameras provide object detection and fence inspection. The unit can operate autonomously or using remote control, but the driver’s seat is preserved so a person can ride in the vehicle or operate the vehicle manually. Using sensor data and software algorithms, the automated vehicle can also identify and document wildlife, including animal species such Figure 27. Perimeter security autonomous vehicle with no driver currently in development. Photo: Edmonton International Airport, 2018.

98 Advanced Ground Vehicle Technologies for Airside Operations as coyotes and deer (Sarkonak, 2018). The technology allows employees to focus on tasks other than airport perimeter patrol (no change in airport personnel is expected), streamline maintenance checks, and as a result, the technology may save hundreds of thousands of dollars in security costs (Sarkonak, 2018). The Ben Gurion Airport in Tel Aviv, Israel, also uses an automated security ATV to secure the perimeter fence and runways, with an initial report in 2010 (Estrin, 2010), and another report of a second vehicle in 2011 (Israel Defense, 2011). Although limited information is available about current capabilities, the 2011 vehicle is electric and can carry sensors, cameras, identification devices, and even a weapons system. The vehicle can traverse terrains from sand to mountains with automated operation or via remote control from up to 6 miles (10 km) away (Israel Defense, 2011). Operational times are up to 18 hours using two engines and a charging backup generator, with capabilities to work alone or as part of a pack (Israel Defense, 2011). The unmanned vehicle can identify hazardous materials, and can be linked to a control room as well as to a smart fence or other sensors that provide alerts of unauthorized entrance (Israel Defense, 2011). AV with no driver may provide a valuable tool to enhance airport perimeter security, which would address one component of the GAO’s recommendations for improved airport peri meter and access control security (AC 150/5200-30D, 2016). Deployment of AVs for perimeter security may be a valuable tool that can enhance airport safety and contribute to the philos ophy put forth by TSA for layers of security. Israel Homeland Security uses AGVT as one component to address the challenges of perimeter security. “The response is a proactive perimeter security array combining physical security layers with virtual, technological layers, with high levels of coordination and synchronization among them through a smart command and control mechanism” (Israel Homeland Security, 2019). Automated without a driver would include cameras and microphone for remote video and audio feed, LiDAR and radar for environment perception, and infrared for thermal imaging and enhanced detection of people and wildlife, especially useful at night. The AGVT vehicle would travel the perimeter fence or other defined path and conduct an automated enhanced visual inspection for insufficiencies (e.g., fence holes) or suspicious activities (e.g., presence of unauthorized people). If a potential problem or discrepancy is identified, an alert will be provided to the operations center and/or security for additional investigation and remote operation. When a person is observed, it would be possible for the AGVT to determine whether the person is an authorized badge holder based on the SIDA badge displayed, the security desig- nation of the area, confirmation via facial recognition, and a crosscheck from an onboard database of authorized personnel. This could be facilitated with an onboard RFID reader and RFID SIDA badges at airports where this technology is deployed. Implementation Scenario The proposed implementation of automated perimeter security without a driver would be a full deployment, which is consistent with the technologies currently used at IMS and Ben Gurion. Operational Area The operational area of automated perimeter security without a driver would be the perimeter of the airport, where terrain is compatible. The proposed deployment would not be in the movement area or runway safety area or interfere or cross paths with aircraft opera- tions, which may limit operations at some airports.

Detailed Evaluation Results 99 Airport Characteristics Autonomous perimeter security would be useful at airports that currently conduct multiple perimeter inspections a day and have perimeter terrain that is compatible with vehicle capa- bilities. Given the high capital cost, it is more appropriate for a commercial airport; however, GA airports with increased security risks (e.g., some joint use airports, airports with security conscious tenants, or airports with enhanced security requirements) may also be appropriate. Project Partners Potential project partners for automated perimeter security may include TSA, law enforce- ment, airport tenants with increased security requirements, and industry partners such as defense companies, and oil and gas companies, which are also leaders in technology development for security and perimeter protection. Key Considerations Key Considerations for perimeter inspection include the following: • Benefits. Potential benefits for automated perimeter security include increased airport safety and security, decreased operational costs and reduced labor requirements, and deploy- ment data for future AGVT. The proposed deployment may also reduce the risk for airport personnel since the AGVT vehicle can safely gain information about a potentially dangerous situation, reducing the risk to people and providing information that can help ensure an appropriate response. • Technical feasibility. The technical feasibility of autonomous AGVT perimeter security is a TRL 8 or 9, which reflects capabilities in other sectors (e.g., IMS and military applications) and deployment at Ben Gurion Airport, and but no documented use of an off-the-shelf unit for an airport environment. The Sharp INTELLOS A-UGV used at IMS was introduced in 2016 but has not been used in the airport environment. Limiting factors may include a lack of features tailored to airports (e.g., proven capability for wildlife management functions) and lack of a proven track record regarding equipment capabilities, reliability and malfunc- tion vulnerabilities. For perimeter security monitoring, automated without a safety driver requires the vehicle to follow the perimeter airport path (or other predefined path) to inspect the integrity of the fence and gates and to provide notification of any unauthorized person or animal. This involves automatically following a planned path, avoiding obstacles, identifying and localizing objects of interest and sending the information back to the airport operations and security offices, in compliance with geofencing rules consistent with predefined vertexes on the airport GIS map stored on the vehicle. The vehicle will utilize cameras, a microphone, radar, GPS and IMU, and is able to localize itself and identify obstacles and discrepancies in the environment based on the GIS map and associated database on the onboard computer. In terms of technology adaptability, the optical sensors (such as cameras) may be more susceptible to dust and dirty environments, which may degrade their imaging quality. The cameras will need to be able to adapt to a wide range of lighting conditions, and the vehicle will be equipped with infrared illumination bulbs for night vision. In terms of safety considerations, the automated vehicle will operate on a defined path but will also have redundant geofencing to constrain operation and allow movement only within a defined range bounded by a polygon defined by vertexes that are coincident with the path following the perimeter of the airport. The programming for the robot should have geofencing constraints as the highest priority to ensure the vehicle can only operate in

100 Advanced Ground Vehicle Technologies for Airside Operations approved and safe areas. Having both GPS and IMU onboard poses backup capabilities in case one of the motion sensors fails. If the vehicle is in critical failure, it should be able to come to a complete stop immediately, set a parking brake, mark itself as incapacitated and communicate this status as incapacitated, which will shut down its autonomous driving power. The vehicle will have the capability of being remotely controlled, so it can be safely retrieved in the event of a critical system failure. This justifies the need for LAN using 4G-LTE or 5G telecommunication standards, which can provide stable data transmission over rela- tively long ranges. The network security issue for a LAN is tolerable as long as it is physically isolated from an unsecured network (e.g., the public Internet) and the software leaks are properly patched. • Operational impacts. Automated perimeter security will alter the way airport operations personnel conduct inspections. Inspections will be automated, although operations personnel can oversee the inspection via camera feed, which can be provided to both airport operations and security. Initially, it may be desirable to have the automated inspections supplement conventional inspections, and replace conventional inspections as the technology is proven. For example, an airport that formerly conducted eight perimeter inspections a day may initially conduct seven AGVT perimeter inspections and one conventional inspection. It will be valuable to identify the impact of the automation not only on safety and security but also on wildlife management. Another operational impact is the management of alerts from the AGVT. The expectation is that the AGVT will handle routine inspections and provide an alert for unusual circum- stances; however, there may be a “learning curve” that requires software modifications for the airport environment. In the short term, labor hours previously devoted to manual inspec- tions may be used to respond to alerts, document AGVT capabilities and work with the vendor to assure that alerts are provided when appropriate. A final operational impact and a potential benefit is that the airport may choose to use the sentry feature as a secondary and random method of providing augmented security and observation at access points or in selected airport areas. The capability to provide a remote video and audio feed to the operations and security centers may prove useful in a variety of situations including an airport emergency. • Infrastructure impacts. Infrastructure impacts of automated perimeter security are expected to be minimal. The automated vehicle can follow the same path currently used for a conven- tional vehicle but if the vehicle is electric, a charging station would be needed. Given the small size of the vehicle, space may be available in an existing hangar or garage close to the planned inspection route. The deployment would require a GIS map and communications. The GIS map would include all features required for inspection including boundary information (e.g., fence characteristics, gate and access points). Communications can utilize 4G-LTE or 5G signal, which may require additional transmitters for remote areas of large airfields. The number of transmitters would vary depending on the size of the airport and the existing service coverage. Seven microcells covering radii of 1/2 to more than 1 mile (1 to 2 km) or 1 macrocell with service radii of up to 20 miles (35 km) (Zhang, 2012) would cover the entire airport in many cases. Long term, the software could be upgraded to include additional information to support wildlife management (e.g., species characteristics for automated identification). • Stakeholder acceptance and ease of adoption. Adoption of AGVT perimeter security would not be expected to pose any significant problems stemming from stakeholder acceptance or ease of adoption. The greatest challenge may be documentation of the system capabilities and safety to facilitate acceptance by FAA for compliance with Part 139 requirements and by TSA for compliance with Part 1542 requirements. Outside influences might have a mixed effect on stakeholder acceptance of perimeter security AGVT. Perimeter security AGVT would not be expected to pose a significant threat

Detailed Evaluation Results 101 in terms of job elimination, but it would change the nature of the work required, diminishing the need for physical inspections and replacing it with surveillance of video feeds as well as management and maintenance of the autonomous fleet (some of which would likely be contracted out). If these changes to duties result in pay decreases, there could be some resistance from labor. At many airports, there is a shortage of available and qualified labor, reducing expected resistance from labor. Perimeter security AGVT would be expected to provide advantages to the adopting airports. The capability to perform continuous perimeter inspection autonomously and alert personnel to potential threats or vulnerabilities would reduce the need for in-person inspections and the need for conventional vehicles to carry out these inspections, while providing increased security coverage. The deployment data can help inform airports regarding the implemen- tation of future autonomous systems that may be more complex (e.g., AGVT that require greater interaction with human workers). Additionally, there might be useful data gained regarding the type and quantity of wildlife incursions (Sarkonak, 2018). The proposed AGVT has a high level of compatibility with existing operations and requires minimal infrastructure changes (e.g., communications, a GIS map and a charging area in a garage or hangar for storage). Since this AGVT is fully automated, it requires little human interaction, and would be expected to have low levels of complexity, with the greatest complexity when interactive capabilities are used (e.g., when two-way communication is used). The technology can be trialed (e.g., it can be used to supplement required inspections initially) and can be integrated into an airport’s operations and security plans over time as its capabilities are confirmed; this should lead to higher levels of stakeholder acceptance. Its high TRL 8 or 9, along with known deployments in airport and other environments, suggest relatively low levels of uncertainty for use. In terms of individual considerations, some workers may be glad to be relieved from a monotonous duty (suggesting good allocation of function); other workers may resent any automation and prefer conventional, manual inspections. The degree of task-technology fit for perimeter security AGVT depends on its ability to effectively conduct its duties and interact in the airport environment. It is, by design, adequately suited for perimeter inspec- tion and surveillance, but it is not clear how it will respond to the wide range of possible situations and security vulnerabilities that it could encounter. An inability to adequately identify and appropriately react to airport situations would reduce the degree of task- technology fit. There is the chance that trust issues could arise due to excessive or inadequate alerts, hacks, malfunctions, or missed security threats. Perimeter security AGVT should be designed to be robust against cyberattacks, minimize the possibility of equipment malfunc- tion, and have a low enough threshold to detect and report any potential threat. Perimeter security AGVT are likely to benefit from a high level of perceived usefulness, as they allow workers to focus on tasks other than perimeter security. Additionally, they allow security personnel to view and/or respond to a potential threat remotely, reducing the risk of harm to personnel and allowing responders to prepare for unusual situations before arriving on the scene. Possible threats to perceived ease of use may be encountered when initiating operation or when using remote interactive capabilities. Since individual workers will have limited interactions with the perimeter security AGVT, there are no expected obstacles in terms of perceived ease of use. • Human factors. Human factors considerations for automated perimeter security include allocation of function, computer-human interaction, compatibility with existing or proposed procedures, appropriate allocation of function, teamwork and communication, compat- ibility with environment, and impact on situational awareness. AGVT for perimeter security would be expected to have appropriate allocation of function since it utilizes technology to perform a routine and repetitive task, freeing up personnel for other tasks. Although the AGVT is automated and would perform the inspection independently, there is a need for

102 Advanced Ground Vehicle Technologies for Airside Operations good CHI for interaction with people that may be encountered and for use by the remote operator (alerts, two-way communication capabilities, operation of the remote boom, etc.). There are no expected problems with respect to compatibility with existing procedures as perimeter inspection is fairly simple and does not require complex operational procedures that involve other employees, vehicles, or aircraft. AGVT for perimeter inspection may enhance communications and teamwork as well as situational awareness, since both airport operations and security personnel can have access to the live feed. In some cases, airports have even reported using technology as a lever to enhance coordination and communication, which is one possible ramification of the proposed AGVT, if it is implemented as a shared deployment. AGVT for perimeter inspection is compatible with the environment as it will remove the need for people to do physical inspections outside. Situational awareness would ideally increase with the proposed AGVT, since the vehicle can provide continuous inspections, and allows remote observation and interaction by a human, when necessary. Documentation of system capabilities would be necessary to assure that removing the physical presence and capabilities of a human (e.g., peripheral vision, sense of smell, etc.) does not reduce overall situational awareness, since it would be very difficult to program software to be attuned to the wide range of unusual circumstances that could possibly occur. Potential Challenges Generally, the use of any autonomous vehicle in an airport environment must consider and overcome several impediments unique to the environment and some that are not unique but require consideration and response. These include rough terrain, extreme weather, RF inter- ference, regulatory approval (FAA and TSA), and cybersecurity. Long-term challenges would also include integration with other security systems and vehicles. In terms of rough terrain, airport perimeter roads may include unpaved or otherwise rugged sections. Similarly, all cameras, sensors, audio equipment, onboard computers, and other equipment must also be rugged and protected for all weather use, including wind, rain, ice, snow, sleet, thunderstorms, and tornadoes. Aircraft Pushback and Aircraft Tug/Taxi to Runway Aircraft tugs are used for a variety of aircraft movements, including aircraft pushback from the terminal gate, aircraft tug (or tow or taxi) to the runway, and re-arranging aircraft, including repositioning aircraft and moving aircraft to or from remote stands or maintenance facilities. The proposed AGVT applications for evaluation of aircraft pushback include: • Remote from the ramp • Automated with safety driver The proposed AGVT applications for evaluation of aircraft tug/taxi to runway include: • Remote from cockpit • Automated with a safety driver There are a variety of options available for moving aircraft, including conventional towbar tractors and tugs, towbarless tugs, automated towbarless tugs (e.g., TaxiBot), and remote control electric equipment (e.g., Mototok). All four of these are shown in Figure 28. It is important that the tow equipment be matched to the aircraft in terms of size, power, and connecting equipment. Conventional tugs (e.g., Figure 28a) typically require a person in the cockpit to apply the aircraft brakes, if needed, one person in the tug, and if on an active ramp, adequate guide

Detailed Evaluation Results 103 personnel (typically two wing walkers to ensure wingtip clearance; in some cases, a tail walker may be appropriate). Because of the size of a conventional tug and its placement in front of the aircraft, it is difficult to maneuver the aircraft in tight turns or in confined spaces. The person in the aircraft cockpit and the tug driver both must complete training and have appro- priate certification. Ground personnel communicate with the pilot through manual operating signals, as well as through verbal communication via radio, which is possible when the ground crew connect their headphones to the external communications panel (aka com panel) on the nose of the aircraft, which must be disconnected when the aircraft movement is complete. Conventional tugs also typically require that the ground crew install a bypass pin to dis- connect the nosegear from the aircraft steering systems; the bypass pin needs to be removed after the after aircraft is moved. One exception is TaxiBot, which does not require a bypass pin; TaxiBot uses a patented system without a bypass pin that enables quick aircraft release in preparation for flight and to allow full automation. The first towbarless tug (e.g., Figure 28b) was used in France in the 1980s, and it allowed a single tug to be used with a wide range of aircraft since it picks up the aircraft wheel (using (a) (b) (c) (d) Figure 28. Aircraft pushback and aircraft tug and taxi: (a) conventional towbar tractor or tug, (b) towbarless tug, (c) TaxiBot provides semi-autonomous towing operated by the pilot from the cockpit, and (d) Monotok provides remote operation with a single person on the ramp. Photo: Monotok, 2016; TaxiBot, 2017.

104 Advanced Ground Vehicle Technologies for Airside Operations hydraulic power) and places it on the tug, eliminating the need for an adapter such as a towbar (Mototok, 2016). A towbarless tug provides greater equipment flexibility, higher speeds and increased control and maneuverability; however, it typically still requires a brake rider in the cockpit for emergency braking for large aircraft, a driver in the tug and guides or wing walkers to assure wingtip clearance in congested spaces. More recent towbarless tugs, such as TaxiBot (Figure 28c), provide semi-autonomous towing and allow the pilot to control the tug from the cockpit. The tug can automatically connect to the aircraft (the tug lifts and cups the nose gear of the aircraft) and the pilot can control the tug using standard cockpit controls; as the nose wheel turns from pilot tiller-manipulation, sensors on the tug turn the tug wheels. A safety driver is on board to operate the tug for pushback and to oversee the return to the terminal after separation from the aircraft at the departure runway. The operation of TaxiBot has been successfully demonstrated in trials with Lufthansa at Frankfurt Airport, and has been approved by FAA and EASA for 737 aircraft and the A320 family of aircraft (AviationPros, 2017b). Deployment of TaxiBot is currently planned for multiple airports in India, and successful tests have been completed (Times of India, 2019). TaxiBot may reduce the need for wing walkers since the TaxiBot driver can see the aircraft wing tips. The driver operates TaxiBot for aircraft pushback, and remains in the TaxiBot to operate the TaxiBot for the return to the terminal. During the pilot controlled taxi to the runway, all pilot input directly controls taxiing through TaxiBot and the tug driver does not have a role or any control of the tug or aircraft, by design. Another recent innovation in aircraft movement technology is remote control electric equipment for aircraft pushback. Remote control equipment for pushback must be sized for the aircraft, and a variety of vendors have developed products. Mototok has models for GA aircraft as well as narrow and wide body aircraft up to 430,000 pounds, and can accom- modate the 737 and A320 family of aircraft (it cannot accommodate larger aircraft such as the A380 and B747). Mototok has been deployed by British Airways at Heathrow, where it has reduced delays by 54% (due to less waiting for a conventional tug) as well as reduced emissions due to the electric power (Mototok, 2018). One electric tug can be used for multiple gates; at Heathrow, one electric tug provides pushbacks for three gates, which is consistent with the idea that if an aircraft is being pushed back, the aircraft on either side should remain stationary to reduce the risk of collisions. A remote tug system may allow operation by a single user who can posi- tion themselves (and move as necessary) to assure obstacle clearance, reducing the need for additional guides in uncongested areas. Remote operation may increase the maneuverability since the tug is small in size and is positioned immediately below the aircraft wheel. In aircraft hangars and in manufacturing plants, systems such as Mototok can utilize automatic controls guided by ground markings (this is also referred to as AGV or automated guided vehicle function- ality) (Eckert, F., personal communication, March 6, 2019). In the long term, it may be possible to utilize AGVT capabilities on the ramp using a painted retroreflective line, an embedded magnetic strip, leader sensors or reference GPS. In the near term, however, it is necessary to have a person operate the tug via remote control and provide pushback clearance to ensure safety. Fully automated operation in the ramp environment is currently impractical due to the variety and range of challenges. Challenges include operational complexities, the need for coordination with other ramp activities, and the associated potential for collisions with other aircraft, ground vehicles and ramp workers, and the need for faultless functionality in all weather conditions (rain, snow, and fog, which may affect both visibility and the ramp surface visibility and friction) (Eckert, F., personal communication, March 6, 2019).

Detailed Evaluation Results 105 The challenges of operation in the ramp environment must consider aircraft damage, ground vehicle damage, ramp worker safety, and the need for reliable operation not only for safety but also to ensure adherence to airline schedules, since delays can be costly. Advanced sensors are also used to reduce the likelihood of aircraft damage due to over- steering. The movement of an aircraft via tug requires caution because nose wheels have limited turning radius; conventional tugs may provide notification when oversteering has occurred but typically do not provide warnings prior to oversteering. TaxiBot and Mototok utilize advanced sensors and reverse steering to avoid oversteering and damage. For example, since the TaxiBot system does not use a steering pin, there is no steering of the aircraft gear; as a result, there is no chance of damage to the aircraft steering system. The steering input from the cockpit is transferred to the TaxiBot tug rather than the aircraft nose landing gear; the rotating turret in the TaxiBot tug also has a much greater steering range than the aircraft steering system. In the long term, the need for aircraft tugs will change since aircraft manufacturers are adding electric power for taxiing without the use of aircraft engines (AviationPros, 2017a; Safran, 2017; WheelTug, 2019). Safran has a motor for the main gear and WheelTug has a motor for the nose gear. However, this may not address the issue of visibility from the cockpit and may not immediately provide a solution for the large number of older aircraft that require tugs. Furthermore, progress for these systems has not been as fast as originally expected (e.g., Dubois, 2015). Currently some proposed technology can be forward fit to new aircraft and may be retrofit to existing aircraft for selected manufacturers and recent model years (e.g., 2016 or newer for Boeing and Airbus). It might take a significant amount of time for retrofit, however, and it may not be possible or practical for some older aircraft. Given the broad range of ages in the aircraft fleet, electric and automated tugs may be a more practical solution. For example, the average age of the American Airlines fleet is 10.7 years, and the average age of some segments within this fleet is much older. The Boeing 757 fleet is 19.3 years, the average age of the Boeing 767 fleet is 20.2 years, the average age of the Airbus A320 is 17.9 years and the average age of the MD-80/90 fleet is 20.5 years (Airfleets, 2019). Information to support aircraft tow or tug activities is provided in the following advisory circulars: • AC 00-34A, Aircraft Ground Handling and Servicing, 9/10/1974; • AC 00-65, Towbar and Towbarless Movement of Aircraft, 11/8/2010; and • AC 150/5210-20A, Ground Vehicle Operations to include Taxiing or Towing an Aircraft on Airports, 9/1/2015. Aircraft pushback with remote control allows ramp workers to operate the aircraft tug from the ramp. Automated aircraft pushback with a safety driver allows automated operation with a person serving as an observer and remote backup safety driver who can monitor and take control, if needed. These two applications would support future deployment of automated aircraft pushback with the safety driver located in a ramp tower, providing oversight for multiple gates or an entire terminal. This framework is currently proposed for AVs in the roadway sector. Aircraft tug/taxi to runway with remote control from the cockpit allows the pilot to operate the aircraft tug from the cockpit with sensors that obtain information from pilot input using the standard aircraft controls. This would be implemented for the taxi to the departure runway to reduce aircraft fuel burn and emissions.

106 Advanced Ground Vehicle Technologies for Airside Operations Automated aircraft tug/taxi to runway with a safety driver allows automated operation with a human serving as an observer and backup safety driver who can monitor and take control, if needed. This would be implemented for the taxi to departure runway to reduce aircraft fuel burn and emissions. Aircraft Pushback Description Aircraft pushback is standard practice for aircraft parked nose-in at a terminal since there is limited visibility from the cockpit and jet blast from reverse thrust may damage nearby buildings and equipment, pose a threat to people, and increase the likelihood of FOD damage. Remote control of electric tugs and automated tugs with a remote safety driver are proposed for evaluation. Implementation Scenario Remote control of electric tugs is proposed for a demonstration. Automated with a remote safety driver is proposed for investigation since the technology is still under development. Operational Area The operational area for evaluation is the ramp area, which is a non-movement area. The equipment could also be used in other areas (e.g., tenant areas); however, use in these other areas is not a consideration in this evaluation. Airport Characteristics Remote control for aircraft pushback may be used at either GA airports or commer- cial service airports. Remote control for aircraft pushback would also be appropriate for commercial service airports of all sizes, as long as the remote tug could be used with the variety of aircraft served by the airport. The use of an autonomous tug with a safety driver may be most appropriate at a medium-hub or large-hub airport in which the entire pushback occurs in the non-movement area. Larger airports with high levels of traffic might benefit from pushback AGVT (remote control or automated with safety driver) more than smaller airports with less traffic due to increased economies of scale in terms of aircraft pushback and because any benefits in terms of avoiding delays associated with prolonged pushback maneuvers may be more critical at larger airports. In the long term, remote aircraft pushback could be controlled for multiple gates from the ramp tower at large airports, in which case cameras and/or LiDAR may be mounted on the terminal to provide a bird’s eye view of equipment and personnel and to ensure coordination of activities at multiple gates. Full deployment of automated pushback with safety driver may be most appropriately implemented in conjunction with a new airport terminal, so operations can benefit from design and procedures that are tailored to the technology, supported by full simulation prior to final design and construction. Project Partners Airlines and ground service providers would be appropriate partners since these organi- zations own the aircraft, have contracts with the aircraft pilots, and are responsible for aircraft

Detailed Evaluation Results 107 pushback. Airlines or ground service providers that have responsibility for multiple gates and flights during the day may be appropriate partners. Coordination with FAA Flight Standards Service to ensure their acceptance of operations when commercial passengers are on board would also be appropriate. Key Considerations Key considerations for aircraft pushback include the following: • Benefits. Potential benefits for AGVT for aircraft pushback, including remote operation and automated tug with a safety driver, include reduced emissions, reduced damage and increased safety due to better visibility. Both pushback AGVT applications are expected to provide relative advantages in efficiency, reducing delays by making pushback operations more consistent and reducing turn times. Pushback may have an impact on delay that has not been fully recognized (Stergianos et al., 2015), and recent deployments of remote operation have reduced delay by more than 50% (Ibekwe and Siciliano, 2018). In the long term, integration of AGVT may reduce the need for ramp personnel, which would result in cost savings. Use of remote electric tugs would reduce emissions and fuel consumption. Automated tugs with a safety driver would provide useful data to support additional AGVT deployment. The inclusion of collision avoidance technologies in both pushback AGVT applications will lead to increases in safety (and associated increased savings due to reductions in crash/ injury-related expenses) through the circumvention of collisions. Remote control pushback could also lead to relative advantages in safety due to an increased ability for the operator to have a better vantage point than in traditional pushback operations, whether on the ramp or viewing the pushback from a ramp tower or control room. If automated pushback with a safety driver eliminates the need for wing walkers during pushback, this pushback AGVT application could lead to relative advantages due to increases in productivity, assuming these displaced wing walkers are assigned to other ramp duties. Though initial costs might be higher, the move to electrification from internal combustion for pushback could lead to reduced complexity in terms of vehicle maintenance, since electric vehicles require less maintenance and are economically competitive with internal combus- tion engines for extended use (e.g., Feng and Figliozzi, 2013). There could be some increases in operational complexity with the use of electrified pushback AGVT though, in terms of routing to charging stations and scheduling charge times. • Technical feasibility. The technical feasibility for remote control of a tug for aircraft push- back with the operator on the ramp is high since this technology has been deployed. This corresponds to a TRL 9. The TRL for an autonomous tug with a safety driver is 7 since similar technology has been demonstrated in a manufacturing setting but has not been deployed in the airside environment; the greatest challenge is the complexity of ramp activities, including a wide range of aircraft, GSE and personnel. The sensors involved have a relatively high level of readiness, since millimeter wave radar has been used for commercial passenger vehicles (TRL 9), and solid-state LiDAR, though not widely used, has been tested under outdoor conditions (TRL 7 to 8). IMUs have been widely used for aerospace industries (TRL 9), and SLAM algorithms have been tested on road environments, and would be suitable for opera- tions close to airport buildings (TRL 8) for the similarity of environments. In terms of technology adaptability, a vision-based depth-measuring sensor, such as solid-state LiDAR or radar, mounted at the rear of the tug that is able to see both landing gears and the wingtip would provide the tug with sufficient capability of collision avoidance at two sides. A motion-based sensor such as IMU is proposed for motion detection and

108 Advanced Ground Vehicle Technologies for Airside Operations planning, in addition to the SLAM from the vision-based sensor. Since such an operation could be close to the terminal building, GPS signals may not be satisfactory. Data from a number of sources is used to ensure the aircraft pushback is safe. The aircraft dimensions can be obtained by uploading the aircraft tail number with a query to a standard aircraft dimension database; this dimension data can be combined with aircraft position and orientation information based on the position of the landing gear. Reference data regarding the ramp, airport terminal and fixed ground facilities can be obtained from a map and depth- measuring vision sensors such as onboard radar. The location and geometry of the aircraft adjacent to the target aircraft can be obtained using information from the ADS-B transponder combined with aircraft dimension database. The pushback path planning is based on path- planning results given the shape of target aircraft and available free space, using available soft- ware in industry such as MoveIt! (PickNik Consulting, 2019), without direct measurement or 3D reconstruction; safety in terms of obstacle obstruction with GSE and ramp personnel is ensured through obstacle detection via sensors on the tug and robotic wheel walkers. The proposed technology is compatible with current ATC and ground operations protocol, since the technology can rely on voice recognition software and verbal interaction with the pilot, with responsibility for monitoring and safe operation provided by the safety driver. There are a variety of voice recognition application programming interfaces (APIs), such as Google Speech API (Internet connection required) (Google, 2019) and CMU Sphinx (off-line, optimized for mobile devices) (Walker et al., 2004). Voice recognition programs can translate verbal commands to text and text to speech, and can use these verbal commands as a catalyst for machine input to initiate tasks. There may be a need for slightly different commu- nications protocol with different airlines since procedures may vary slightly. Tugs are heavy vehicles (and even heavier with an aircraft), which means they can accu- mulate a dangerous amount of kinetic energy even when moving slowly. As a result, hardware speed limiters and emergency stops should be mandatory. One example of a hardware lock is a vehicle parking brake that will be set anytime the safety driver is not in the tug. • Operational impacts. In the short term, operational procedures would need to be revised to accommodate remote operation or automated operation with a safety driver. The integra- tion of automation and personnel would be minimal due to the presence of a person who would be responsible for monitoring activities and intervening, if needed. For automated with a safety driver, the current operational procedures require people to do a number of activities and manual checks in addition to the aircraft pushback. Activities may include placement and removal of the bypass pin, a check that the cargo door is secured, the pre-conditioned air is disconnected, chocks are removed, the cradle is secured, and the safety zone is clear of personnel, equipment and FOD. In some cases, people need to confirm that equipment has operated properly. For example, the towbarless tug may automatically capture the aircraft and raise the bucket off the ground, however, the assist agent needs to check to see that the strap is taut after the auto cutoff has been engaged, and if not, then it may be necessary for an agent to use the winch override switch. Until all tasks associated with aircraft pushback can be automated, it is necessary for people to be involved. AGVT may still provide great value by increasing safety, efficiency and reliability, and by removing people from unsafe or difficult tasks, but it is important to recognize that in the near term AGVT may augment rather than replace people due to the complexity of tasks, the wide variety of aircraft, and the current reliance on manual labor for ground services. Another operational consideration is the use of hand signals for communications, both between the ground crew and the flight deck and within the ground crew (e.g., between the tug driver and the guide agents). Signals are also used after the headsets have been discon- nected from the aircraft, and as a backup if communications are inoperable. Automation would

Detailed Evaluation Results 109 require all communication be verbal, or would require a way for the automated tug to communicate with the flight deck and the ground crew members using visible signals such as flashing lights. • Infrastructure impacts. Infrastructure impacts include charging stations for electric tugs, with one tug and charging station required for every three gates. For the automated vehicle with safety driver, a high-resolution map of the airport ramp and pushback area is needed, and this can be stored onboard the vehicle. In the future, the airport maps could be downloaded from a database stored in an airport server for autono- mous vehicles, in which case the map could be updated in real time based on information from tugs and other vehicles that are collecting 3D radar data and syncing to the database. Communications between the tug and the following robots that serve as automated wing walkers can utilize local area networking such as DSRC technology. If it is preferable to use static radar sensors rather than following robots, then it would be necessary to tempo- rarily mount the sensors on the ramp tower, the terminal or the jet bridge. These static radar installations would require radar, power, and possibly network setup, and would require calibration prior to use. If following robots are used, it may require distinctive paint marking on the ground so the robot can position and initialize itself for the collaborative observation and obstacle avoidance. In the future, a more robust use of automated tugs would utilize communications that enable V2X connectivity supporting communication between auto- mated tugs at different gates (e.g., on opposite sides of the apron, pushing back to a shared path to the taxiway). Communications may also be required to connect automated tugs with an airline ramp tower, an APOC, and a central server. Although an embedded magnetic strip or pavement markings with enhanced retroreflec- tivity could provide guidance for the pushback maneuver on a pre-planned path, this was not considered for evaluation since different pushback paths may be appropriate depending on the aircraft and the aircraft parked at nearby gates. • Stakeholder acceptance and ease of adoption. Stakeholder acceptance and ease of adoption is impacted by organizational acceptance (including regulatory issues and labor consider- ations) and individual acceptance. The use of remote control tugs will retain the aircraft pushback function with ramp personnel, which is consistent with current procedures; this would facilitate stakeholder acceptance and ease of adoption. Resistance and change management would be necessary since the remote operator will be able to walk on the ramp to ensure clearance from a variety of perspectives and will no longer ride the tug. Individuals may have different reactions to this change; some people may prefer sitting on the tug to standing and walking on the ramp. In the short term, the crew necessary for pushback during air carrier operations would likely remain the same because wing walkers will still be required in a congested ramp area. For aircraft pushback and repositioning during off peak hours, it may be possible to use a reduced crew. Long-term, increased use of remote and automated tugs would facilitate a smaller crew, increasing benefits and acceptance by management and likely increasing resistance from labor. Regulatory issues would include the need for FAA certification of the equipment and approval for operations. Initial deployment for cargo aircraft and/or to move aircraft without passengers (e.g., to support maintenance and aircraft repositioning) would reduce regulatory obstacles and overall risk, while providing a safe framework to explore the limits and capabilities of the system. Regulatory challenges would be reduced since all activities occur on the ramp, which is a non-movement area and outside the jurisdiction of ATC. Since both technologies retain a person to provide oversight and responsibility for aircraft pushback, safety would be increased and resistance reduced.

110 Advanced Ground Vehicle Technologies for Airside Operations Automated with a safety driver might have resistance from workers and unions, if it is seen as an intermediate to facilitate future automation that eliminates jobs; the same pos- sibility would likely garner support from airports and airlines due to the reduction in personnel costs. In the near term, displacement of ramp workers is unlikely since workers can be tasked with other ground crew duties, and the complexity of ramp activities requires a number of manual tasks and manual confirmation of automation; these manual activities are not easily replaced by automation. As an analogy, auto manufacturing utilizes robotics to a much greater extent than aircraft manufacturing, due to the difference in production levels and quality control requirements. The inclusion of automation that might lead to wage reductions due to reduced workload and responsibilities would likely meet union resis- tance (Teamsters Speak at Congressional Roundtable on Automation, 2018). Automation of pushback with a safety driver will have less resistance since it is an inves- tigation, and since full implementation is not imminent due to the complex environment. When technical advancements support full deployment, significant challenges may be expected because of reduction in labor (safety driver plus two wing walkers to one safety driver). The most appropriate application for full deployment of an automated pushback with safety driver in the future would likely be a new terminal construction project. This strategy is demonstrated by Changi Terminal 4, which implemented advanced technology in the terminal and on the ramp, which has its own ramp tower. In terms of compatibility, infrastructure investments needed for pushback (e.g., charging stations) may reduce ease of adoption. Remote pushback may be a good precursor to auto- mated pushback with safety driver as the data and operational changes may be useful to inform more advanced automation in ramp operations. This progression might help increase perceived trialability1 of pushback AGVT and reduce uncertainty, which would have a benefi- cial effect on organizational acceptance for AGVT to support aircraft pushback. Remote control pushback has the potential for high levels of perceived usefulness among its users. Whether implemented on the ground or in a control room, increases in area visibility and situational awareness afforded by moving the operator out of the pushback vehicle might be greatly appreciated and improve the individual worker’s level of trust. Moving operators from the loud and potentially dangerous ramp to control rooms could also have beneficial effects on workers’ perceived usefulness by means of improving the comfort and safety of their daily work, as ramp agents face high exposure to high volume noise that can be damaging to hearing (Tubbs, 2000). As occupational hearing loss can lead to a host of other issues including fatigue, depression, stress, and sleep disturbances (Hétu et al. 1995), this repositioning might in turn lead to higher well-being and retention rates of ramp agents as well. Individual-level acceptance of remote control pushback might also benefit from high perceived ease of use to the extent that the remote control design follows ergonomic prin- ciples and best practices, and proper training with how to operate the controller and maintain situational awareness has been received. Automated pushback with safety driver might have lower levels of perceived usefulness to experienced ramp agents who have no difficulty in performing conventional pushback operations, relative to newer ramp agents who lack experience. Like other L3 AGVT applications, low levels of perceived usefulness are likely if the individual does not feel the need for automation of a task. The effectiveness of the control algorithm and its accuracy, as well as the adequacy of the sensors to accurately perceive and navigate the ramp area, will determine the individual’s perceived ease of use and trust in automated pushback with safety driver. If the safety driver needs to intervene frequently, individual’s trust in this pushback AGVT application will suffer; if the method of intervention is awkward, individual’s perceived ease of use will suffer. 1 Trialability refers to how easily an application can be tried, which is important since airport stakeholders may need to experi- ence an innovative technology or practice in order to be convinced of its value and usefulness.

Detailed Evaluation Results 111 As both pushback AGVT applications do not eliminate the need for a human in the loop, and only stand to increase worker safety and well-being and decrease workload, the potential for misuse or abuse of these AGVT applications is low. The threat to employment is also low. • Human factors. Human factors considerations for remote tug technology and automated with a safety driver include whether the proposed AGVT complement the activities of people and is easy to use. Ideally, AGVT equipment must integrate ergonomic principles, support good computer-human interaction, be compatible with procedures, and represent appropriate allocation of function. AGVT equipment must also enhance situational awareness, support communication needs, facilitate teamwork, be compatible with workload requirements and reduce human error. For remote operation, training times to use existing remote equipment are minimal; this presumably reflects good computer-human interaction and good ergonomics. The greatest ergonomics consideration may be the increased time spent standing, rather than seated in the tug, which may cause some fatigue over the course of an eight-hour shift for some workers. For an automated tug with a safety driver, a significant concern is the ability of the safety driver to adequately monitor and quickly take over if needed. Many of the challenges asso- ciated with L3 operation in the roadway sector would not be expected in this environment, since the airside environment is dynamic and complex and each pushback operation takes a relatively short time (this contrasts with the roadway environment in which monitoring can become monotonous since the vehicle provides lane-keeping and speed control). The physical remote, whether it be a console inside a building or a handheld device, must be ergonomically compatible. Design features should be intuitive and lead the user to proper actions. The actual interface must be similar as it must be easy to use. Potential Challenges There are minimal challenges expected for remote operation of tugs for pushback; the greatest challenge may be regulatory if passengers are on board, and potential resistance from ramp personnel. Airlines may be hesitant to invest in this technology, since the method of aircraft pushback is rarely noticed by passengers. The greatest challenges for an automated tug with a safety driver are technical, since the application has not been deployed in the ramp environment, where the complexity of activities and the wide range of operating conditions with respect to weather may present challenges. Aircraft Tug/Taxi to Runway Description Taxiing aircraft propelled by jet engines are uneconomical, increasing fuel consumption, engine time and maintenance costs, as well as emissions. Negative impacts increase with long taxi distances and with delay, which can be significant at large airports where aircraft queue, waiting for their turn to takeoff, especially during peak hours. Using jet engines during taxi also increases the risk of FOD damage, since FOD can be drawn into the jet engines during taxi. Utilizing tugs to deliver aircraft to the runway (and ultimately retrieve aircraft from the runway) potentially presents a more economical, safe and environmentally friendly alternative for moving aircraft from the terminal to the takeoff runway. Implementation Scenario The proposed implementation scenario for remote operation includes operation from the cockpit by the pilot to move the aircraft from the terminal to the runway; this would be a

112 Advanced Ground Vehicle Technologies for Airside Operations demonstration project. This would require a person to return the tug to the terminal. The proposed implementation scenario for automated tug with safety driver would be an investigation since the required technology for automated braking has not yet been developed. Operational Area The operational area for remote from cockpit and automated with safety driver tugs would be the path between the terminal (after pushback) and the runway threshold, including the ramp and taxiways in the non-movement area and taxiways in the movement area. Operation in the non-movement area would require coordination with the ramp tower, and operation in the movement area would require coordination with ATC ground control. Remote and automated tugs could also support aircraft movement and repositioning for maintenance and fleet management. Airport Characteristics The use of remote from cockpit tug and automated tug with a safety driver would be most appropriate at select large and medium hub airports. Airports that have long taxi distances and significant delays may realize the greatest benefits, as would airports for which environmental considerations and sustainability are a priority. Remote from cockpit tugs could potentially reduce CO2 emissions from 7,040 pounds per typical taxi operation to less than 132 pounds (Alcock, 2016). Favorable airport geometrics and space and a path for tugs to return to the terminal is also an important consideration. Project Partners Potential project partners for both remote from cockpit and automated with safety driver include airlines, tug equipment vendors, and NASA, which has worked with airports and airlines in the deployment of technologies to reduce airport ground delay (e.g., ATD-2) and to advance automated aircraft taxiing. Key Considerations Key considerations for airport tug/taxi to runway include the following: • Benefits. The benefits of remote from cockpit include reduced fuel consumption, engine time and emissions, reduced risk of FOD damage, and increased gate availability. Trials of remote from cockpit operation in India suggest that the aircraft can delay engine start-up time by 10 to 12 minutes (Phadnis, 2018), saving fuel and reducing emissions and FOD risk. Gate avail- ability also increases as aircraft engine start up can occur during taxi to the runway. During conventional operations, aircraft engines must start up 2 to 3 minutes prior to pushback, and newer engines such as LEAP on A320NEO and 737MAX require longer start-up times. Shifting these start-up times so they occur en route during the taxi to the departure runway frees up the gates for arriving aircraft. The benefits of clearing the gate quickly are even greater when movements at one gate restrict movements for adjacent and nearby gates; this may be more likely at older airports with space and geometric constraints. One airline reported that almost half of their gates had procedural gate exceptions due to irregular geometry and other operating constraints. The benefits of automation with safety driver would include these benefits, as well as the potential for future benefits to increase the system efficiency, if all aircraft at the airport use automated tugs. Both remote from cockpit and automated with safety driver will also create valuable data that can be applied to the same application or other future AGVT plans.

Detailed Evaluation Results 113 In the future, implementation in conjunction with aircraft pushback can reduce bottlenecks near the terminal (TaxiBot, 2013). • Technical feasibility. The technical feasibility of remote from cockpit operations of a tug is between levels 8 and 9 as the technology has been demonstrated at several airports. The feasibility of tug automation with a safety driver is much lower, with an estimated TRL of 4 or 5. Current technologies that allow remote from cockpit operation may be capable of integrating some autonomous features in the future; however, there remain technical obstacles to fully autonomous tugs in the near future. Autonomous tugs are challenged by the need to perform adequately in a highly congested environment, ensure complete communication and control between the tug and aircraft, and successfully cross taxiways and runways. Operation of autonomous vehicles needs to be validated in movement areas before any implementation of autonomous tugs can begin. In terms of technology adaptability, airports usually have clear markings that autonomous tugs can follow to find the desired taxiway and target runway (FAA, AC 150/5340-IL, 2013). The communication between tower and aircraft also has standard terminology, which is relatively easy for speech recognition software to recognize. The tug would tune into the frequency between the tower and target aircraft, decode the designated taxiway, and use its onboard camera to find and follow the desired path. The tug would use a vision-based depth-measuring sensor, such as radar, mounted at the rear of the tug that is able to see both landing gears and the wingtip. Alternately, solid- state LiDAR could be used in place of radar, but this option was not evaluated. This would provide the tug with sufficient capability of collision avoidance at two sides. A motion-based sensor such as GPS provides motion detection and planning, in addition to the SLAM from the vision-based sensor. To ensure aircraft clearance, aircraft dimensions are obtained from a database, using the aircraft tail number, while aircraft position and orientation are inferred based on the position of the gear connected to the tug. The outline of the airport terminal and ground facilities can be obtained from the map and depth-measuring vision sensors such as radar on board. The geometry outline of the aircraft adjacent to the target aircraft can be obtained through ADS-B or airport dispatch records, which means the free space to move the target aircraft is obtainable without direct measurement or 3D reconstruction. The technology is compatible with current ATC and ground operations, since only the original tug driver has been swapped by an intelligent assistant on the microphone. The safety driver will also have communication capability with aircraft crew. Tugs are heavy vehicles, which means they can accumulate a dangerous amount of kinetic energy when moving even slowly. This is important to keep in mind during operation and hardware speed limiters and emergency stops should be mandatory. One example of a hard- ware lock is a vehicle parking brake that will be set anytime the safety driver is not in the tug. The greatest constraint in terms of technical feasibility may be the capability to tow aircraft fast enough and still be able to stop the convoy of aircraft and tugs fast enough to avoid damaging the aircraft landing gear. With remote control using TaxiBot, this is possible since the system uses the aircraft main landing gear to absorb the kinetic energy if the aircraft needs to stop. Technical feasibility for automated tugs is more challenging since an interface to remotely apply the aircraft brakes would be needed; there are no reports that such a system is being developed or investigated. • Operational impacts. Impacts to operations from remote tugs will include decreased opera- tional costs, reduced fuel burn, and lowered emissions. Full automation will also see a decrease in required staff. Overall airport capacity and throughput will be increased due to the efficient nature of aircraft towing. Actual procedural elements to the taxi operation will change with the addition of a tug/ aircraft combination. ATC, if required to oversee tug movements, will need to ensure they are

114 Advanced Ground Vehicle Technologies for Airside Operations directed away and around from regular aircraft operations or simply put them in normal traffic lines as if they were small aircraft. The tugs that deliver aircraft to the runway for takeoff cannot immediately provide service to landing aircraft since busy airports use different runways for landing and takeoff, and since aircraft taking off are delivered to the downwind end of the runway and landing aircraft need to be collected at the upwind end of the runway. Tugs for landing aircraft may have fewer benefits since aircraft queuing is more likely prior to departure than after landing. Both landing and takeoff are constrained by runway capacity; aircraft waiting for takeoff queue on the ground and aircraft waiting to land are placed in a holding pattern in the air. • Infrastructure impacts. Infrastructure requirements include new aircraft tugs, compatible airfield geometry for tug operation, and charging stations. In terms of compatible airfield geometry, tugs must have a safe route back to the terminal from the runway without nega- tively impacting aircraft operations. This may be possible through use of existing access roads for ground vehicles (aka service roads), as shown in Figure 29, or alternate taxiways. Ideally, tugs can deliver aircraft to the threshold and return to the terminal for the next aircraft without crossing a runway. The width of a standard ground vehicle access road is probably adequate; it may be necessary to confirm that the access road has adequate structural strength to handle the weight of the tug, which is heavier than many airport operations vehicles. Charging stations would be needed for electric tugs. Charging stations could be set up near the terminal, space permitting, since electricity may be more readily available and since tugs servicing any runway would come to the terminal area. In the future, a control room for control and oversight of tugs may be appropriate. In the future, pilots may control tugs remotely from the cockpit with automated return to the terminal after release. In this case, robust networking is required to allow human intervention during a malfunction. To set up networking devices covering the airfield, 4G-LTE or 5G signal trans mitters should be set up. Typically, 7 microcells covering radii of 1 to 2 km or 1 macrocell with service radii of up to 35 km (Zhang, 2012) will cover the whole airport area. • Stakeholder acceptance and ease of adoption. Remote from cockpit tugs will not require significant changes in some regards, since the pilot in command will retain operational control Figure 29. Infrastructure for tug to return to terminal. Image: Annotated Google map.

Detailed Evaluation Results 115 of the aircraft. It is necessary to ensure that the airport infrastructure will allow the tugs to return to the terminal using taxiways or access roads. With the use of automated tugs, procedures will need to change, although a person will remain in the tug as a safety driver for the return trip of the tug to the terminal, and the pilot will remain as the safety driver in the aircraft, since there is currently no mechanism for remote application of aircraft brakes. A change in procedures may cause there to be a varied reaction among pilots and ramp employees in terms of control. Ramp personnel may be upset that they only have to monitor the tug or tow situation rather than have control while pilots may be annoyed that they have extra controls to learn and a slightly higher workload to deal with. Certification of TaxiBot by FAA, EASA, Civil Aviation Authority of Israel, and Directorate General of Civil Aviation, Government of India, for narrow body aircraft reduces regulatory obstacles in terms of stakeholder acceptance and ease of use for remote from cockpit. Auto- mated tugs with a safety driver would require additional regulatory approval and may meet resistance from airlines and airports due to the challenges associated with an investigation of an immature technology; these challenges may be exacerbated since it is a temporary deployment, and perceived benefits (primarily progress of the technology) may not outweigh perceived costs (inconvenience and risk). Automated taxiing technologies would not reduce labor needs compared to the current situation since pilots currently taxi aircraft, and the role of pilots is not threatened; this implies that labor opposition would be limited. Because remote and automated tugs may utilize a safety driver, it would be valuable to frame the duties and contracts for this position in the near term to reflect a future scenario in which this position may not be necessary. The advantages of engine-free taxiing in terms of environmental benefits (reduced fuel and emissions) and the reduced risk of FOD damage (Aviation News, 2009) would support stakeholder acceptance, although the current low prices for fuel may limit the associated benefit in the near term. The requirement to change taxi operation procedures, although proven at airports such as Frankfort and in India, and the need for paths for tugs to return to the terminal may prove burdensome in terms of stakeholder acceptance and ease of use. Remote operation with pilot control benefits from a low level of complexity, since the technology leverages existing cockpit controls; pilot control is analogous to pilot taxiing with control via the aircraft nose gear. Implementation of automated tug with a safety driver has been reported to be ready for demonstration. This application, however, faces greater barriers in terms of ease of use and stakeholder acceptance since the technology has not been deployed and since it introduces a new framework for responsibility. The pilot in command is responsible for the aircraft; however the automated tug is in control. This may present liability challenges and resistance. Human takeover from the tug safety driver is necessary from a safety standpoint but may present challenges due to the pilot’s role as pilot in command. It may be necessary to temporarily transfer responsibility from the pilot to the automated tug, similar to the responsibility during pushback. The biggest concern with regards to acceptance and ease of use may be addressing the opera- tional solution with respect to how the additional tug traffic will be managed. Individual-level considerations for acceptance include acceptance by the pilot, and by the tug safety driver. Remote from cockpit allows operational oversight for the pilot, which bodes well, but pilots may be averse to any change and may not prioritize the benefits of fuel savings and reduced emissions; a belief that it reduces the likelihood of FOD damage, an important safety issue, would facilitate individual acceptance by pilots. The tug safety driver would likely be a ramp worker; this would represent an increase or change in duties. The increased responsibility for this task, including the need for certification to operate in the movement area, certification to operate a new piece of equipment, and the fact that this equipment moves an aircraft, which is a high value asset, may reduce stakeholder acceptance and ease of use.

116 Advanced Ground Vehicle Technologies for Airside Operations Automated with driver may meet more individual-level resistance than remote from cockpit since it is a higher level of automation, requires more changes to the operational procedures, and may imply a greater threat to jobs. Evidence has shown that humans are poor monitors of automation (e.g., Sheridan, 2002; Bainbridge, 1983), and the need to maintain situational awareness during automated taxiing might lead to low levels of perceived usefulness (i.e., if the automation must be monitored, what is the value of automation). Perceived ease of use for an automated tug might receive a lower rating than remote from cockpit taxiing if the means for the pilots to resume control are not well designed for timely takeover. Autonomous tug traffic may be seen as intrusive (by ground vehicle drivers, ATC, and pilots) which would reduce stakeholder acceptance and ease of adoption. • Human factors. Human factors considerations for remote tug technology include whether the proposed AGVT reflects ergonomic principles, good computer-human interaction, compatibility with procedures, allocation of function, situational awareness, communication, teamwork, operational suitability, and compatibility with workload and existing procedures, and also whether the proposed AGVT reduces human error. For remote from cockpit tugs, the human factors considerations are minimized since the pilot uses the standard equipment in the cockpit and the tug sensors translate the input from the nose gear to the tug. The greatest consideration may be the transitions, which could leverage the protocol used in Frankfort and India, as well as current protocol on the ramp. Another significant human factors consideration is that the tug driver would be operating in the movement area, which is outside the traditional domain of ramp workers and requires addi- tional training and protocol with ATC for the return trip. For automated tugs with a safety driver, the greatest concern may be that the addition of tug vehicles in the movement area introduces complexity, which would present challenges for all users, including ATC, ground vehicle operators and pilots. The increased complexity could affect workload and situational awareness. The addition of tug vehicles in the movement area increases the need for communication with ATC, and associated protocol for the tug to return to terminal after delivering the aircraft for takeoff. Potential Challenges The most imposing challenge for both remote and autonomous tugs is the addition of more vehicles to the movement area and the associated potential for congestion and additional complexity. Infrastructure and procedures will need to be developed to ensure that tugs do not compromise safety and efficiency or disrupt normal operations. Baggage Carts Description Baggage carts and the tug (aka tractor) that pulls them transfer bags to and from the aircraft as well as to and from the airport. Airlines operate their own equipment or contract the provision of ground services to a third party, with different arrangements at different airports. Although baggage carts are the responsibility of airlines and ground service providers, airports have an interest in their safe operation since they operate on airport property and use airport service roads. The tug and trailing baggage carts must avoid hazards including other vehicles, pedestrians, and aircraft. AGVT has the potential to increase airport safety and improve the efficiency of baggage cart operations as well as overall ramp operations. The proposed AGVT application for evaluation is: • Safety assist

Detailed Evaluation Results 117 Efficient movement of baggage is critical to efficient and safe aircraft turns. Information related to aircraft ground handling and servicing is provided in the following advisory circular: • AC 00-34A, Aircraft Ground Handling and Servicing, 7/29/1974, provides information for ground handling activities and personnel, although tug and baggage carts are not specifically addressed since the primary focus of the advisory circular is on equipment that directly services the aircraft. Implementation of AGVT for baggage carts is consistent with trends for AGVT to support GSE. IATA’s Airport Handling Manual requires anti-collision safety assist technologies for all new belt loaders, cargo loaders, passenger stairs and catering trucks by July 2018, and retrofit for belt loaders by July of 2020. Although not legally binding in the United States, IATA recommenda- tions reflect industry trends that will likely be adopted in the future, and reflect the need to pre- vent costly damage to aircraft and support employee safety. This evaluation of AGVT for baggage tugs and carts reflects an AGVT GSE application of greater interest to airports. Because tugs with baggage carts traverse airport service roads traveling to different ramps and terminals, they have a greater impact on overall airport safety and operations. Belt loaders and other GSE generally remain on the ramp, which is the responsibility of the airline or GSE provider, and their operation and movement have a lesser impact on overall airport safety. Progress has been made in terms of the availability of AGVT for GSE. The anti-collision technology in Textron’s SmartSense belt loader is one safety assist feature. In this case, input from ultrasonic sensors reduces the speed of the tug as it approaches the aircraft to eliminate aircraft damage caused by operator error (Textron GSE, 2019). The technology must be turned on when the loader is in the aircraft envelope; some operators do not turn the system on, to avoid the resulting reduced speed. Retrofitting the system so that the “on” light is visible from all sides can increase compliance and allow ramp supervisors to see whether the system is engaged when the loader is in the envelope. Other tug technologies in use include automatic brakes that assure the brake is on anytime the tug is not moving. This feature and other AGVT related to vehicle movement (e.g., variable speed limits) may be easier to implement as equipment transitions to electric power from the more traditional hydraulic transmission. Another example of AGVT is TractEasy, shown in Figure 30. This driverless electric tow tractor can traverse a predefined fixed path or respond to an on-demand request from a cell phone app to interface with software that provides mission management, route optimization and navigation. The equipment uses LiDAR, radar, stereo cameras, GPS, odometers, IMU, and V2X for path following and obstacle avoidance. This technology has an obstacle detec- tion range of up to 300 ft (100 m) in front of the vehicle and 120 ft (40 m) to the sides, has been used indoors to support warehouse and industrial applications, and is intended for use outdoors at airports. Figure 30. TractEasy driverless tug used to transport baggage at airports. Photo: EasyMile, 2018.

118 Advanced Ground Vehicle Technologies for Airside Operations Another technology that has been successfully implemented is Power Stow (2018). This tech- nology features a moving conveyer belt that can turn corners, and reduces the lifting required to load and unload baggage onto aircraft. Sensors keep the belt horizontal, even when the aircraft shifts as the baggage is unloaded. The equipment reduces injuries, baggage damage, and aircraft turn time, reducing costs and labor requirements, and has been successfully implemented by airlines and ground service providers. Successful deployment demonstrates that appropriate technology with good allocation of function will overcome initial employee resistance to new technology. Plans are also underway for the use of smart watches that provide real-time gate assignments and GPS tracking for ramp workers. As this technology and additional real-time data on ramp activities becomes available, the potential to improve operations and safety through AGVT and V2X capabilities increases. Safety assist features for baggage carts include collision avoidance, lane keeping and variable speed limits based on vehicle location and geofencing technologies supported by GPS transponder technologies. Implementation Scenario The proposed implementation of safety assist for baggage tugs and baggage carts would be a full deployment. Operational Area The operational area would include everywhere the baggage tug and carts travel, including the ramp, aircraft envelope, access roads between terminals, and in or near the terminal as required to collect and drop off passenger baggage. Airport Characteristics Safety assist for baggage carts would be most valuable at larger commercial airports where baggage carts traverse longer distances on airport access roads, including airports that require travel between terminals, and/or have hazardous areas such as narrow tunnels. Safety assist technologies combined with GPS geofencing technologies would ensure safe operation and allow baggage cart tugs to operate at higher speeds in remote areas and on road segments where visibility is good with few obstacles, and to operate at reduced speeds in congested and/or constrained areas, near tunnels and on tight radius curves. Project Partners Partners may include vehicle companies, GSE providers, and tech and software firms. Companies such as Ford may be good partners since AGVT for baggage movement may utilize baggage tractors built on a conventional pickup chassis, such as the Eagle Bobtail Tractor, which is built on a Ford chassis (Eagle Tugs, 2019). GSE providers may be good partners, including GSE companies such as Textron, manufacturer of the SmartSense belt loader, and Cavotec, a GSE provider that has implemented extensive technologies to support advanced technologies and automation at ports. Software companies, such as GroundStar, may also be good partners to support efficient deployment, tracking, and operation of GSE. Key Considerations Key considerations for baggage carts include the following: • Benefits. Benefits of baggage carts with safety assist functions include improved airport safety, fewer crashes, reduced damage, and increased efficiency, which will be realized by using

Detailed Evaluation Results 119 historic and real-time GPS data to improve operations. The proposed safety assist technology would improve airport safety by ensuring compliance with airport speed restrictions, reduce damage by using anti-collision technology, and support safe operation in narrow tunnels and other restricted areas. • Technical feasibility. The technical feasibility of safety assist technologies for baggage tugs and carts is high since safety assist technologies have been widely implemented in the roadway sector (TRL 9); however, confirmation is needed for the airside environment (TRL 8), including the algorithms for issues related to trailing carts, including path following capabilities and the potential for tipping (TRL 3). As an active safety assist feature, AEB brakes the vehicle at the very last moment, so the vehicle will achieve maximum deceleration and end up with a mini- mal designated clearance with the obstacle. The sensors involved have been used for com- mercial vehicles and have been tested under operational scenarios (TRL 8 or 9 for millimeter wave radar, TRL 8 or 9 for ultrasonic sonar). FAA has established standards for Airport GIS systems (associated TRL 8 or 9). Safety assist technology to prevent the chain of trailing carts from jackknifing or tipping will require specially developed control algorithms that reflect system dynamics, which will vary depending on the load and center of gravity for each cart (TRL 3).Utilizing force sensors at the hitch for automated trailer alignment and docking has been implemented in other sectors, but in this application would have an estimated TRL of 3. Safety assist technology includes collision warning and avoidance, lane keeping (on airport service roads), and GPS tracking in conjunction with speed limits for different airport areas. For collision avoidance, front- and rear-facing mid-range millimeter wave radar will pro- vide distance sensing to potential obstacles, while ultrasonic sensors (sonar) can be used as a short-range proximity sensor for docking maneuvers and parking assistance, when the vehicle is moving at relatively low speeds and there are blind spots. A rear-facing camera will eliminate potential blind spots behind the baggage cart, and act as visual assistance for the operator during alignment for cart connection as well. For geofencing purposes, GPS transponders in each vehicle will be used to provide vehicle location information. This information will be considered in conjunction with features and zones identified in a GIS airport map on the in-vehicle computer. In terms of technology adaptability, no major adaptation of commercial collision avoidance technology is foreseen. However, the mounting position and angle of the obstacle sensors may need to be recalculated and adjusted to ensure that their detection resolution can distin- guish aircraft obstacles, since conventional safety assist is not calibrated for aircraft. For baggage carts traveling on service roads to other terminals or other remote locations, a portable transponder integrated with a GPS-based software geofencing system to support variable speed limits for different areas can be mounted in the vehicle. In terms of technology safety and system redundancy, the safety assist technology is physi- cally local to an in-vehicle computer system (i.e., does not involve any networking and com- munication requirements) and thus is resilient to network attack. The invasion of malicious software or adversarial data to update media (such as a flash drive) can be prevented by using encrypted (electronically-signed) executable programs and detected by using a checksum of the upgrading program, protocol that is already widely adopted by the software industry. Because the system is limited to warning (passive) and only takes preventive control (active) of the vehicle, and the operator is able to override and disable the system at all times, no redundancy is required; however, the system should have self-diagnostic functionality and inform the operator of upcoming maintenance requirements. There are no expected infrastructure impacts, other than signal coverage for GPS. If electric baggage carts are used, then charging stations would be required. • Operational impacts. Operational impacts include improved safety on airport access roads and the potential for improved efficiency, since GPS data can be used to track and support efficient resource allocation.

120 Advanced Ground Vehicle Technologies for Airside Operations • Infrastructure impacts. Infrastructure impacts are expected to be minimal, and include good coverage for the GPS signal, an airport GIS map for the operating area, and good pavement markings (e.g., on access roads where LKA will be used). Pavement lane markings will provide confirmation of the information provided by an airport GIS map, in which the coordinates of the associated borders will support variable speed limits with software geofencing. A central computer and supporting software could be used to monitor the fleet of baggage tugs, and supervisor oversight should include mobile capabilities such as a cell phone app to ensure that supervisors have access to the required information wherever it is needed. • Stakeholder acceptance and ease of adoption. Stakeholder acceptance and ease of adop- tion for safety assist for baggage tugs and carts is facilitated by the fact that it will increase safety and will not affect labor force requirements or require approval from FAA. Safety assist may meet some resistance due to the inclusion of a GPS component and speed regulation. If employees are required to wear smart watches that include GPS capabilities, the capability to track equipment with GPS capabilities would become moot. As safety assist features become increasingly prevalent on passenger cars, operators may become increasingly reliant on these features, which will ease stakeholder acceptance and ease of adoption. Additional consider- ations related to organizational and individual factors are discussed below. In terms of organizational characteristics, acceptance and adoption will be influenced by the benefits, which will correlate with the operational characteristics due to the physical airport layout and the paths traversed by the baggage carts. Airports with longer distances between terminals, and airports with congested and sight restricted or other hazardous areas may particularly benefit from safety assist baggage tugs and carts. The increase in benefits and perceived benefits, and the associated reduction in operator workload, would increase stakeholder acceptance and ease of adoption. Acceptance and ease of adoption are also facilitated by the fact that the proposed AGVT is compatible with current operations, requires little to no infrastructure changes, and reflects relatively low complexity in terms of use and maintenance. Furthermore, this AGVT is also highly trialable, since safety assist equipped tugs and baggage carts can easily be phased in incrementally. Since collision avoidance systems are preventive devices, some benefits might not be immediately observable (Rogers, 2003), which does reduce the acceptance and adop- tion. This AGVT has a relatively low level of uncertainty from a technology perspective, and the greatest concern may be operator complacency and overreliance on safety assist features, which is likely best dealt with through training. In terms of individual considerations, worker familiarity with similar safety assist technolo- gies in their personal vehicles will support adoption and acceptance. Individual acceptance and willingness to adopt is also supported to the extent the AGVT improves worker safety and performance and without threatening to eliminate jobs. This will reduce misuse, abuse, and disuse (Parasuraman and Riley, 1997) although some employees may feel threatened by the GPS feature. Ideally, the system design should be such that an employee cannot turn off, disable or vandalize the GPS transponder. The proposed safety assist would be expected to benefit from a high level of task-technology compatibility and perceived ease of use, since it enhances task performance. User trust will be enhanced by system performance, and by organizational sharing of data regarding efficiency gains and accidents and damage avoided by the safety assist technology, which will also support perceived usefulness. • Human factors. Human factors considerations include whether the AGVT reflect appro- priate allocation of function, good computer-human interaction, procedural compatibility, communications, and compatibility with staffing needs. Since many safety assist features are well developed in the roadside environment, many of the human factors considerations have been adequately addressed. Since there will be minimal procedural changes, human factors concerns are reduced. The greatest human factors concern may be complacency, and specifi- cally the need to assure that operators maintain vigilance rather than rely on the safety assist

Detailed Evaluation Results 121 technologies. Other human factors concerns may be workload management issues for supervisors, who now have another tool, but also additional information that must be processed in an already complex and demanding environment. Operators must understand the functions and limitations of each safety assist feature, and receive training for appropriate use. The use of GPS and GIS will allow safety assist features to be tailored to the zone, which will support the provision of appropriate warnings and ensure that warnings do not distract the operator’s attention from a necessary task. Potential Challenges The greatest challenge is the need to demonstrate technical capabilities of safety assist in the airside environment; this is not considered a significant challenge given the deployment experi- ence in the roadway environment.

Next: Chapter 7 - Summary of Key Findings and Further Research »
Advanced Ground Vehicle Technologies for Airside Operations Get This Book
×
 Advanced Ground Vehicle Technologies for Airside Operations
Buy Paperback | $87.00
MyNAP members save 10% online.
Login or Register to save!
Download Free PDF

Recent advancements in automated and advanced driving technologies have demonstrated improvements in safety, ease and accessibility, and efficiency in road transportation. There has also been a reduction in costs in these technologies that can now be adapted into the airport environment.

The TRB Airport Cooperative Research Program's ACRP Research Report 219: Advanced Ground Vehicle Technologies for Airside Operations identifies potential advanced ground vehicle technologies (AGVT) for application on the airside.

Appendices B Through S are online only. Appendix A, on enabling technologies, is included within the report.

READ FREE ONLINE

  1. ×

    Welcome to OpenBook!

    You're looking at OpenBook, NAP.edu's online reading room since 1999. Based on feedback from you, our users, we've made some improvements that make it easier than ever to read thousands of publications on our website.

    Do you want to take a quick tour of the OpenBook's features?

    No Thanks Take a Tour »
  2. ×

    Show this book's table of contents, where you can jump to any chapter by name.

    « Back Next »
  3. ×

    ...or use these buttons to go back to the previous chapter or skip to the next one.

    « Back Next »
  4. ×

    Jump up to the previous page or down to the next one. Also, you can type in a page number and press Enter to go directly to that page in the book.

    « Back Next »
  5. ×

    To search the entire text of this book, type in your search term here and press Enter.

    « Back Next »
  6. ×

    Share a link to this book page on your preferred social network or via email.

    « Back Next »
  7. ×

    View our suggested citation for this chapter.

    « Back Next »
  8. ×

    Ready to take your reading offline? Click here to buy this book in print or download it as a free PDF, if available.

    « Back Next »
Stay Connected!