The Impacts of Vehicle Automation on the Public Transportation Workforce (2022)

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.

13   C H A P T E R 2 The background information presented in this chapter will help readers better understand some of the potential benefits of transit vehicle automation, what factors and forces should be considered regarding automation timelines and anticipated impacts, and what technologies support transit vehicle automation. The sections in this chapter are: • Levels of vehicle automation. • Baseline supporting technologies for vehicle automation. • Supporting technologies for transit vehicle automation. • Potential impacts of automating transit services. • Challenges facing transit service automation. • Handling uncertainty. 2.1 Levels of Vehicle Automation SAE classifies vehicle automation in six incremental levels of autonomy from 0 to 5 in its J3016™ standard (SAE International 2021). Table 7 shows each automation level and its defini- tion. This research examined the transit automation use cases at L4 or L5 automation in which greater workforce effects would be expected. Significant workforce effects for lower levels of automation are unlikely due to the continued need of a professional driver on board the vehicle (USDOT 2021). In most current applications of L4 and L5 vehicle automation, human drivers continue to be present; however, in both levels of automation, human drivers are unnecessary, assuming an L4 vehicle stays in its operational design domain (ODD). Vehicles at these two levels of auto- mation are often referred to as highly automated vehicles (HAVs). 2.2 Baseline Supporting Technologies for Vehicle Automation HAVs need a significant amount of supporting technology and systems, and these systems have workforce implications. Automation-supporting systems will require qualified special- ists and technicians for system monitoring, troubleshooting, maintenance, replacement, and modification—potentially increasing the number of needed qualified technicians and adding to technical training and certification programs. This subsection of this chapter provides a cursory overview of the technology supporting vehicle automation to provide the reader with a basic understanding of the systems required and the levels of expertise that may be needed from tran- sit vehicle maintainers and technicians. This subsection is not meant to be an exhaustive review of automation-supporting technologies or the current state of those technologies. Transit Vehicle Automation Overview

14 The Impacts of Vehicle Automation on the Public Transportation Workforce Although there are some exceptions (e.g., automated xed-guideway people movers), a base- line technology architecture supports vehicle automation. is architecture is comprised of a multitude of environmental perception sensors [i.e., light detection and ranging (LIDAR), camera, and radio detection and ranging (RADAR)], localization sensors [i.e., global position- ing system (GPS) and inertial measurement unit], actuators for vehicle automation, and an electronic control unit for control and decision-making. ese technologies are all connected in a network setting, as shown in Figure 3 (Gelbal et al. 2017). As the environmental percep- tion sensors gather data, the automation system’s built-in intelligence detects and tracks objects (including obstacles, lane markings, and trac control devices) and responds by actuating lateral or longitudinal control. e architecture may also include modems for vehicle-to-vehicle (V2V) Level of Automation Definition Level 0: No Automation There is zero autonomy. The driver performs all driving tasks. Level 1: Driver Assistance The vehicle is controlled by the driver, but some driving assist features may be included in the vehicle design. Level 2: Partial Driving Automation The vehicle has combined automated functions, like acceleration and steering, but the driver must remain engaged with the driving task and monitor the environment at all times. Level 3: Conditional Driving Automation The driver is a necessity but is not required to monitor the environment. The driver must be ready to take control of the vehicle at all times with notice. Level 4: High Driving Automation The vehicle is capable of performing all driving functions under certain conditions. The driver may have the option to control the vehicle. Level 5: Full Driving Automation The vehicle is capable of performing all driving functions under all conditions. The driver may have the option to control the vehicle. Source: SAE International (2021). Table 7. SAE automation levels and denitions. Source: Gelbal et al. (2017). Note: IMU = inertial measurement unit, RTK = real time kinematics, CAN = controller area network. Figure 3. Computing, sensing, communication, and actuation architecture.

Transit Vehicle Automation Overview 15   and vehicle-to-infrastructure (V2I) connectivity. HAV technology is still developing, and pilots and demonstrations will help manufacturers refine how the vehicles deal with challenges like the weather. (For example, snow conditions reduce the accuracy of LIDAR.) 2.3 Supporting Technologies for Transit Vehicle Automation In transit vehicle automation, there are additional supporting technologies needed. In some cases, automation technologies available in light- or heavy-duty vehicles may not translate auto- matically to transit vehicles. For example, although advances in technology are occurring at a rapid pace, there are still technological advances necessary to support the automation of a full-size transit bus—most notably, advances in electronic control of bus braking and steering systems (Nasser et al. 2018). Additionally, for transit vehicles to be fully automated (i.e., not requiring a human onboard), several other bus driver tasks must also be automated. For example, automated transit vehi- cles will need interfaces to communicate with transit agency systems and service stakeholders (including passengers), exterior sensing to detect passengers waiting at a stop, interior sensing to detect the presence of passengers and the conditions inside the vehicle, interfaces to facilitate human interaction (e.g., passengers requesting a stop or needing assistance), and systems to allow for fare payment (FTA 2018). Transit services for people with disabilities [e.g., Americans with Disabilities Act (ADA) paratransit service] have many additional requirements in terms of both vehicle design and human-machine interfaces to ensure that people with disabilities will be able to use automated transit services. In fact, there is some uncertainty as to whether addressing these additional requirements (e.g., providing automated wheelchair securement onboard the vehicle) is even feasible without a human operator or vehicle attendant. HAV accessibility is an important area of research, with many advances still needed to ensure HAVs (transit or otherwise) are accessible to people with disabilities. For example, the U.S. Access Board held a series of virtual forums on the topic of inclusive design of autonomous vehicles and published the results of these forums in a summary report (U.S. Access Board 2021). The additional supporting technologies and systems required for transit vehicle automation add more uncertainty to the timeline for automated transit service implementation. However, assuming that transit vehicle automation does move ahead, it is also likely that these automation- supporting technologies and systems will result in increased workloads and training and certification requirements for affected bus mechanics and technicians. 2.4 Potential Impacts of Automating Transit Services Automating transit services continues to garner significant interest and discussion in the tran- sit industry, and HAV applications in transit are being intensely advanced by the technology and vehicle manufacturer industries—particularly in the area of low-speed automated shuttles—but ultimately, why is transit service automation receiving this attention? Policy makers and transit agency decision-makers hope that automated transit services will lead to improved operations, enhanced safety, and more efficient cost structures (USDOT 2021). On the other hand, many transit stakeholders, including transit unions and front-line transit employees, have significant concerns about the potentially negative workforce effects that might come with automating transit service. As of the writing of this report, these potentially positive service impacts and the potential workforce effects remain highly uncertain.

16 The Impacts of Vehicle Automation on the Public Transportation Workforce 2.4.1 Potentially Positive Service Impacts In terms of safety, by shifting responsibility for driving from humans to technology, HAVs could reduce opportunities for behavioral errors blamed in most road crashes. Although infre- quent, transit vehicle crashes can be a significant source of risk and cost for transit agencies (Lutin et al. 2016). Therefore, if transit vehicle automation reduces the risk of crashes and colli- sions, then transit agencies, their passengers, and other users in the right of way receive benefits. However, the safety improvements of HAVs are still highly uncertain. HAV developers are pursuing different strategies and technologies—and making different claims, in different ways, about the safety of their systems. Also, not all pilots or demonstrations reveal their safety out- comes. These challenges make it hard to compare vehicle safety across companies and evaluate the safety of HAVs overall. Due to the unique requirements for automated transit vehicles, new tests may need to be established to certify that HAVs are safe for transit agency use (Hughes- Cromwick and Dickens 2019). In fact, FTA issued a report that provides guidance for determin- ing the requirements for automated transit bus testing facilities (Brewer et al. 2019). Business interests in HAVs (from transit agencies, auto manufacturers, and technology firms) are driven by assumptions or expectations that HAVs would be significantly cheaper to operate than conventionally driven vehicles (CDVs). Although uncertain, cost savings, and therefore improvements in transit service cost efficiency, could provide benefits to passengers and com- munities. Transit agencies could potentially continue providing higher levels of coverage-type service with HAVs that would not be cost-effective using CDVs (USDOT 2021). Transit agencies could also improve service levels (e.g., increased frequencies) on high-ridership corridors without as significant an operating cost increase as increasing service with CDVs. However, there is conflicting evidence as to whether cost savings are actually achievable. For example, contradicting much of the early research on the topic, recent research has found that it may be difficult for HAV technology to gain a price advantage over CDVs, depending on opera- tional modes and types of services (Bösch et al. 2018; Nunes and Hernandez 2020). Much of the uncertainty on anticipated cost savings is due to the fact that studies use different operational models (e.g., mass transit, taxi fleets, private cars), vehicle types (e.g., cars, vans, minibuses, buses), and cost categories (e.g., operating expenses, purchase price, depreciation, maintenance, tires, fuel, cleaning) in analyses. The main driver of the anticipated cost savings from automating transit services comes from the reduction in bus operator positions or in the reduction of bus operator wages if bus operator positions become lower-skilled, passenger-service-oriented positions (e.g., an onboard atten- dant). As previously mentioned, bus operator jobs include many non-driving tasks, which are not necessarily automated by automating the driving task (e.g., passenger assistance and fare collection). The USDOT explains the situation clearly: At high levels of automation, there are many reasons why a human presence may continue to be needed onboard, including operators’ non-driving duties. Transit agencies may elect to create a new job category of non-driving onboard attendant to handle customer assistance and other tasks. In that situation, out- right displacement would be limited, particularly in cases where these positions are offered to existing operators. However, given the less-specialized skills required for such positions, the pool of potential candidates would be larger, and thus wages would likely be lower. (USDOT 2021, 58) Other forms of cost savings could be found without necessarily relying on a reduced number of operators. For example, automated buses could be built bidirectionally, eliminating the need for end-of-route turnaround loops (USDOT 2021). The reduced number of crashes could reduce the cost of repairs and insurance premiums. The improved working conditions for operators (e.g., being able to move about the vehicle, not having to attend to passengers and drive at the same time) could reduce operator absenteeism and turnover, etc. However, these potential cost savings are hypothetical, and whether or not they materialize will become known as automated transit vehicles enter fully into service.

Transit Vehicle Automation Overview 17   2.4.2 Potential Workforce Effects Transit service automation has the potential to bring about wide and complex impacts on the public transit workforce. However, what those impacts are is very difficult to predict. As the USDOT notes: The automation of tasks affects the structure of work and the demand for human labor, but the actual trajectory of employment is complex and difficult to predict. Considerable uncertainty exists when attempt- ing to predict the impacts of driving automation on employment, particularly given the multi-decade timescale involved and the current state of deployment. (USDOT 2021, 16) Although more nuanced than can be fully dealt with in this brief discussion, there are con- cerns that automating transit services would cause some employees to lose their jobs or expe- rience wage reductions. Intuitively, transit operators are the largest group of employees who would face the most significant risk of job disruption and change. However, other parts of the transit workforce would likely also be affected by transit service automation. Transit vehicle automation and use in automated transit services could result in new or modified transit jobs, for example, jobs to maintain and operate HAVs (Kalra 2017; Pettigrew et al. 2018) and jobs to program, troubleshoot, and continuously advance HAV technology. In most cases, these new or modified jobs are likely to require a higher level of skills and job training (Pettigrew et al. 2018; USDOT 2021) and may also have higher wages (USDOT 2021). Members of the transit workforce who are in jobs where automation is expected to reduce working hours may need to be retrained for new or modified jobs, may choose to voluntarily leave their employment, or may even experience involuntary separations. Any effect on the transit workforce—whether changes in jobs, wages, or reductions in force— would be subject to requirements under 49 U.S.C. §5333(b), also known as Section 13(c) of the Federal Transit Act (Cregger et al. 2018; USDOT 2021). Although somewhat complex, Section 13(c) requires that transit agencies enact “fair and equitable” protections for transit employees affected by federal assistance (Woodman et al. 1995). Already often strained, labor and management relations could be further stressed during implementations of automated transit services. Even if the leadership of transit unions and transit agencies partner to maximize the benefits and minimize the negative impacts of transit vehicle automation, the transit employees affected by transit automation may feel significant discomfort and uncertainty with the potential changes that automated transit vehicles may bring to their day-to-day jobs. The workforce effects of transit vehicle automation can be divided into the hierarchy shown in Figure 4 based on effect type and outcome. Job gains or losses are changes in the number of jobs required to support the automated transit services that may replace conventionally driven Outcomes Effect Types Overall Impacts Workforce Effects Job Gains or Losses Job Gains Job Losses Job Description Changes Wages Tasks KSAs Figure 4. Workforce effect hierarchy.

18 The Impacts of Vehicle Automation on the Public Transportation Workforce transit services. Job description changes are changes in the responsibilities, tasks, KSAs, and compensation for positions that now support automated transit services. This hierarchy of workforce effects will be used throughout this report, and the outputs of the workforce effect calculator focus on the number of jobs affected by automated transit services, including job gains or losses and job description changes (i.e., changes in wages, tasks, or KSAs). 2.5 Challenges Facing Transit Service Automation Although there is interest in automating transit vehicles of different types, and there is federal support of transit automation research activities, there is still a significant degree of uncertainty facing transit vehicle automation (FTA 2018). The potential challenges may include: • Public acceptance. • Transit agency acceptance. • Labor impacts. • Capital investment. • Research and technology availability. • Safety and security. • Standards and regulations for AVs. These challenges should not be taken lightly, and each challenge may delay or significantly impact the deployment of automated transit vehicles. A recent assessment of the market for automated transit buses found that the relatively small size of the transit bus market and the high levels of vehicle customization at different transit agencies were significant barriers to the development of automation technologies that apply more directly to transit vehicles (Cregger et al. 2019). Other research found that, for full automation of bus services, the business case (i.e., whether investing in automated buses would be cost effective for transit agencies) was fraught with uncertainty associated with whether unstaffed scenarios were feasible, or even likely, given bus operators’ critical roles in customer assistance and fare collection (Peirce et al. 2019). How- ever, there is still an expectation that the costs of automation technology will decrease over time as functionality improves. Also, the business case may be stronger in certain transit automation use cases than in others, and the safety and operational benefits may be significant enough to still motivate transit agencies to automate transit services. Requirements for FTA grant recipients, including limits on the federal share, limits on fleet spare ratios, and standards for asset management and useful life, will also impact transit agencies as they look to explore automated transit vehicles while minimizing agency risk and maintaining compliance with FTA rules and regulations. Automated buses will need to be fully tested and certified—using both traditional means such as Altoona testing (see https://www.transit.dot.gov/ research-innovation/bus-testing) and new testing facilities or processes to certify that auto- mated systems meet yet-to-be-finalized testing criteria (e.g., see Brewer et al. 2019). As previously mentioned, there are specific legal protections for transit labor in Section 13(c) of the Federal Transit Act and potentially in state law and/or collective bargaining agreements, which could create legal complexities for agencies seeking to achieve labor cost savings through automation and the concomitant elimination or restructuring of transit operator jobs (Cregger et al. 2018). 2.6 Conclusion: Handling Uncertainty Despite the challenges and uncertainty facing transit service automation, this study ultimately assumed that transit service automation will eventually be adopted by many transit agencies and that the current transit workforce will be affected by the adoption of automation.

Transit Vehicle Automation Overview 19   2.6.1 Transit Agency Types and Modes The research team realized that use cases might have varying effects on transit jobs at different agency types and sizes. Although not a perfect way to differentiate transit agencies, the research team used the categories of rural/tribal, small urban, and large urban as agency types: • Rural/tribal transit agencies (referred to simply as rural in this report) are those that serve a predominately rural area and are listed as rural or tribal reporters in the NTD. • Small urban transit agencies are listed as urban reporters in the NTD and serve urbanized areas that are less than 200,000 in population. • Large urban transit agencies are listed as urban reporters in the NTD and serve urbanized areas that are 200,000 or more in population. To differentiate what services would be impacted, the research team used the NTD’s official modes that are associated with bus (non-rail) service. The NTD modes (and their official NTD abbreviations) are listed as follows (NTD 2020): • Motor bus (MB): A transit mode comprised of rubber-tired passenger vehicles operating on fixed routes and schedules over roadways. Vehicles are powered by a motor and fuel or by electricity stored onboard the vehicle. Also includes any route-deviated or point-deviated service. • Trolley bus (TB): A fixed-route service that uses manually steered, rubber-tired passenger vehicles powered by an electric current from overhead wires using trolley poles. • Bus rapid transit (RB): A fixed-route bus system that (a) operates over 50 percent of its route in a separate right of way dedicated for transit use during peak periods; (b) has defined sta- tions that are accessible for persons with disabilities, offer shelter from the weather, and provide information on schedules and routes; (c) uses active signal priority in a separated guideway and either queue-jump lanes or active signal priority in a non-separated guideway; (d) offers short headway, bidirectional service for at least a 14-hour span on weekdays and a 10-hour span on weekends; and (e) applies a separate and consistent brand identity to stations and vehicles. • Commuter bus (CB): A local, fixed-route bus transportation that primarily connects outlying areas with a central city. It has multiple stops in outlying areas, limited stops in the central city, and at least 5 miles of closed-door service. • Demand-response (DR): A transit mode operating on roadways in response to requests from passengers or their agents to the transit operator, who groups rides together when possible and dispatches a vehicle to provide the rides. Vehicles do not operate over a fixed route or on a fixed schedule unless temporarily satisfying a special transit need. 2.6.2 Replacing Current Services In reality, automated transit services may replace some current conventionally driven transit services, support expansion of transit service coverage, and even create completely new routes or higher levels of service in already-served areas. Each of these uses of automated transit services will have potentially different effects on the transit workforce. To correctly size the analysis in this study and to focus the results on one particular aspect of workforce effects, the research team decided to focus on the workforce effects associated with replacing current convention- ally driven transit services with automated transit services. This analytical approach focuses the results onto the potential impacts on current transit employees and avoids adding complexity that is created by hypothesizing about unknown amounts of potential service increases. However, because the extent of adoption and the details of automated transit operations are so uncertain, the research team and panel decided to explore four scenarios: two different adop- tion scenarios and two different operational scenarios.

20 The Impacts of Vehicle Automation on the Public Transportation Workforce 2.6.3 Automated Transit Service Adoption Scenarios The research team explored two adoption scenarios: • The partial adoption scenario assumes that implementation of different automated transit use cases at different transit agency types (i.e., rural, small urban, and large urban) will be limited. Partial adoption is somewhere between no adoption and full adoption, and the assumed lev- els of adoption used in this study were based on a combination of industry feedback, panel feedback, and professional judgment. • The full adoption scenario is when all transit agencies implement an automated transit use case to fully replace conventionally driven transit services (assuming they have convention- ally driven transit services to which the automated transit service could be applied). The full adoption scenario provides an estimate of the maximum potential workforce effects of the transit automation use cases. 2.6.4 Automated Transit Service Operational Models Whether or not most automated transit services will ultimately have a human on board every vehicle all the time is still uncertain. There are many valid reasons to keep an operator on board, and there are also valid reasons to imagine a future where humans are not on board every vehicle (assuming vehicles are safe and non-driving operator tasks can be automated or remotely per- formed). The discussion of the pros and cons of both approaches was beyond the scope of this report. In fact, there are many potential variations possible in the degree to which humans would be present on or engaged with automated transit vehicles, and there are many possible ways that transit agencies could delegate non-driving tasks to other transit agency staff (e.g., street supervisors and maintenance staff) (USDOT 2021). Because of this high degree of uncertainty, the research team and panel decided to develop two operational models for automated transit services: • The in-person operations model assumed that there is a human on board every automated transit vehicle, and the onboard person is not dedicated to driving the vehicle. Because the person on board is no longer operating the vehicle, the research team referred to this position as an attendant. • The remote operations model assumed that humans must remotely monitor (and potentially control) automated transit vehicles and that humans would monitor multiple vehicles at the same time. In the remote operations model, a person referred to as a remote operator would be housed in a control-center-like location with high-speed wireless connectivity with each HAV to allow real-time video and audio transmission, communication, and control of auto- mated transit vehicles. More details about both adoption scenarios and operational models are provided in Chapter 6.