National Academies Press: OpenBook

Low-Speed Automated Vehicles (LSAVs) in Public Transportation (2021)

Chapter: Appendix D - LSAV Case Studies

« Previous: Appendix C - Interview Outline
Page 57
Suggested Citation:"Appendix D - LSAV Case Studies." National Academies of Sciences, Engineering, and Medicine. 2021. Low-Speed Automated Vehicles (LSAVs) in Public Transportation. Washington, DC: The National Academies Press. doi: 10.17226/26056.
×
Page 57
Page 58
Suggested Citation:"Appendix D - LSAV Case Studies." National Academies of Sciences, Engineering, and Medicine. 2021. Low-Speed Automated Vehicles (LSAVs) in Public Transportation. Washington, DC: The National Academies Press. doi: 10.17226/26056.
×
Page 58
Page 59
Suggested Citation:"Appendix D - LSAV Case Studies." National Academies of Sciences, Engineering, and Medicine. 2021. Low-Speed Automated Vehicles (LSAVs) in Public Transportation. Washington, DC: The National Academies Press. doi: 10.17226/26056.
×
Page 59
Page 60
Suggested Citation:"Appendix D - LSAV Case Studies." National Academies of Sciences, Engineering, and Medicine. 2021. Low-Speed Automated Vehicles (LSAVs) in Public Transportation. Washington, DC: The National Academies Press. doi: 10.17226/26056.
×
Page 60
Page 61
Suggested Citation:"Appendix D - LSAV Case Studies." National Academies of Sciences, Engineering, and Medicine. 2021. Low-Speed Automated Vehicles (LSAVs) in Public Transportation. Washington, DC: The National Academies Press. doi: 10.17226/26056.
×
Page 61
Page 62
Suggested Citation:"Appendix D - LSAV Case Studies." National Academies of Sciences, Engineering, and Medicine. 2021. Low-Speed Automated Vehicles (LSAVs) in Public Transportation. Washington, DC: The National Academies Press. doi: 10.17226/26056.
×
Page 62
Page 63
Suggested Citation:"Appendix D - LSAV Case Studies." National Academies of Sciences, Engineering, and Medicine. 2021. Low-Speed Automated Vehicles (LSAVs) in Public Transportation. Washington, DC: The National Academies Press. doi: 10.17226/26056.
×
Page 63
Page 64
Suggested Citation:"Appendix D - LSAV Case Studies." National Academies of Sciences, Engineering, and Medicine. 2021. Low-Speed Automated Vehicles (LSAVs) in Public Transportation. Washington, DC: The National Academies Press. doi: 10.17226/26056.
×
Page 64
Page 65
Suggested Citation:"Appendix D - LSAV Case Studies." National Academies of Sciences, Engineering, and Medicine. 2021. Low-Speed Automated Vehicles (LSAVs) in Public Transportation. Washington, DC: The National Academies Press. doi: 10.17226/26056.
×
Page 65
Page 66
Suggested Citation:"Appendix D - LSAV Case Studies." National Academies of Sciences, Engineering, and Medicine. 2021. Low-Speed Automated Vehicles (LSAVs) in Public Transportation. Washington, DC: The National Academies Press. doi: 10.17226/26056.
×
Page 66
Page 67
Suggested Citation:"Appendix D - LSAV Case Studies." National Academies of Sciences, Engineering, and Medicine. 2021. Low-Speed Automated Vehicles (LSAVs) in Public Transportation. Washington, DC: The National Academies Press. doi: 10.17226/26056.
×
Page 67
Page 68
Suggested Citation:"Appendix D - LSAV Case Studies." National Academies of Sciences, Engineering, and Medicine. 2021. Low-Speed Automated Vehicles (LSAVs) in Public Transportation. Washington, DC: The National Academies Press. doi: 10.17226/26056.
×
Page 68
Page 69
Suggested Citation:"Appendix D - LSAV Case Studies." National Academies of Sciences, Engineering, and Medicine. 2021. Low-Speed Automated Vehicles (LSAVs) in Public Transportation. Washington, DC: The National Academies Press. doi: 10.17226/26056.
×
Page 69
Page 70
Suggested Citation:"Appendix D - LSAV Case Studies." National Academies of Sciences, Engineering, and Medicine. 2021. Low-Speed Automated Vehicles (LSAVs) in Public Transportation. Washington, DC: The National Academies Press. doi: 10.17226/26056.
×
Page 70
Page 71
Suggested Citation:"Appendix D - LSAV Case Studies." National Academies of Sciences, Engineering, and Medicine. 2021. Low-Speed Automated Vehicles (LSAVs) in Public Transportation. Washington, DC: The National Academies Press. doi: 10.17226/26056.
×
Page 71
Page 72
Suggested Citation:"Appendix D - LSAV Case Studies." National Academies of Sciences, Engineering, and Medicine. 2021. Low-Speed Automated Vehicles (LSAVs) in Public Transportation. Washington, DC: The National Academies Press. doi: 10.17226/26056.
×
Page 72
Page 73
Suggested Citation:"Appendix D - LSAV Case Studies." National Academies of Sciences, Engineering, and Medicine. 2021. Low-Speed Automated Vehicles (LSAVs) in Public Transportation. Washington, DC: The National Academies Press. doi: 10.17226/26056.
×
Page 73
Page 74
Suggested Citation:"Appendix D - LSAV Case Studies." National Academies of Sciences, Engineering, and Medicine. 2021. Low-Speed Automated Vehicles (LSAVs) in Public Transportation. Washington, DC: The National Academies Press. doi: 10.17226/26056.
×
Page 74
Page 75
Suggested Citation:"Appendix D - LSAV Case Studies." National Academies of Sciences, Engineering, and Medicine. 2021. Low-Speed Automated Vehicles (LSAVs) in Public Transportation. Washington, DC: The National Academies Press. doi: 10.17226/26056.
×
Page 75
Page 76
Suggested Citation:"Appendix D - LSAV Case Studies." National Academies of Sciences, Engineering, and Medicine. 2021. Low-Speed Automated Vehicles (LSAVs) in Public Transportation. Washington, DC: The National Academies Press. doi: 10.17226/26056.
×
Page 76
Page 77
Suggested Citation:"Appendix D - LSAV Case Studies." National Academies of Sciences, Engineering, and Medicine. 2021. Low-Speed Automated Vehicles (LSAVs) in Public Transportation. Washington, DC: The National Academies Press. doi: 10.17226/26056.
×
Page 77
Page 78
Suggested Citation:"Appendix D - LSAV Case Studies." National Academies of Sciences, Engineering, and Medicine. 2021. Low-Speed Automated Vehicles (LSAVs) in Public Transportation. Washington, DC: The National Academies Press. doi: 10.17226/26056.
×
Page 78
Page 79
Suggested Citation:"Appendix D - LSAV Case Studies." National Academies of Sciences, Engineering, and Medicine. 2021. Low-Speed Automated Vehicles (LSAVs) in Public Transportation. Washington, DC: The National Academies Press. doi: 10.17226/26056.
×
Page 79
Page 80
Suggested Citation:"Appendix D - LSAV Case Studies." National Academies of Sciences, Engineering, and Medicine. 2021. Low-Speed Automated Vehicles (LSAVs) in Public Transportation. Washington, DC: The National Academies Press. doi: 10.17226/26056.
×
Page 80
Page 81
Suggested Citation:"Appendix D - LSAV Case Studies." National Academies of Sciences, Engineering, and Medicine. 2021. Low-Speed Automated Vehicles (LSAVs) in Public Transportation. Washington, DC: The National Academies Press. doi: 10.17226/26056.
×
Page 81
Page 82
Suggested Citation:"Appendix D - LSAV Case Studies." National Academies of Sciences, Engineering, and Medicine. 2021. Low-Speed Automated Vehicles (LSAVs) in Public Transportation. Washington, DC: The National Academies Press. doi: 10.17226/26056.
×
Page 82

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.

57 Appendix D presents three case studies of completed LSAV pilots. These LSAV services operated in a different operational design domain at speeds of under 15 MPH. The three cases studied are the following: • City of Arlington, TX, Milo LSAV shuttle, that was available to the public and used a multiuse pedestrian and bicycle path through a city park. • Fort Bragg, NC (Applied Robotics for Installations and Base Operations), where the LSAV traveled on roadways with mixed traffic, pedestrian pathways, and parking lots on a closed military base and provided service to/from medical appointments. • Las Vegas, NV, AAA Self-Driving Shuttle, which was available to the public and traveled in mixed traffic on downtown streets, crossing eight intersections. Cones marked a curbside lane or path. The case studies include interviews with the key stakeholders responsible for setting up these services. Each case study covers the transportation context, the planning process, the funding and procurement process, and what the project leads learned during operations. More specifically, • The City of Arlington, TX, case study was informed by more than 40 conversations with city staff, the North Central Texas Council of Governments (NCTCOG), First Transit, planners at Dallas Area Rapid Transit (DART) and the City of Frisco, Drive.ai, Via, May Mobility, Houston-Galveston Area Council (H-GAC), City of Arlington Tech Team Challenge members, University of Texas Arlington researchers, and others. • The Fort Bragg, NC, case study is primarily based on discussions with the Robotics Research team (the lead robotics experts who converted and operated the vehicle), the project lead from the United States Army Tank Automotive Research, Development, and Engineering Center (TARDEC), and the project manager who oversaw technology development and planning of the deployment, who is also technology deployment advisor to the research team. • The Las Vegas, NV, case study was based on interviews with six people involved in the planning, operation, and assessment of the LSAV project. These six interviews were augmented by discussions with the sponsor of the project, the technology deployment specialist (who is a member of the research team), the operational point of contact at the Regional Transportation Commission of Southern Nevada (RTC), a review of police reports and National Transporta- tion Safety Board reports, and an unpublished paper with regard to law enforcement and public safety considerations of the pilot. A P P E N D I X D LSAV Case Studies

58 Low-Speed Automated Vehicles (LSAVs) in Public Transportation D.1 City of Arlington, TX, Milo Case Study1 Project Profile From August 2017 to August 2018, the City of Arlington, TX, piloted a low-speed shared automated vehicle (LSAV) shuttle service, known as “Milo.” (Figure D.1.1). This was the first low-speed automated shuttle providing regular service to the public in the United States. The City’s twin goals were 1. Better understand the technology and 2. Expose residents and visitors to LSAV technology. The route is depicted in Figure D.1.2. The City of Arlington’s Office of Strategic Initiatives led the planning, project development, procurement, implementation, operation, and evaluation for the pilot. First Transit, a transit and mobility service provider, oversaw regular operation of the vehicle, which included training and supervision of on-board attendants. Two EasyMile EZ10 Gen-12 vehicles were selected for the pilot. The vehicles are capable of SAE Level 4 (High Driving Automation) travel and were operated with an on-board attendant and monitored from a central operations center. EasyMile also provided technical support and training. From August 2017 to August 2018, the shuttle operated for more than 100 events, including 78 events at AT&T Stadium and Globe Life Park (football, baseball, and concerts), 17 public demonstrations for interested citizens, and 18 demonstrations for special-interest groups (e.g., schools, local engineering groups, interested companies). The Office of Strategic Initiatives reported that the project’s goals had been met without significant safety incidents. See the Milo project overview in Table D.1.1. Geographic Context With a population of 396,394, Arlington, TX, forms the geographical center of the Dallas-Fort Worth metroplex. The climate in Arlington is a subtropical humid climate with summer tempera- tures of around 95°F (35°C). The heat combined with high humidity makes walking outside uncomfortable for many and difficult for some with limited mobility. Figure D.1.1. Milo LSAV shuttle service. 1 All photos and graphics in this case study reprinted with the permission of the City of Arlington. The City of Arlington is not an author or sponsor of this case study. 2 The EZ10 was deployed in Arlington. Since 2017, EasyMile has replaced the EZ10 Gen-1 with second- and third-generation systems. All observations reflect this older technology.

LSAV Case Studies 59 Figure D.1.2. Milo route map. Topic Milo LSAV Shuttle Overview Use Case Shuttle (off-street) connecting remote parking during special events Route Description 3 Fixed Routes, with 5 distinct stops (1 shared stop at south end) Route Lengths (1) South: ¾ mi; (2) North: ½ mi; (3) Public demo: ½ mi Vehicle Capacity 15 passengers, seated and standing (based on specs) Speed 15 mph maximum speed, 8 mph average operating speed Operator First Transit Vehicle Provider EasyMile Vehicle Type EZ10 Gen-1 Project Manager City of Arlington, TX Next Phase Drive.ai on-demand shuttle in Entertainment District Table D.1.1. Milo LSAV shuttle overview. As of 2015, over 14 million visitors come to Arlington each year, most of them going to the Entertainment District. The Entertainment District is an area with over 15,000 jobs along with the Dallas Cowboys Stadium (AT&T Park), Texas Rangers Stadium (Globe Life Field), a Six Flags amusement park, a major convention center, hotels, and other regional attractions. The University of Texas (UT)–Arlington is in the nearby College Park District. Most of these attractions, along with several office buildings, are located together in the northern part of the city. Access to these destinations is almost exclusively by car. The Entertainment District has more

60 Low-Speed Automated Vehicles (LSAVs) in Public Transportation than 30,000 parking spaces in both surface and structured parking. For large events, some of the remote parking lots are located over one-half mi away from travelers’ destinations. Milo shuttle operated in two routes alongside Johnson Creek and Mark Holtz Lake between AT&T Park, Globe Life Field, and the Arlington Convention Center. The route was primarily north to south in the Richard Greene Linear Park and crossed under East Randol Mill Road with fixed stops along the route to connect the destinations with their respective parking lots. The Milo operations were on an off-street on a geofenced multiuse path without other motorized vehicles. The path was approximately 10 ft wide and used by both bicyclists and pedestrians. The typical operating speed was 8 mph because the route was on a multiuse pathway. This is below normal operating speeds of the EZ10. Transportation Context Arlington is the largest city in the United States without regular scheduled fixed-route transit service. While Dallas Area Rapid Transit (DART) is expanding an extensive light-rail system nearby, Arlington is not a part of that network. The Trinity Railway Express (TRE) also runs adjacent to the city to the north. In 2013, the City of Arlington contracted with DART to operate the Metro Arlington Express (MAX) commuter bus service between UT-Arlington and the TRE DFW/Centreport station. In 2017, the City of Arlington terminated that service although it provided the only shared service on that corridor. Shortly after the termination of the MAX, the City of Arlington initiated the Milo project in the Entertainment District and began a separate on-demand rideshare service provided by the private company Via. Via also provides an on-demand platform for transportation in Arlington, which will facilitate the dispatch of paratransit service in the future. The City of Arlington planned to transfer operations of this paratransit service, known as Handitran, to Via in late 2019. Arlington’s LSAV shuttles and on-demand services are both designed to provide connec- tions within and to key activity centers in the city. Phase 1 Milo service provided connection to event venues. The next phase was with Drive.ai (as described under Planning for Subsequent Initiatives); that service provided connections to hotels, the convention center, offices, and the stadiums. The City of Arlington planning team has also assessed routes and use cases on the UT-Arlington campus and adjacent residential areas, as well as the route now served by Via. City leaders are discussing how Via could use its platform and/or its operational service to provide on-demand LSAV service. Planning The Milo pilot was the first phase of the City of Arlington’s LSAV pilot, and the Office of Strategic Initiatives led the planning, project development, procurement, implementation, operation, and evaluation over an 18-month period. The planning of the Milo pilot took place as the City of Arlington wanted to promote mobility innovations to meet key transportation needs. At the same time, the regional metropolitan planning organization, the North Central Texas Council of Governments (NCTCOG), pioneered efforts in regional AV/LSAV planning. Three things occurred during this time: In the fall of 2016, the City of Arlington appointed a Transportation Advisory Committee (TAC) to develop a city mobility innovation plan. The TAC developed its recommendations over a multi-month time period and published its final report in September 2017. AV shuttles were a key recommendation in this plan. Also, in 2016, the NCTCOG designated a lead for automated vehicle planning (which encom- passes LSAV) and began coordinating with the Texas Innovation Alliance on its Automated

LSAV Case Studies 61 Vehicle Proving Ground efforts. NCTCOG sponsored related pilots, incorporated a technology chapter in its 2045 Regional Long-Range Plan, included automated vehicle pilots in its TIP, and constructed a first-of-its-kind automated vehicle strategy and deployment initiative due to be launched in 2020. As the TAC continued its stakeholder and community meetings in 2017, the City of Arlington moved forward with an LSAV pilot. In January of that year, it decided to pursue an automated vehicle pilot to become more familiar with the technology. In February, the City of Arlington identified the EasyMile technology, and in August it launched a pilot service in the Enter- tainment District. In a similar time frame, the City of Arlington also looked for an on-demand solution to replace a 6-mi route connecting the Trinity Railway Express to the business core of Arlington. Mobility Innovation Plan Development The Arlington City Council established a stakeholder/citizen committee, the TAC, to craft a mobility innovation plan to strengthen Arlington’s transportation network, to provide trans- portation amenities for its residents and visitors, and to promote innovation. The TAC examined currently available and emerging transportation technology, gathered stakeholder input, and recommended which modes would work for Arlington’s current and future needs. The September 2017 strategic report, “Connect Arlington: A Transportation Vision Connecting People and Places,” provided a strategic plan for new transportation investments and explained how on-demand platforms and automated shuttles could enhance the city’s mobility system. Connect Arlington set priorities for both new fixed-route and on-demand services, including identifying how automated vehicles could be best implemented to serve local transportation needs. The report calls for • Bus rapid transit/high-intensity bus (with a dedicated lane) • Demand-response rideshare (e.g., Microtransit services like Via) • Personal rapid transit • Rubber-tired automated shuttle service, either at grade or on an elevated guideway The report identified six key corridors for public transportation improvements, which could incorporate automated vehicles, on-demand service, and/or bus rapid transit. The six corridors are listed here and shown in Figure D.1.3: Corridor 1: Centreport TRE Station to Entertainment District Corridor 2: Entertainment District to South Arlington, along Cooper Street Corridor 3: Entertainment District to Tarrant County College, along Route 360 Corridor 4: Interstate 30 Corridor Corridor 5: Interstate 20 Corridor Corridor 6: Pioneer Parkway/Highway 303 In addition, the Connect Arlington document envisions services that connect to the high- speed corridors (corridors 4, 5, and 6) such as the commuter rail line, the TRE, and corridors along freeways. In addition to these transit growth strategies, the City of Arlington has pursued a variety of options to provide shared service to, from, and within the city. As previously mentioned, a commuter bus service called MAX was contracted from DART and connected to the TRE Centreport Station. However, during its four years of operation, the MAX had low ridership, averaging just 240 one-way rides per day.3 As previously indicated, the City of Arlington canceled its contract and replaced it with on-demand service provided by Via in 2017. The City 3 Cadwallader, Robert, “Time Running Out for Arlington’s Three-Bus Public Transit Service.” Fort Worth Star-Telegram, July 1, 2016, http://www.star-telegram.com/news/local/community/arlington/article87143757.html.

62 Low-Speed Automated Vehicles (LSAVs) in Public Transportation of Arlington envisions that automated shuttles will represent a larger part of the core transporta- tion network, as local connectors and as part of the Via demand-response microtransit platform. Regional AV Planning The Milo project was not included in the region’s long-range plans at the time of its incep- tion. NCTCOG’s leadership supported the effort through its two-person AV planning unit and complementary programming through the Texas Innovation Alliance. Additional work is being coordinated with the Texas Automated Vehicle Proving Grounds, organized by the Texas A&M Transportation Institute (TTI), the University of Texas–Austin, and the Southwest Research Institute. After the start of Milo, NCTCOG made AV pilot projects eligible for Conges- tion Mitigation and Air Quality (CMAQ) funds in its TIP. The City of Arlington tapped this CMAQ funding for Phase 2 of its AV pilots. In May 2018, NCTCOG adopted the 41-city area’s 2045 regional transportation plan, which identifies goals that align with the Milo pilot and estab- lishes a more comprehensive framework for AV planning and implementation in the region.4 Project Planning for Milo In early 2017, the Arlington City Council instructed city staff to identify a route, preferred technology, and operational requirements for a shuttle pilot. The regulatory environment in Texas has been encouraging of automated vehicle deployments, and City of Arlington staff did not need to receive permission from the State of Texas before the project’s start. The project also did not require an FMVSS waiver because the vehicle did not operate on public roads. Rather, the Milo operated on off-street pathways on city-owned land. City staff launched the service in August 2017, after 7 months of site assessment, modest infra- structure enhancements, and development of operational and safety protocols (in conjunction with EasyMile and First Transit). During those 7 months, city staff also worked on procurement, contracting, safety attendant training, and coordinating with law enforcement. Figure D.1.3. Key new mobility corridors. 4 North Central Texas Council of Governments, Mobility 2045. http://www.nctcog.org/trans/plan/mtp/2045.

LSAV Case Studies 63 Planning for Subsequent Initiatives In October 2018, the City of Arlington began Phase 2 of its AV effort with on-street services provided by Drive.ai. This time, it chose its vendor through an RFP for turnkey operations and tapped CMAQ funds allocated by NCTCOG. This was Drive.ai’s second deployment in North Texas; the first was a short self-funded pilot within a business park in nearby Frisco. Drive.ai’s service ended on May 31, 2019, as a result of Apple purchasing the company. A total of 765 automated trips was provided during the length of the pilot. Separately, NCTCOG approved a three-part initiative providing a total of $30 million for automated vehicle deployments (including LSAVs). These funds would support its member cities in managing deployments, planning initiatives, and developing strategies for AV readiness, including management of revenue, parking, and curbside impacts. Procurement and Preparation For the Phase 1 Milo shuttle, the City of Arlington required that the LSAV shuttle have a wheel- chair ramp. Staff determined that EasyMile had the only commercially available LSAV at the time that included an automated ramp, justifying a sole-source contract. The City of Arlington then used general funds to contract with EasyMile to lease two EasyMile EZ10 Gen-1 vehicles for a total of $265,213. The cost of the shuttle’s operations was funded with the Convention and Event Services account using tourism-derived revenues. This funding source reflected the City of Arlington’s goal to become a technology leader in the Dallas-Fort Worth region, which has demonstrated widespread community interest in automated vehicles. With City Council approval, the Office of Strategic Initiatives contracted with EasyMile for the vehicle and software as a 6-month lease, with an option for a 6-month extension. The City of Arlington had initially planned to operate the vehicles themselves upon receiving them from EasyMile and getting the relevant training. However, the City of Arlington staff realized that this would be an unsustainable model and decided to contract with First Transit to handle regular operations at a stipulated number of events and to provide daily service. Arlington’s Convention and Visitors Bureau issued the operations contract. Insurance The City of Arlington self-insured the project with backing from its government insurance risk pool. In addition, it required additional coverage by EasyMile. Within the EasyMile contract, the City of Arlington stipulated that the vendor secure appropriate general liability insurance to cover passenger service operations, and EasyMile indicated that this could be covered by its general umbrella policy. Later, the city’s risk management staff determined that this level of coverage was inadequate and requested additional insurance from the vendor. This was readily resolved by EasyMile and largely resulted from differences in insurance require- ments in the United States versus those required in France. Site Assessment Before the vehicles arrived on site, one of EasyMile’s engineers identified and documented potential risks and mitigation strategies along the proposed route. This information aided in the development by the EasyMile engineer of a Site Assessment Report, which summarized EasyMile’s requirements and recommendations as well as the scope and conditions of the operations on this specific site. EasyMile completed the assessment in May 2017 as a predicate to completing operational planning and launching the service in August.5 In other projects, these 5 The process for EasyMile’s Site Assessment was described in interviews with project stakeholders.

64 Low-Speed Automated Vehicles (LSAVs) in Public Transportation steps typically occur before procurement and are an aid to defining technology requirements. In this instance, the site assessment was conducted by EasyMile and can be seen in Table D.1.2; it focused on how to adjust the proposed route (as seen in Figure D.1.4) to meet the vehicle’s requirements, namely • Impediments to vehicle localization; and/or • Issues that could cause the vehicle to unnecessarily perform emergency stops, like the clear- ance of surrounding plants, bridge structures, open fields, and the steepness of the grade. Infrastructure Requirements In its Site Assessment Report to the City of Arlington, EasyMile recommended a series of changes to the physical infrastructure. EasyMile did determine that the digital infrastructure was sufficient to deploy the service as the vehicle had been designed so it could operate on existing 4G networks and GPS service. Further, the City did not include testing of dedicated short-range communications (DSRC) and related RSUs as part of this pilot. Note that this route did not include traffic signals. Infrastructure modifications are summarized in Table D.1.2. Infrastructure Changes Required Reason Concrete crossing turnouts The existing multiuse pathway is not wide enough for two vehicles to pass. Rocks along lake edge This gave the vehicle additional reference points to minimize risk associated with localization deviation. This path followed a small lake that had insufficient reference points for the EasyMile vehicle. Birdhouses along path These provided fixed reference points for the vehicle to identify its location when clear edges are not available. Tree/grass trimming This kept the path clear for the EZ10 shuttle and ensured that grass clippings/branches would not be interpreted as obstacles that slowed the operation of the vehicle. Signage and path marking This is for users of the path and shuttle riders, to help them go to the right location and to share the path with the EZ10. Table D.1.2. Required infrastructure changes (excerpt EasyMile assessment). Figure D.1.4. Milo route map.

LSAV Case Studies 65 Operations As stated, First Transit was contracted to provide regular operations support for the shuttle. As the operator, First Transit developed operating and safety protocols including6 • Developing roles and responsibilities for an operations team. • Creating a daily operational schedule. • Developing a service and maintenance plan (based on guidance from EasyMile). • Setting emergency guidelines and protocol (based on guidance from EasyMile). • Developing and conducting operator training and certification (based on guidance from EasyMile). • Operating the service (based on guidance from EasyMile). • Reporting key performance indicators. For daily operations, First Transit dispatched the two vehicles, operated them during the stipulated hours during events at the Entertainment District, and performed regular preventative maintenance. Consistent with EasyMile guidance, First Transit performed a graffiti/vandalism check, cleaned inside surfaces with Lysol, and sprayed down the vehicle on a regular basis. First Transit also hired and trained safety attendants to monitor operations in the vehicle during regular service. First Transit and City of Arlington operators manually logged passenger counts, the number of soft and emergency stops, and errors/disengagements in a Google Form that could be accessed by EasyMile, the City of Arlington, and First Transit. Each of the project partners could log in and access the data when necessary. Separately, the vehicles had a black box that recorded soft- ware error information, which could be accessed by EasyMile. The City of Arlington did not receive this LSAV performance information. EasyMile provided technical support for the City of Arlington and First Transit, especially at the start of the pilot. It provided the vehicle specifications, instructions on operation and maintenance, and offered certification courses and training documents for safety attendants. The First Transit operations manager used these materials from the vendor to provide a 5-hour training course for new safety attendants. The City of Arlington arranged storage and electric charging for the vehicles. One vehicle was kept at the convention center and one was kept at Globe Life Park. In both locations, the vehicles were charged on wall outlets at no additional cost to the city. The City of Arlington also secured the use of their park maintenance vehicle to provide a portable hose to wash down the vehicles in the parking lot next to where they were stored. Additionally, a City of Arlington planning department employee was trained as a certified operator to provide backup and demonstrate the technology to other stakeholders. Safety and Safety Training EasyMile provided detailed specifications and instructions on how the EZ10 Gen-1 vehicle worked. First Transit adapted best practices from its rail operations, including standardized protocols that followed checklists, dispatching, and maintenance. The City of Arlington provided oversight with an emphasis on safety. Texas has no specific requirements for AV safety operators. At the direction of the City, First Transit developed and executed safety training for the operators. Before commencing service, the First Transit manager for this project, who was also their Lead Automated Vehicles Operations 6 EasyMile noted in its interviews that it provides guidance to First Transit.

66 Low-Speed Automated Vehicles (LSAVs) in Public Transportation Manager, completed a 4-day training course with EasyMile. Based on that information, she then prepared a training program for the operators, which was adapted from other transit manuals. During this training, operators were taught how to perform normal operations. This included starting the vehicle in manual mode, switching from manual to automatic and vice versa, and getting sign-off from the lead operator. They were also taught how to evacuate riders in an emergency and call security. Coordination with Law Enforcement, Public Safety, and Emergency Responders Before the launch of the Milo service, planning staff briefed law enforcement, the fire depart- ment, and other first responders. The briefing included familiarization with the vehicle and information on incidents particular to electric vehicles, such as a fire with the electric motor. During operations, no incident occurred requiring contact with law enforcement or other safety or emergency personnel. Information was obtained that a group of attendees of a baseball game shook the vehicle while passengers were inside. (Some would refer to this as “robot bullying”). The Lead Automated Vehicles Operation’s Manager (as the on-board safety attendant) exited the vehicle and defused the situation. She indicated that this incident was not unlike events that occur on rail and bus in regular daily operations on rail systems. She did not contact law enforcement. Incidents First Transit logged all incidents and shared that information with the City of Arlington. According to the incident log reviewed by the research team, there were interruptions involving the automated driving feature, though none resulting in serious injury. The typical incidents related to loss of localization, especially under a bridge, challenges navigating a section of path with limited clearance, and interference from debris in the path. In one incident, the vehicle failed to maintain speed along a stretch of steep incline. EasyMile engineers were dispatched to review the vehicle’s operation, and after that incident, they recom- mended that only 10 passengers be included in future shuttle rides to reduce weight and facilitate higher speeds. Project Evaluation The Milo shuttle was intended to be an innovative transportation project, showcasing auto- mated technology to a wide variety of regional residents and visitors to the Entertainment District. In the City of Arlington’s words, the specific goals were to (1) test automated vehicle technology in a real-world setting and (2) educate the public and raise awareness of automated vehicle technology. The City of Arlington determined that the pilot met these goals over the course of the year-long operation and estimated that over 1,600 rides were provided across more than 110 events. Key quantitative indicators collected included • Number of Rides • Number of Passengers • Number of Disengagements • Number of Events Served User Acceptance The City of Arlington aimed to expose residents and visitors to automated technology in order to both gain familiarity with automated technology and to understand how riders experienced

LSAV Case Studies 67 the service. The City of Arlington conducted surveys of riders during the year-long pilot. Among those who rode the Milo shuttle, 99 percent enjoyed their experience and felt safe. 97 percent of riders who were surveyed support the use of AV technology more broadly. Other comments noted that the experience had changed their perception of automated technology and expressed a desire for the City of Arlington to expand the service area. The City of Arlington did not survey nonriders. Automated Technology Performance The City of Arlington reported that overall performance of the LSAV technology met its goals. First Transit and the City of Arlington indicated that there were frequent vehicle stops and disengagements caused by grass trimmings and other obstacles on the path. Further, when Milo passed under the AT&T bridge, the vehicle would lose GPS localization, which sometimes resulted in the operators shifting to manual operations. Transportation System Integration Milo was a 1-year pilot project in an off-street campus environment, with a clearly defined use case and a public educational mission. The City of Arlington did not charge fares and did not integrate the Milo service with other regional transportation services. In the future, its staff anticipates integrating automated shuttle services with other on-demand options (e.g., Via) in a single platform/app. Rider Accessibility/Universal Design The City of Arlington selected the EasyMile EZ10 Gen-1 shuttle because the vehicle has an automated wheelchair ramp, allowing it to accommodate wheelchair users. The vehicle did not have tie-downs. The vehicle also did not include assistive technology such as a human– machine interface allowing for people to communicate directly with the vehicle through haptics, display screens, or voiced communications from Milo. An attendant was available to answer passenger questions. Lessons Learned The City of Arlington identified a series of lessons learned: A contract with a third-party operator for operation of low-speed automated vehicles is crucial. Given that the City of Arlington had no transportation operations services, it recommends a turnkey operation with one contract for vehicles and for operations. Staff continuity in on-board attendants can be difficult for event-based service with irregular hours. The City of Arlington experienced high turnover among on-board attendants. This may be attributable to the irregular hours, combined with relatively low pay. Operators can prepare for this in their training and service plans. Site assessment and mapping are required before initiating service. The site or route assess- ment will reveal any requirements for infrastructure modification. In the case of Arlington, EasyMile noted that the absence of firm edges typically provided by the built environment, whether curbs or buildings, was not present. Accordingly, Arlington installed landscaping features to help with localizations. Practitioners need to plan for maintenance and repair of vehicles. Staff from the City of Arlington recommended ensuring that parts are available locally and having towing equipment available if vehicle repairs are needed. Off-street, event-based shuttles can operate in an off-street operating environment. The City of Arlington found that the EZ10 functioned satisfactorily for event-based services on the

68 Low-Speed Automated Vehicles (LSAVs) in Public Transportation off-street paths of the Entertainment District. Residents and visitors experienced the technology. Arlington reported that the pilot demonstrated value in transporting people from the stadium to remote parking. The service was particularly valuable for those susceptible to Arlington’s more extreme heat in the summer. D.2 ARIBO Program at Fort Bragg Case Study Project Profile The Applied Robotics for Installations and Base Operations (ARIBO) project was first conceived in 2011. The ARIBO program introduced automated ground vehicle systems into a semi- controlled environment. Pilots were selected based on a high potential for the automated ground system to perform a task safely and reliably as well as to add quantifiable value to the selected installation. The objectives for ARIBO include • Socializing users and nonusers with automated systems. • Identifying operational challenges and mitigation strategies, and generating empirical data including, but not limited to, performance, reliability, and maintenance. Collaboration with Industry and Other Federal Agencies ARIBO was intended to accelerate the practical use of automated vehicles. ARIBO became the first significant implementation of automated vehicles to provide transportation services in the United States. The program started early demonstrations to stakeholders in 2014 using Cushman-Robotic Research shuttles (depicted in Figure D.2.1) and was led by the United States Army (Army) and the Army’s Tank Automotive Research, Development and Engineering Center (TARDEC). The Fort Bragg project was the first full implementation, in which “Wounded Warriors” were transported to their medical appointments at Womack Army Medical Center (WAMC) (see route map in Figure D.2.2). “These frequent, short-distance door-to-door trips between the medical barracks and appointments at the medical center provided an environment to better understand the systemic impacts of driverless vehicle integration.”1 Source: K. E. Schaefer, A. N. Foots, and E. R. Straub, “Applied Robotics for Installations and Base Operations: User Perceptions of a Driverless Vehicle at Fort Bragg,” Presented at U.S. Army Research Laboratory, January 2018. Figure D.2.1. ARIBO Fort Bragg LSAV shuttle. 1 Schaefer, Kristin, et al., “Applied Robotics for Installations and Base Operations: User Perceptions of a Driverless Vehicle at Fort Bragg.” Army Research Laboratory, Jan. 2018, https://apps.dtic.mil/ditch/tr/fulltext/u2/1044446.pdf.

LSAV Case Studies 69 This implementation site was a roughly 1-square-mile area. The route included five pick-up/ drop-off locations. The ARIBO Fort Bragg project overview is in Table D.2.1. Geographic Context Fort Bragg is an Army base in North Carolina. It has an on-base residential population of 29,183, and 50,000 active duty personnel work there, making it one of the largest bases in the world. It is home to several departments and groups with diverse mobility needs. Notably, it is the site of a Warrior Transition Battalion, one of nine used by the Army, which aided the transi- tion of injured soldiers to subsequent opportunities. The area is a humid subtropical climate. Winters tend to be mild, though some snowfall occurs. Summers, however, can be quite hot with July average temperatures in the 90s with high humidity. The base covers 251 square miles with on-base transportation primarily being individual automobiles. ARIBO at Fort Bragg represents a confluence of military research and development into auto- mated vehicles and systems (such as the DARPA Grand Challenge) and the increasing traffic problems at military bases. TARDEC led the development of this program to further commer- cialize and use applied robotics to solve real challenges for the military. Figure D.2.2. ARIBO Fort Bragg route map. Table D.2.1. ARIBO LSAV shuttle overview. Topic Applied Robotics for Installations and Base Operations (ARIBO) Fort Bragg Project Use Case Paratransit (on-demand route) Route Description Defined routes with on-demand service with 5 stops inside a military base Vehicle Capacity 3, but ridership was 1 passenger normally + safety attendant Speed Max speed 25 mph, average speed 8–10 mph Operator Robotic Research Vehicle Provider Robotic Research Fort Bragg Project Team Army, Fort Bragg Transportation, WAMC, TARDEC, DOT ATTRI, Comet, UMTRI, University of MI SMART ARIBO Project Team Army, TARDEC, Induct/Navya, NASA, Stanford, SLAC, DOE, DOT, West Point, Comet, UMTRI, University of MI SMART, UTARI, TTI

70 Low-Speed Automated Vehicles (LSAVs) in Public Transportation Transportation Challenges at United States Military Bases In the United States, military bases are closer in scale to small cities with correspondingly high population densities. They have the same problems of traffic congestion (Figure D.2.3) and parking constraints that exist in other cities.2 One of the goals of ARIBO was to explore how automated transportation could decrease the need for individual vehicles on base. The approach was to add more travel options so that indi- viduals could explore public transit options such as shared vans and other carpooling. ARIBO was intended to introduce the shared mobility concept to a military base and accelerate shared usage in the future. Better Management of Vehicle Fleets Through Sharing The Army has an extensive fleet of vehicles, over 50,000 in total. Most of these are passenger vehicles. Expanded sharing of these assets has been discussed to make this fleet more efficient. Automated shuttles and vehicles were suggested to facilitate that process—another reason for implementing ARIBO at Fort Bragg. Economic Development and Innovation ARIBO was also considered to bolster innovation and economic development in high-growth industries. Proponents argue the program could spark the development of new markets and facilitate advanced manufacturing in batteries, automated technology, and vehicle platforms.3, 4 Planning Fort Bragg was selected for the first ARIBO implementation because of its distinct transpor- tation needs and use cases. As noted previously, Fort Bragg is the site of one of the largest of nine Warrior Transition Battalions. The WAMC and the barracks associated with the Warrior Photo courtesy of Northwest Florida Daily News. Figure D.2.3. Military-base-related congestion in NW Florida. 2 A 2011 report highlighted worsening traffic congestion as a significant issue inhibiting military efficiency. 3 Clothier, Corey, “ARIBO Robotic Vehicle CPS Test-Beds in 2014,” https://smartamerica.org/wp-content/uploads/2014/05/ ARIBO.png. 4 Defense Advanced Research Projects Agency (DARPA), “The DARPA Grand Challenge: Ten Years Later,” Mar. 13, 2014, https://www.darpa.mil/news-events/2014-03-13.

LSAV Case Studies 71 Transition Battalion are close in proximity, but still too far for some people to walk (especially those with mobility challenges). Additionally, traditional shuttle service results in long travel and wait times for the Warrior Transition Unit service members. As a result, service members missed medical appointments, costing the Warrior Transition Battalion money and time while the service members lost the opportunity for much-needed care. Planning, Permitting, and Procurement Phase 1 of the ARIBO project started in May 2016, with Phase 2 initiated in January 2017. Procurement was handled through military contracts, primarily making use of appropriated research and development dollars. The planning for the ARIBO project was unique regarding the permitting process. Military bases are given wide latitude to modify their physical environment at the discretion of the Installation Commander. Further, common U.S. DOT regulations, local land use, and municipal codes do not apply. As a result, ARIBO was not required to follow the approval processes of typical civilian agencies or private landowners. There were some specific requirements for implementing ARIBO at Fort Bragg. Permission had to be obtained from multiple organizations on Fort Bragg (transportation, hospital, legal) and ultimately the Installation Commander had to approve the ARIBO pilot. Insurance was provided by the automated vehicle vendor with support of the ARIBO team identifying an insurance company. One outcome of the project was to accelerate the develop- ment and issuance of commercial automated vehicle insurance by identifying and partnering with a reputable insurance company to issue a policy. ARIBO procured a set of Cushman-6 shuttles and contracted with Robotic Research, an auto- mated technology vendor to outfit the shuttles with automated driving sensors and software. Site Assessment The ARIBO team worked with Fort Bragg planners and leaders to assess and determine the optimal route to meet the ARIBO objectives and provide the most benefit to Fort Bragg stake- holders. The selected route covered the distance between the medical barracks and the WAMC. The route included • A mix of private roads and parking lots and a signalized interchange. • Stops (5 in total) 1⁄3- and ¾-mi intervals at designated routes between the barracks and the WAMC. • Pedestrian sidewalks. • A crossing of a four-lane divided road, which had several blind spots. • Traffic circles at the WAMC entrances. Infrastructure Requirements Stationary LiDAR sensors were installed to cover the blind spots and augment the perception sensors on the shuttles. No major physical infrastructure changes were made. Operations Program operations were planned and managed by Army personnel. A TARDEC consultant led initial operations and planning in conjunction with the University of Michigan’s Connected Vehicle Proving Center and Transportation Research Institute SMART (Sustainable Mobility & Accessibility Research & Transformation) teams. Regular operations for the program were from Monday to Friday between 8:00 a.m. and 3:30 p.m. The operations team stored and charged vehicles in the parking structure next to the Warrior Transition Battalion barracks.

72 Low-Speed Automated Vehicles (LSAVs) in Public Transportation Accessibility Many of the passengers were injured and needed a vehicle that could accommodate their mobility limitations. As a result, the vehicles were wheelchair accessible. Human–machine interaction deployed to aid accessibility were discussed to include screens showing vehicle and route status. Safety Testing and Automated Development Before the pilot was operational, extensive testing was carried out for the project to ensure all requirements were met and program goals could be safely executed. Project leaders decided to phase in the self-driving features over time to build trust in the safety of the automated systems. Phase 1 of the project was a human-driver-operated shuttle service using the Cushman-6 shuttles. During this phase, vehicle data and driving data were collected for testing the under- lying architecture and autonomy algorithms. This was described as having a dual purpose. First, because a trained human driver operated the vehicle, ARIBO staff was able to evaluate the performance of the sensing and control systems in real-world conditions without increasing risk to any passengers. Second, they were able to not only evaluate the vehicle’s performance against real-world conditions but also compare performance and decision making against the human-driven baseline using a recording scheme and automated analysis based on researcher- defined upper and lower limits of acceptable performance. Results of the analysis provided data for the decision to proceed to the next phase of the project. Phase 2 of the ARIBO project transitioned the role of the driver to a safety operator. The role of the safety operator was to intervene with the vehicle’s autonomy only in case of an emergency or potential vehicle error. The comparison of the human-operated system versus the automated system found that the system reliability was comparable and that AVs adhered to the rules of the road at a level comparable to human drivers. The vehicle was subsequently developed further for the commercial market. Project Evaluation The main goals for the project were demonstration of the technology as a viable form of trans- portation and further development of the automated technology. There was limited ridership, with only 112 riders during Phase 1 of the program and only 15 riders during Phase 2, but data were collected, analyzed, and reported. User Acceptance A survey of users of the vehicle demonstrated the following: • 88 percent were positive on the vehicle’s perceived intelligence • 85 percent were positive on perceived autonomy • 81 percent were positive on perceived trustworthiness • 77 percent were positive on perceived safety • 69 percent were positive on perceived usefulness Automated Technology Performance The automated control software performed well, and participants were generally satisfied with the vehicle’s ability to follow traffic laws and navigate through parking lots and inter sections. No incidents occurred during the pilot as a result of the AV technology or external factors impacting the shuttle.

LSAV Case Studies 73 Transportation System Integration The ARIBO on-demand shuttle operated independently of other transportation systems on base and focused solely on transporting people to/from/within the WAMC and nearby barracks. It continued to operate in this capacity through the duration of the ARIBO pilot. Rider Accessibility/Universal Design The customized design of the Cushman-6 shuttle by the vendor provided an accessible design and wheelchair accessibility as a central objective. The vehicle operated as fully wheelchair accessible. However, more than one person mentioned that there was insufficient legroom in the second row for tall passengers or those with serious injuries.5 Lessons Learned The initial ARIBO pilot at Fort Bragg tested a use case for LSAVs that is applicable to military bases and civilian contexts. The processes, research, and lessons learned formed the basis for additional pilots, early safety operational practices and standards, development of an automated vehicle data recorder and analytics tool, and led to the creation of an automated vehicle insurance product. Further, the Army and other ARIBO partners made the following observations about lessons learned from this project:6 Users believed that the automated capabilities of the ARIBO vehicle were acceptable or would be in the future. Those who doubted safety had only a minimal level of understanding of the system’s operations, suggesting further education would help increase trust. Design features to improve passenger comfort for wounded soldiers would improve comfort and safety for all riders. These improvements include improving seats, extending legroom, safety belts, and effective communication of safety features and options. Additional alerts or warnings would improve pedestrian safety. This includes verbal (auditory) acknowledgment when the vehicle is paused for a pedestrian, sounding a horn to let pedestrians know the vehicle is waiting for them, or providing signs explaining that this is a driverless vehicle. Discrete design modification could increase passenger trust or confidence. These changes would communicate in multiple ways that a vehicle is driverless. This could include changes to vehicle size, color, and signage that allow the vehicle to communicate with other road users. The vehicle system would benefit from a user display. This display would communicate vehicle awareness and actions, including navigation-specific feedback (e.g., estimated arrival time, changes in behavior or routes), error reporting (e.g., ability to report issues or errors, count of errors), and a means to communicate to a human monitor. The ARIBO program developed automated vehicle technology across the country. Members of the team have consulted widely, including pilot projects in seven states across the United States and sites in two other countries. The project has also influenced the SAE Automated Vehicle Safety Consortium, impacting standards development. The project has led to the develop- ment of automated vehicle performance measuring tools and influenced the development of the first automated vehicle insurance product. In addition, the introduction of automated shuttles into the U.S. market (e.g., Navya, EasyMile, and Local Motors) began with ARIBO’s demonstrations on military bases. 5 Schaefer, Kristin, et al., “Applied Robotics for Installations and Base Operations: User Perceptions of a Driverless Vehicle at Fort Bragg.” Army Research Laboratory, Jan. 2018, p. 16. https://apps.dtic.mil/dtic/tr/fulltext/u2/1044446.pdf. 6 Ibid.

74 Low-Speed Automated Vehicles (LSAVs) in Public Transportation D.3 Las Vegas AAA Free Self-Driving Shuttle Case Study Project Profile Beginning in 2017, the City of Las Vegas (City) and the Regional Transportation Commis- sion of Southern Nevada (RTC), along with Keolis North America (Keolis) and the American Automobile Association Northern California, Nevada and Utah (AAA), operated the nation’s first public transit AV service in mixed-use traffic on a downtown loop. A single Navya ARMA low-speed automated vehicle (Figure D.3.1) provided rides to residents and Las Vegas visitors in mixed-use right-of-way and transit-like driving conditions. The shuttle operated for 1,515 hours and had an overall ridership of 32,827 during the pilot.1 AAA’s main goals for the pilot were 1. Gauge the public’s satisfaction and acceptance of automated vehicles. 2. Gather relevant rider feedback for the service itself. In addition, The City focused on how DSRC, connected vehicle, and infrastructure technology that was under development for the Las Vegas Valley would be relevant in the case of shared AV service. The City and AAA shared the objective of implementing a regular service geared for the general public to gain real-world experience of the technology. Both the City and AAA report that their goals were met. Geographic Context With a population of over 600,000, Las Vegas forms the center of a metro area of more than 2 million people. At least 42 million people visited the Las Vegas Valley (Valley) in 2019,2 making it one of the most popular tourist destinations in the United States. The Las Vegas Strip to the south attracts most of the Valley’s tourists with its well-known entertainment and resort locations. In recent years, the City rebranded its downtown as an Innovation District and undertook revitalization efforts centered on Fremont Street. The City hopes to leverage Zappos’ headquar- ters to develop a more vibrant tech economy downtown. The City also looks to enhance its other efforts to build the district with innovative transportation technology, including automated vehicles. The AAA Free Self-Driving Shuttle operated in a 0.6-mi fixed route (Table D.3.1) in down- town Las Vegas near Fremont Street, with three fixed stops and no fare. The shuttle ran in a Reprinted by permission of AAA Nevada. Figure D.3.1. AAA self-driving shuttle. 1 Bell, Maurice, Personal Interview. Keolis North America. Dec. 2018. 2 Las Vegas Convention and Visitors Authority, “Las Vegas Visitor Statistics.” https://www.lvcva.com/stats-and-facts/visitor-statistics/.

LSAV Case Studies 75 loop along Fremont, South 8th Street, East Carson Avenue, and South Las Vegas Boulevard and crossed eight intersections, of which six had signals (Figure D.3.2). The shuttle typically operated at speeds of less than 10 mph. Except for very hot days in the summer, interest was high and there was usually a line to ride in the shuttle, with ridership consisting primarily of visitors to Las Vegas. The City has a subtropical desert climate, with extremely hot summer days. Furthermore, heat affected LSAV performance, particularly the battery life, which was diminished significantly by air conditioning constantly running on high. Transportation Context The City and RTC have been planning for years to improve public transportation in the Las Vegas metropolitan area, especially in the congested corridor along the Las Vegas Strip and around downtown Las Vegas. Downtown Las Vegas is currently accessible by I-15 and I-515, as well as numerous other surface streets. Since 2004, RTC has been expanding a system of bus rapid transit lite express services (known as MAX), with enhanced shelters, limited stops, and transit Topic AAA Free Self-Driving Shuttle Use Case Shuttle (on-street) Route Description 1 fixed route on a closed loop, 3 stops Route Lengths 0.6 mi Vehicle Capacity 15 (11 seated; 4 standing) Speed 15 mph max speed, 8 mph average operating speed Operator Keolis Vehicle Provider/Vendor Navya Project Managers City of Las Vegas Next Stage GoMed shuttle, funded by U.S. DOT BUILD grant Table D.3.1. AAA LSAV shuttle overview. Figure D.3.2. AAA LSAV shuttle route.

76 Low-Speed Automated Vehicles (LSAVs) in Public Transportation signal priority. It is currently undergoing a transit planning process called OnBoard, asking for input from the community on future service. In the past few years, during January when Las Vegas hosts the Consumer Electronics Show, Las Vegas has become the center of the transportation technology world. At this large-scale event, Las Vegas has hosted numerous demonstrations of AV technology. The AAA Free Self-Driving Shuttle project represented the convergence of an active and willing private sponsor and an interested local government that had already been working with a transport operator (Keolis) and automated shuttle vendor (Navya). Las Vegas was an appealing location to hold the pilot given the large number of tourists who would be exposed to the service as well as a state and local government eager to test the technology and demonstrate its value in a real setting. The next phase of the City and RTC’s LSAV pilots, branded “GoMed,” will be a 4.5-mi shuttle to connect downtown Las Vegas and a nearby medical campus (Figure D.3.3). This project was awarded federal funding through a BUILD grant of $5.3 million in December 2018 and dubbed the Las Vegas Medical District Automated Circulator and Pedestrian Safety Project. The project includes enhanced stop shelters and pedestrian improvements, as well as auto- mated vehicle service. Four vehicles will serve the shuttle route providing both a downtown circulator and point A to B shuttle between downtown and the medical campus across I-15. Reprinted by permission of City of Las Vegas. Figure D.3.3. GoMed schematic.

LSAV Case Studies 77 The project’s goal is to connect residents “to healthcare, employment, education, and other vital services.”3 In addition to the GoMed shuttle project, the Las Vegas area is also hosting a robo taxi project jointly done by Lyft and Aptiv. Aptiv is providing BMW vehicles modified with its automated software to be placed on Lyft’s platform. This project started in January 2018 and remains ongoing, surpassing 50,000 total rides as of June 2019.4 Planning The State of Nevada, the RTC, and the City of Las Vegas have each engaged in a multiyear plan- ning and policy effort to advance smart mobility, including LSAVs. The Department of Motor Vehicles by statute has immediate regulatory oversight of all types of AV deployments and has created a permitting/licensing process for AVs. Nevada has special licensing and Research and Development plates for registered vehicles.5 As part of an economic development effort, the Department of Commerce and a special advisor to the Governor are charged with promoting smart mobility initiatives to accelerate advanced technology in Southern Nevada. Metropolitan Transportation and Mobility Planning The RTC serves as the Las Vegas MPO, leads the traffic engineering system, and manages public transit operations in the urbanized area. RTC and the City have been investing in connected vehicle infrastructure in line with state policies to promote AVs. The RTC has reflected AV technology in its long-range transportation plan. Adopted in 2017, Access 2040 mentions the potential impacts of AVs on the regional transportation system and includes AV mode share as a 2040 transportation system indicator under the secondary strategy of innovative planning and emerging technologies. In its maintenance program, RTC calls for maximizing pavement quality, improving lane markings, and lane narrowing (potentially) in preparation for automated vehicles.6 The City and RTC are also working in the short term to improve the regional transportation system and bolster the deployment of AVs. Within the region, there are three main transportation programs to bolster innovation: • DSRC regional network: The City has equipped over 70 intersections in downtown Las Vegas with this technology. Clark County, in partnership with Aptiv, outfitted an additional 43 inter- sections along the Las Vegas Strip with DSRC. • Automated vehicles: RTC and the City seek to expand AV piloting to enhance mobility, especially downtown and between key activity nodes. • Other transportation improvements: RTC is implementing bus rapid transit expansion along key corridors. 3 Regional Transportation Commission of Southern Nevada, “RTC, City of Las Vegas Receive $5.3M Federal Grant for Downtown Autonomous Circulator Project.” Mass Transit, Dec. 11, 2018, https://www.masstransitmag.com/alt-mobility/autonomous- vehicles/press-release/21036657/regional-transportation-commission-of-southern-nevada-rtc-rtc-city-of-las-vegas- receive-53m-federal-grant-for-downtown-autonomous-circulator-project. 4 Korosec, Kristen, “Aptiv’s Self-Driving BMWs Have Made More Than 50,000 Rides on the Lyft App in Las Vegas.” TechCrunch, June 3, 2019, https://techcrunch.com/2019/06/03/aptivs-self-driving-bmws-have-made-more-than-50000-rides-on-the-lyft- app-in-las-vegas/. 5 Nevada Department of Motor Vehicles, “Autonomous Vehicle Testing Registry Application.” https://www.dmvnv.com/ pdfforms/obl326.pdf. 6 Regional Transportation Commission of Southern Nevada, “Access 2040: Enhancing Mobility for Southern Nevada Residents.” Feb. 9, 2017, p.15.

78 Low-Speed Automated Vehicles (LSAVs) in Public Transportation In addition, the City has established an innovation district branded as “Innovate Vegas.” This initiative has focused on economic development and exploring new AV options to bring people to destinations and employment in the vicinity. Project Planning By mid-2017, the City sought to expand from demonstrations to a longer-term pilot of AVs. At the same time, AAA wanted to find a city to sponsor an AV shuttle project designed to study the public’s perception after they gained firsthand, real-world experience with AV technology. AAA’s objective in sponsoring a public “self-driving” shuttle was to demonstrate the potential for automated driving technology to improve traffic safety and potentially reduce the more than 37,000 traffic fatalities annually. AAA, Keolis, and the City joined together in a public–private partnership to plan the shuttle in a loop route downtown. Keolis and the City executed a contract for the shuttle operations. AAA hired technology deployment specialists to identify the use case, transportation market, and technology partners. Keolis was hired to develop a concept of operations and to run the year-long pilot. The City Engineer served as the overall project manager from the City, and the RTC played a coordinating role given its operational responsibilities for the transit system and traffic engineering system. AAA had key requirements for this pilot project; these included that the service needed to be permitted per local regulations, and the project had to have support from a local agency and sponsor. To boost visibility and ridership, the technology also had to meet the requirements of the site, which was in an area of high foot traffic. All of AAA’s considerations were met by the City, and the Navya vehicle was well suited to the busy downtown district, which had large numbers of tourists who would be introduced to the technology. The City and RTC wanted the service to be tested with the new DSRC network to inform their future investments in support of automated vehicle deployment. Procurement and Preparation AAA Northern California, Nevada, and Utah funded the project. The City contracted with Keolis to provide a 1-year pilot service along the downtown loop to demonstrate this technology, using one Navya ARMA shuttle vehicle (now rebranded as AUTONOM). Permitting and Waivers Navya required two levels of permission—federal and state—for its vehicle to operate on public roads and to provide rides in Las Vegas. Navya sought approval from NHTSA to operate on public roads.7 Keolis and Navya also were required to register and license the vehicle with the state of Nevada.8 Navya’s AUTONOM shuttle does not have side mirrors, a brake pedal, and a driver’s seat as required by FMVSS. Navya completed an FMVSS gap analysis, explaining which specific sections the vehicle does not meet, and demonstrating how they can still meet an equivalent level of func- tion and provide an equal level of safety. For example, the eight LiDAR sensors provide much 7 National Highway Traffic Safety Administration, “Importation of Motor Vehicles and Motor Vehicle Equipment Subject to Federal Motor Vehicle Safety, Bumper, and Theft Prevention Standards.” https://www.nhtsa.gov/sites/nhtsa.dot.gov/files/ hs7_r.v.7.pdf. 8 Nevada State Legislature, Nevada Revised Statutes Chapter 482A—Autonomous Vehicles. https://www.leg.state.nv.us/NRS/ NRS-482A.html.

LSAV Case Studies 79 more information about road and traffic conditions than the human eye. (LiDAR is a surveying method that measures distance to a target by illuminating the target with pulsed laser light and measuring the reflected pulses with a sensor.) The HS-7 waiver, originally intended for exotic sports cars and auto shows, had been used until recently to facilitate research through temporary trials and pilots. Keolis and Navya met state-level requirements in Nevada in order to operate on public roads. They filed for a state permit and acquired two sets of special AV license plates for the Navya vehicle. There are no licensing requirements for the safety attendants. The City had no permitting requirements for the Las Vegas project, but law enforcement and emergency responders needed to be notified that the shuttle was operating (more in Safety Training and Coordination with Law Enforcement section). Insurance Insurance on the Navya vehicle was provided by three main companies: • Zurich American for general commercial liability • RLI for automotive vehicle liability • Westco for workers compensation liability Site Assessment In this project, Navya performed the site assessment. Keolis deemed the ODD appropriate for the operation proposed for the Navya AUTONOM. The route was in mixed traffic, in a lane separated by cones. There were eight intersections of which six were signalized and two were stop-controlled. Navya further identified “a large amount of pedestrian traffic.” Average traffic speeds were low, typically below 10 mph. Infrastructure Requirements The main infrastructure requirement by the project team was the inclusion of DSRC RSUs. These RSUs were procured and installed by the City. There were six DSRC units along the LSAV route, and the RTC had a lead role in ensuring information technology robustness and cybersecurity. Operations During the trial period, the AAA Free Self-Driving Shuttle generally operated between 11:00 a.m. and 7:00 p.m. During the summer, Keolis adjusted the schedule to between 1:00 p.m. and 9:00 p.m. to avoid the heat. The extreme heat necessitated constant air conditioning, compromising battery performance and adversely impacting peoples’ willingness to wait for the shuttle outside. In August 2018, Keolis suspended operations entirely because of excessively high heat. During the period when Keolis suspended service for excessive heat, their vehicles were sent back to Navya and were retrofitted with equipment to mitigate range degradation caused by the heat. Keolis resumed service in September and October of 2019. Mobility Service Provider Keolis operated the AAA Free Self-Driving Shuttle. Keolis, a mobility/transit service provider operating bus and rail transit service in the United States and Europe, recruited, trained, and supervised all safety attendants during the pilot. Keolis also oversaw storage and charging of the vehicles in a city-contracted facility. Keolis also provided AAA, Navya, RTC, and the City information on ridership, trips, and battery utilizations. Separately, Keolis provided Navya with technical disengagement reports and other logs. Neither the City, RTC, nor the project sponsor AAA received these reports on disengagements or other incidents.

80 Low-Speed Automated Vehicles (LSAVs) in Public Transportation Operational Challenges Keolis and AAA reported that, generally, operations ran smoothly. AAA noted that it was often a challenge to keep the stop locations free of transportation network company vehicles (e.g., Uber/Lyft). If there were other vehicles in the shuttle stop, the automated control algorithm would not be able to decide on another stop location, and the operator would have to shift into manual mode. Reported wait times during peak periods often were up to 30 to 40 minutes. During periods of high heat, the air conditioning significantly affected battery range and impacted the feasible hours of operation. Development of Safety Protocols Navya provided specifications and other information on how the ARMA shuttle worked. Keolis developed a detailed safety operations protocol for the AAA Free Self-Driving Shuttle, based on its extensive experience in other transit systems. The RTC followed its own safety protocols with respect to transit. Safety Training and Coordination with Law Enforcement Keolis and AAA identified safety as a key concern for the pilot. This was reflected in the selection and training of the shuttle’s in-vehicle attendants. Keolis developed an extensive training program to provide a detailed understanding of the technology and its capabilities. The company noted that this allowed the attendants to be ambassadors for the technology and to provide an addi- tional layer of safety. AV operators received certifications upon completion of the training. Keolis briefed law enforcement officers and first responders at a workshop on the vehicle’s technology as well as planned operations before the launch of the service. The workshop focused on what to do in the case of crashes and whom to contact in the case of an incident. It involved training the fire department on how to respond to an electrical vehicle fire, including whether to use water. Keolis explained how police could get footage from the vehicle in case it was needed for investigation. A later review by the City indicated that training should be extended to other law enforce- ment personnel who might cooperate in a joint emergency response. The need for this training was made clear during the response to the attack on the Route 91 Harvest music festival in October 2017, when outside law enforcement encountered the Navya shuttle and was not familiar with its sensor setup and capabilities. Incidents Less than an hour into operations on November 8, 2017, a truck crashed into the Navya ARMA shuttle. No one was injured and no other serious safety incidents were recorded. After a review of the data, Navya indicated that the automated control software performed as designed, and no changes were needed in the algorithms.9 NTSB investigated the crash (HWY18FH001) “to better understand how self-driving vehicles interact with their environment and the other human-driven vehicles around them.” They note that “while there have been other crashes of self-driving vehicles, this crash is the first of a self-driving vehicle operating in public service. Our decision to investigate this crash aligns 9 Other analysts raised the consideration that a more adaptive algorithm and louder horn may be needed in a mixed-traffic envi- ronment where dealing with driving culture and communicating with people on the road is necessary to stay safe. Bigelow, Pete. “Self-Driving Shuttle Returns to Service, but NTSB Wants to Examine Vegas Crash.” Car and Driver, November 13, 2017. https:// www.caranddriver.com/news/a15339189/self-driving-shuttle-returns-to-service-but-ntsb-wants-to-examine-vegas-crash/.

LSAV Case Studies 81 with our process of deciding to investigate those highway crashes that can advance our knowledge of safety issues.”10 This investigation was recently completed and the NTSB cited the truck driver’s actions as the probable cause for the collision and the AV attendant’s lack of easy access to a manual controller as a contributing factor for the collision. The NTSB made no recommendations as a result of their investigation but will continue to monitor the develop- ment of those vehicles to better understand their potential safety impacts and any unintended consequences. Project Evaluation The City and AAA both sought regular shared automated shuttle service that maximized ridership and visibility for the technology. The public partners sought to test the performance of the DSRC infrastructure and the ability of the LSAV to operate in conditions like those of public transit. AAA sought broad exposure of a successful AV operation in order to understand consumer acceptance of the technology. AAA measured this through the level of rider satis- faction in post-trip surveys. Ridership By the end of the pilot, over 32,000 people rode on the shuttle. User Acceptance The core objective of this project was to gauge the public’s reaction to an AV operated as a public shared transportation on a public road. There were two surveys conducted in relation to this project. One survey was administered by AAA to survey riders on their experience while the second was an academic study by researchers at the University of Nevada–Las Vegas (UNLV). Riders who completed the AAA survey indicated they enjoyed the experience. They recorded an average response of 4.9/5 stars. Additionally, 96 percent of the respondents stated they would recommend the service to their friends. Based on general feelings toward AVs before and after the ride, there was a 30 percent improvement on overall AV sentiment. The UNLV research effort surveyed riders of the shuttle as well as the general population around Las Vegas, measuring their perceptions and attitudes regarding “autonomous and con- nected vehicles (ACVs).”11 In general, respondents were positive about the potential of the tech- nology with riders of the shuttle even more so. For example, 68 percent of the general population wanted to see automated transportation more widely available compared to 82 percent of shuttle riders. The researchers suggest that firsthand experience is important to alleviate negative and uncertain feelings about ACV technology and that trial periods are useful as an opportunity to build trust. Automated Technology Performance Keolis noted that the Navya shuttle worked as intended, with no incidents with the automated control software. The shuttle successfully navigated a mixed-traffic environment, within a lane marked with cones designated for its operation. The vehicle traveled at speeds of up to 10 mph along the fixed route, with an average speed of approximately 8 mph. 10 NTSB, “NTSB Launches Investigators to Self-Driving Vehicle Crash.” Nov. 11, 2017, https://www.ntsb.gov/news/press-releases/ Pages/PR20171110.aspx. 11 The researchers recorded survey data from 236 respondents from the general population and 153 respondents who were shuttle riders. They used a stratified stated preference approach, with a 30-question survey of people in the general popula- tion and a 17-question survey of shuttle riders. “Perceptions and Attitudes Towards the Deployment of Autonomous and Connected Vehicles: Insights from Las Vegas, Nevada” (Submitted for TRB Annual Meeting 2020, July 31, 2019).

82 Low-Speed Automated Vehicles (LSAVs) in Public Transportation Transportation System Integration The AAA Free Self-Driving Shuttle was designed as a use case that was not integrated into RTC’s transit system (with either integrated fare or coordinated service and scheduling). The vehicle was integrated with local infrastructure such as DSRC RSUs. The City considered the pilot to be a useful validation of the connected vehicle DSRC units, which provided redundancy for the vehicle’s positioning, and relayed information about traffic light status to the vehicle, including signal phase/timing data. Rider Accessibility/Universal Design The Navya AUTONOM shuttle is accessible for people using wheelchairs via a built-in ramp. The on-board attendant oversaw securing the wheelchair once it was on board the vehicle, which he or she could do manually. This vehicle did not have a human–machine interface to allow communication with passengers who had a visual, cognitive, or hearing-impaired passenger. The attendant was available to answer any questions that arose during travel. Lessons Learned AAA, Keolis, and the City of Las Vegas and RTC identified the following lessons learned: • Attendants or ambassadors helped explain the technology. These employees, located at stops and on board, provided a critical path for disseminating information and answering questions at shuttle stops or on board the vehicle. • Curbside management is key. For preprogrammed shuttles with fixed stop locations to operate smoothly, a strategy needs to be implemented to ensure that people are not double parking or standing illegally in these spots. Brand ambassadors often performed this func- tion. AAA suggested that closer collaboration with local law enforcement would also facilitate keeping pick-up and drop-off areas available to the LSAV. • Heat can have a big impact on ridership and vehicles. Because people do not want to wait on the sidewalk in the heat (or other adverse conditions), ridership may be decreased if there is not a covered shelter. Extreme heat conditions also strain electric vehicle batteries, which may impact the duration of daily operations and require more charging than expected. • Deployment of vehicles in mixed traffic with variable speeds may cause anxiety under 25 mph. AAA noted that some passengers may be uncomfortable riding in a vehicle at 15 mph when surrounding traffic is operating at 30 mph. • Practitioners need to have a communications strategy and partnership plan with local businesses to make a new LSAV pilot a success. Community and stakeholder engagement are key to acceptance given the novelty of the LSAVs. Prior familiarity with the vehicle and the goals of the project are especially key for all stakeholders during any crisis event. • Localities or practitioners need to seek incident reports from the vendor. Stipulate this in the contractual agreement. • Training local law enforcement/first responders on how to respond to LSAV shuttle inci­ dents is necessary. To maintain safe operations during the pilot, it was important that project stakeholders had ongoing conversations with police and fire departments in Las Vegas. • Partnership between public entities is often necessary to get an LSAV shuttle pilot up and running. Both the City and RTC had reasons to explore automated shuttle technology through a public pilot and had complementary skills, regarding local infrastructure and transit operations experience. Through this project, local stakeholders were more coordinated and prepared to plan for a Phase 2 shuttle to be funded through an FTA BUILD grant. • Vendors should share more data with project sponsors. Navya shared technical disengage- ment reports and other logs with Keolis. However, neither the City, RTC, nor the project spon- sor AAA received these reports on disengagements or other incidents. All three noted that this information should be shared subject to protections of Navya proprietary information.

Next: Appendix E - LSAV Mini Case Studies »
Low-Speed Automated Vehicles (LSAVs) in Public Transportation Get This Book
×
MyNAP members save 10% online.
Login or Register to save!
Download Free PDF

Interest in driverless vehicles, including low-speed automated vehicles (LSAVs), continues to expand globally and in the United States.

The TRB Transit Cooperative Research Program's TCRP Research Report 220: Low-Speed Automated Vehicles (LSAVs) in Public Transportation presents current use cases for LSAVs and provides a practitioner guide for planning and implementing LSAV services as a new public transportation service.

  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!