Accessible autonomous transportation and services: a focus group study

Abstract Purpose Existing automated vehicle transportation guidelines and regulations have minimal guidance to address the specific needs of people with disabilities. Accessibility should be at the forefront to increase autonomy and independence for people with disabilities. The purpose of this research is to better understand potential facilitators and barriers to using accessible autonomous transportation. Methods Focus groups were conducted with key stakeholders derived from people with disabilities (n = 5), travel companions/caregivers (n = 5), and transportation experts or designers (n = 11). Results The themes include describing stakeholder perceptions across all three groups by identified themes: autonomous vehicle assistive technology, autonomy vs automation, cost, infrastructure, safety & liability, design challenges, and potential impact. Conclusion Specific gaps and needs were identified regarding barriers and facilitators for transportation accessibility and evidence-based guidance. These specific gaps can help to formulate design criteria for the communication between, the interior and exterior of accessible autonomous vehicles. IMPLICATIONS FOR REHABILITATION Accessible and affordable autonomous transportation may increase mobility and the autonomy of people with disabilities to travel spontaneously. Autonomous vehicles and services should be designed to accommodate various types of disabilities such as multimodal and multilingual device communication. Safety and liability regulatory protocols need to be developed for incidents and emergencies. Wheelchair user, especially people who use powered devices, would need systems for ingress/egress, docking, and occupant restraints.


Introduction
Over 24 million adults in the United States have a travel-limiting disability [1].The autonomy to travel opens opportunities for greater participation in community and family activities, job opportunities, and access to healthcare and education.For example, this could enable new employment opportunities for approximately 2 million people with disabilities (PwDs) and could save $19 billion annually in healthcare-related expenditures from missed medical appointments alone [2].Transportation barriers can be due to a lack of accessible transportation [3], limited knowledge or information on available services [4], or, lack of funding in various payment models [5].In a survey of over 1,000 PwDs, � 60% of respondents (n ¼ 612) rated the importance of technology in meeting their personal mobility needs (e.g., home, work, neighborhood) as "critical," and over 40% (n ¼ 438) felt that traveling freely (e.g., vacation, cruise, airline, bus, taxi, train) was also "critical."Some of these participants also provided additional written comments specifically stating self-driving vehicles [6].PwDs are optimistic regarding the potential for enhanced mobility and independence resulting from the emergence of self-driving vehicles and transportation systems [7].The implementation of this technology has largely been under-investigated, especially through accessibility standards and legislation with the Society of Automotive Engineers (SAE's) levels of automation framework [8].
The SAE J3171 informational report includes universal design principles and guidance for the development of automated driving systems-dedicated vehicles (ADS-DVs).It includes a summary of stakeholder interviews with PwDs, advocates, and others who work with PwDs, experts in design for PwDs, and, governments/policy experts for transportation [9].The interviews highlighted specific issues for individuals with visual impairments and wheeled-mobility users including but not limited to entering/exiting the vehicle, in-vehicle communications, and transfers into/out of the vehicle.However, the report was limited to fleet-operated on-demand shared mobility scenarios and did not consider fixedroute transit services or private vehicle ownership.The informational standard points to the seven principles of universal design emphasizing the importance of involving the PwDs throughout the design process.However, there are no perspectives from caregivers or existing accessible vehicle modifiers to illuminate the current and future potential design challenges to making ADS-DVs accessible to all PwDs.
Previous focus group studies such as Hwang et al. [10], explored the potential of automated vehicles (AVs) for PwDs and public transit agencies.The findings lead to three reflections for policymakers and planners: AV accessibility and safety, promoting relationships amongst stakeholders, and mitigating user anxiety through education and outreach programs.Another focus group study that included 24 PwDs, stated there was a lack of access to distant health care services with existing paratransit services.AVs could potentially mitigate this, but concerns were raised about the capacity of AV services, trip cost, and environmental accessible provisions (access to sidewalks, ramps, and curb cuts).Two AV focus group studies with older adults (55þ) stated similar views but stressed upon emergency protocol systems and the ability to trust the AV to ensure transportation reliability and accuracy [11,12].Three focus group studies [7, 13,14] with people with visual impairments discussed accessible features and concerns pertaining to parking, vehicle location, roadside assistance, and smartphone interface.
In these focus groups, some explored the perspective of a provider as well as PwDs, but none discussed the views of other key stakeholders such as caregivers, other types of providers (designers and modifiers), and other types of disabilities apart from persons with visual impairments.
A recent systematic review of grey and scientific literature was conducted for autonomous transportation and services for PwDs and older adults.This work outlined the current trends and their implications for future travel [15].The results highlighted that each disability group may need accessible and usable features that are tailored to meet each of their specific needs.Another study designed an inclusive mobile app for PwDs to book rideshare AVs which included 20 individual interviews across 12 monthly meetings with PwDs and transportation providers [16].It concluded that providers should take an "accessibility-first approach" and how collaborative research extended their knowledge of PwDs.A focus group with 72 AV experts and "nonexperts" was conducted in Italy, Spain, and Germany where participants stated that they do not have reduced mobility or disability [17].This study recorded also reported concerns about increased maintenance costs, emergency protocols, and vehicle accessibility challenges.There are AVs on the road that have some accessibility features; however, there is little published research that compares current ADAS-DVs on the market for accessibility, and previous focus groups and reviews did not include multiple stakeholder populations.There have been efforts to redesign wheelchair accessibility hardware to be compatible with a vehicle floor full of batteries [18].
Therefore, the purpose of this study was to conduct a user needs assessment with key stakeholders (end-users, transportation companions/caregivers, and, accessible transportation providers) based on the Participatory Action Design and Engineering (PADE) model [19], to elucidate their views and concerns around the accessibility of ADS-DVs.

Participants
The study included a sample size of 5-10 participants per stakeholder group as in total 12-15 participants was deemed adequate to achieve thematic saturation in qualitative research using interviewing techniques [20][21][22].All participants were recruited from internal research registries and disability organizations/support groups.All focus group sessions were conducted in February 2022.This study was approved by the University of Pittsburgh's Institutional Review Board (IRB) (STUDY20090111).
General inclusion criteria for all participants were living in the United States, age 16 years or over (parental consent was required for participants under 18 years old), and being able to communicate and participate in small group dialogue online (or over the phone).
Group 1 (end-users), individuals that currently use one or more forms of accessible/non-accessible transportation (e.g., personal, or public) for travel to destinations in the community (currently defined as using transportation mode(s) to travel at least one time per week) and were living with a disability or were an older adult (65 yrs.þ).A disability was defined according to the definition provided by the Americans with Disability (ADA) Act as "a physical or mental impairment that substantially limits one or more major life activities; a record (or past history) of such an impairment; or being regarded as having a disability" [23].All participants highlighted their disability category as per the American Consumer Survey [24].ADA accommodations were provided for those who wanted to participate in the focus group.
Group 2 (transportation companions/caregivers) were defined as those who currently travel with a person with a disability (currently defined as travel to/from a location in the community with a person with a disability at least one time per week).
Group 3 (providers) were defined as persons currently employed and have at least 1 year of experience in accessible transportation services (e.g., vehicle manufacturers and modifiers, driver rehabilitation specialists, adaptive driving instructors, paratransit, and ride-share operators), providing education to PwDs, providing referrals of PwDs for accessible transportation services (e.g., physicians, vocational rehabilitation counselors, social workers) and/or translation of client needs into accessible transportation design concepts (e.g., architects, engineers, designers, etc.).

Survey construction
An interview guide was developed to maintain consistency in the questions being asked of each key stakeholder group: end-users (Group 1), transportation companions/caregivers (Group 2), and accessible transportation providers (Group 3).The guide was created using feedback from a sample survey using Figure 1 that identifies the current set of pain points for consumers and providers throughout a complete trip or journey mapping process [15].Journey mapping clarifies the decision processes and steps individuals make to use and provide mobility services and helps identify why individual decisions or trips may not be taken, and why specific options and choices, or decisions are made [15].The items were drafted by using a Delphi process with a transdisciplinary team of key stakeholders that included researchers, endusers, clinicians, and accessible transportation providers [25].Each interview guide consisted of the interviewer's introduction, and explanation of the levels of automation [26], followed by a set of questions that are specific for PwDs/older adults (Appendix A), transportation companions/caregivers (Appendix B), and transportation experts/designers (Appendix C) pertaining to the planning and execution of a complete trip, the steps of which are depicted in Figure 1.

Focus groups
Three focus groups (one for each group) were facilitated by trained study team members and conducted and recorded online using Zoom.The purpose of conducting the focus group online was to support the engagement of a broad spectrum of disability etiologies, levels of function and severity, and representation from different geographic regions.Participation was not limited only to individuals who have access to the internet.Call-in phone numbers were provided for those who did not have access to the internet and could connect via a cellular phone or landline.Each group interview began with providing an explanation of the purpose of the study, obtaining informed consent, and completing a sociodemographic survey in Research Electronic Data Capture (REDCap) (Vanderbilt University, Nashville, TN, USA) [27,28].All participants were provided with definitions and explanations of autonomous vehicles, and SAE's level of autonomy (levels 1-5), via an interview guide and visual presentation (Appendices A-C).Each of the recorded focus group sessions were automatically transcribed via zoom's transcription features and were then manually verified by a research team member for accuracy.

Data analysis
Three trained study investigators independently reviewed the transcripts to identify initial themes for each of the focus group sessions using content analysis techniques [29].The reviewers coded the themes using grounded theory [30] and then met to compare and contrast their findings and to reach a consensus.After separately reviewing the transcripts, these themes were discussed and mutually agreed upon by all the reviewers (Table 1).The following remarks for each of the questions for each group (Appendix A, B & C) were aggregated into seven themes.

Results
The following section describes the stakeholder perceptions across all three groups by identified themes: AV assistive technology, autonomy vs automation, cost, infrastructure, safety & liability design challenges, and potential impact.

Sociodemographic Data
A total of 21 stakeholders were recruited for the focus group, including 5 end-users, 5 transportation companions/caregivers, and 11 accessible transportation providers.The accessible transportation provider focus group was split into two sessions (5 and 6 people in each group), and the results were combined.All participants in each of the stakeholder groups discussed their experiences and those of others whom they knew, therefore the results are not separated per disability type.Extensive sociodemographic data are described in Appendix D for all stakeholder groups and are summarized in the following sections.

Group 1: End users
End users included four White/Caucasians and one Native American/Alaskan Native who all owned/rented their house or apartment.Four males were married and employed full-time with a household income over $100,000 but 1 female was engaged/in a committed relationship, unable to work, and had a household income of $25,000 -$49,999.Three participants stated ambulatory difficulty, self-care, and independent living required wheelchair transportation, requiring a paid caregiver or family friend for transportation whereas the remaining 2 declared vision difficulty only.

Group 2: Transportation companions/caregivers
The 5 transportation companions/caregivers that participated were two retired parents and 3 paid caregivers.To transport their care recipients, 4 were licensed drivers of the recipient's private vehicle, whilst 1 was a passenger.The remaining aided with ground public or private transportation.

Group 3: Accessible transportation service providers
There were 7 service providers of accessible transportation or adaptive driving options/technology (e.g., accessible adaptive driver training instructors, clinicians, educators, and project  Lastly, 1 stated that they were in the business services to the mobility industry.Six had over 25 years, 3 between 21-25 years, 1 between 1 -5, and 1 between 6 -10 years of experience in their respective industry.

AV assistive technology
To consider or plan a trip or locate the vehicle, all three stakeholder groups stated, "a smartphone would be fine" -E1, C4, S10, where accessible features can be integrated into "existing rideshare apps" -E2, C5, S2 so that it can provide "audible and visual feedback" -E4, C5, S3.End-users expressed how existing features of a "talking GPS for the blind have come a long way" -E5 such as "to see if the vehicle is ready … .or what roadblocks are or not on the road" -E4.They stated that they do work but need to be expanded as some users get "anxiety of trying to locate [the] ride" -E2, especially if "there are 12 grey cars lined up and you don't know which one is the type of car you are supposed to get into" -E5.To reduce such anxiety, vehicle identifiers have been identified as helpful tools such as "color, [exact] location, size [and] … [door] locator sound or lights, dependent on your disability" -E1.End-users pointed out that they, "like to hear people's voices, as opposed to constantly reading screens" -E1 for accessible feedback so features such as door locators that "beep louder and louder, the closer you get to it, or the light flashes" -E4, will enable the ease of entering/exiting an AV.Also, all three stakeholders stressed the ability for the wheelchair "to go into the [vehicle] on its own and come out" -E2, C5, S2, automatically, but discussed how there should be an automated wheelchair locking mechanism that would "fit in all vehicles" -E1, C5,S2.

Autonomy vs automation
One end-user stated, "I think that AVs would accommodate all my needs … but I would want the full automation with the ability to take over, and that's simply for my own peace of mind" -E2.A caregiver's perspective was that "I have a Tesla with the selfdriving feature, but I have not been willing to let it do it yet … but I want to make sure it is safe" -C4.Lastly, one provider queried, "As the levels of automation increase do the need to concentrate on the road decrease?"-S2.

Cost
The notion of ownership of private vehicles for end-users appealed to most participants; however, end-users stated the cost of level 2-4 vehicles "still need my $40,000 worth of adaptive equipment to drive it" -E1 and were "concerned about the cost of a level 5 AV" -E1 .Regarding public transportation, the "fees for Ubers" -E4 were a primary concern, but one participant stated, "there are a lot of things that [are] available that would make my life easier, but insurance doesn't pay for it … if you are on disability, you can't afford it … cost is a factor" -E2.One provider described a client, stating that "sometimes it is cheaper to hire a professional driver, instead of paying for the adaptive equipment" -S7.

Infrastructure
End-users unanimously described the need to "change existing paratransit infrastructure" to allow "for spontaneous travel … otherwise there will be no difference."But "existing rideshare apps have tracking, and route navigation systems set up really well".Some caregivers pointed out deep concerns about existing private taxi companies, as "there aren't any Uber cars for handicapped people in wheelchairs to use ramps" -C4.Furthermore, we still do not have "the infrastructure in terms of powering the vehicles … and environmental accessibility of the roads" -C2.Providers (specifically those working in paratransit) questioned how to replace the "door-door service" -S6 as many still need human interaction.They were especially concerned about " … getting these AVs into [remote] areas with not many people with disabilities" -S7.

Safety & liability
In this section, safety encompasses the potential accidents and protocols, reliability, and subsequent liability concerns.

Safety
End-users expressed how level 2 AVs (such as some that are already on the market) can help "decrease the risk of fender benders," and level 5 AVs can "help to get to destinations [safely] in emergencies especially those with reduced capacity and mobility" -E4.However, participants stated that levels 2-4 AVs are not meant "to mitigate driver fatigue" -E1 and overall have preconceived notions of the reliability of AVs due to accidents seen on media outlets.Caregivers expressed the potential for their care recipient "to be more independent" -C4 without their caregiver.However, their concerns included: "what would happen if he had a medical emergency" -C3, "what happens if the AV goes to the wrong address" -C1 and what happens if "there was an accident" -C5.Providers stated similar concerns where level 2 vehicles could reduce the risk of accidents but added that level 5 AVs "will only work if users can independently enter/exit and travel to the vehicle" -S7."For level 2 AVs if a wheelchair user has a cervical spine injury and not able to turn their head as far, those lanes assist features are useful" -S1.Some providers questioned whether those with "cognitive impairments have the ability to operate the vehicle" -S2, what would happen if "the vehicle catches on fire and how does the person in a wheelchair independently get out of the vehicle" -S3.The vehicle modifiers stressed that "more power is being drawn than ever before for electric AVs and those are becoming very dangerous vehicles to work on if you're not properly trained" -S2.
Reliability was a reoccurring theme.Specifically, end-users questioned whether "AVs can reliably navigate through large crowds to the user itself" -E4, caregivers stated how "reliability is the most important factor" -C1 and lastly providers were concerned more about infrastructure surrounding AVs but questioned the "accountability for uncontrollable variables" -S2.

Liability
Providers were asked to discuss their thoughts about ethics, liability, or ownership in the case of confidentiality (private information), accidents, or AV malfunction.Overall, vehicle modifiers and designers discussed the challenges with data protection and liability determination.Vehicle modifiers stated that there are a few concerned parties for liability determination: "the user, the modifier, vehicle manufacturer and any 3 rd party (i.e., another vehicle/person)" -S2.For example, "users need to get the correct vehicle adaptive equipment for their wheelchairs" -S9 but some users avoid provider recommendations due to cost and "don't take care of their equipment" -S1.Liability determination for "[AV] levels 3-4 will be harder … level 5 AVs reduces the number of parties involved and omits liability for the modifier" -S3 -level 5 AVs would theoretically not require a human driver therefore adaptive driving controls would not need to be installed.However, there was a general acceptance by the entire group after one vehicle modifier stated that "we must accept the wave of liability issues as with any new technology until a balance is found" -S5.

Design challenges
The providers stated three areas that could influence the design challenges: economical influence, vehicle design, and comfort.The economic influence was thought to be due to the disability population being a "small market size" -S1.This would mean that it increases the cost of producing an accessible vehicle and "there are too many variations of adaptive equipment" -S1, especially when trying to "accommodate underpopulated regions" -S1.Minivans are the most popular vehicle type when looking to adapt a vehicle There are also "more SUVs on the market than minivans" -S4, so this could lead to lower production of minivans in the consumer market thus increasing the purchase price of said vehicle type.
The main vehicle design challenge is "battery relocation … to accommodate wheelchair users" -S1.A design challenge is that batteries are heavy and the common displacement across the vehicle is in the base to distribute the weight and create a lower center of gravity.While this creates a challenge for wheelchair accessible modifications [31,32], it has been demonstrated that electric vehicles with batteries located in the floor of the chassis can be modified to improve wheelchair accessibility [33].
A "universal design platform" -S3 was stressed by vehicle modifiers especially for "interior dimensions … .to accommodate wheelchair users, like Ford has a universal design platform" -S2.Creating an automated wheelchair ingress/egress and docking system was primarily discussed as an accessibility feature but a core challenge highlighted by the providers is "vehicle cleanliness … how does it get cleaned out … if somebody leaves garbage in a vehicle" -S10.This could potentially comprise the AV automated wheelchair securement and automated ingress/egress method.
Comfort was considered a primary issue and "huge design challenge" -S4.An adapted vehicle driving instructor elaborated on her experience with her clients in newly adapted vehicles.The instructor stated, "to me, comfort isn't just about … the person riding in a wheelchair.It's not comfortable in a wheelchair-accessible vehicle … [because] the center cut channels, like [a] rear entry [vehicles], [where] they're sitting at a tilt.And they're much lower to the ground and it's really uncomfortable [because] people are being bounced around.Personal vehicle[s] [are] not a super comfortable ride" -S4 .
The instructor then expanded further regarding public transportation: "For mass transit, it's the same issue where they're having … multiple people in there.Can they see outside?They [are] sitting low underneath the islands … and now they're getting nauseous while they're being transported" -S4.Lastly, the specialist concluded with the clinical implications for end-users: "What's the temperature going to be … do I have control?Is it too hot too cold and then the other thing is just the sound, I think, the rattling of ramps … that's still an issue for some people, so … comfort as it's just [about] the frame [of the vehicle]?"-S4.

Potential impact
Apart from being apprehensive about trying AVs, end-users stated the "convenience" it would provide as it "allows for spontaneous travel" -E5.One family caregiver described her current experience, "I had to quit my job to be able to drive her in the car four hours a day to the daycare center … it would change my life tremendously to have an AV because I would have the ability to do regular things like everybody else" -C1.The providers indirectly stated three points where it could provide potential impact, "accessibility for all … by increasing autonomy," "alleviates driver shortage issues," and "opens the possibility for tele-operating multiple vehicles" -S10.

Discussion
This may be the first study to gather input from multiple stakeholder groups into the potential barriers and facilitators concerning AV use amongst PwDs.Each stakeholder group was comprised of a diverse mix of PwDs and experiences allowing for greater insight into creating a design criterion for creating accommodations for various disability groups.This detailed account of potential barriers and considerations for accessible autonomous transportation could provide a foundation for AV design as well as legislation and standards development.
Vehicle accessibility modifications challenges, user comfort, indepth liability determination, and, caregiver potential impact are topics not found in the SAE J3171 standard [9] and other focus groups [7, 10,13,14].
Although caregivers in this focus group reiterated comments from people with physical disabilities, they vocalized the challenges that people with intellectual or cognitive disabilities would encounter.A universal agreement amongst the stakeholders was stated that there should be an attempt to "accommodate all" types of disabilities when developing accessible autonomous transportation.Aside from the accessible AV design elements, infrastructure was a key area highlighted regarding public and private transportation.This includes public transportation schedules (either paratransit or bus schedules) to "allow for spontaneous travel."End-users stressed that current public transportation/paratransit schedules do not always allow spontaneous travel.Optimistically, the end-user focus group agreed upon this statement, "Personally I quite look forward to the independence and the opportunity to be a little bit more spontaneous in either local trips or long journey" -E5.Furthermore, participants supplemented this by remarks about existing private vehicle transportation and how future AVs and services would need to factor in wheelchair-accessible vehicle availability, service reliability, and cost.
Key stakeholder collaboration is essential as policymakers will need to work with AV and EV manufacturers, accessible vehicle modifiers, clinicians, and end-users to create universal design platforms and standards.Through the lens of stages of a journey, the results form design criteria for accessible AV communication, interior, and exterior design.

Accessible AV -user communication
End-users and providers recommended expanding on existing rideshare apps used by existing private transportation companies such as Uber/Lyft.Currently, existing rideshare apps utilize GPS vehicle-smartphone tracking; some are compatible with voice assistants (e.g., Siri) and have wheelchair-accessible vehicles in a few areas across the US.
Accessible AV features such as communication with the AV/ATS (e.g., request for human assistance or in-vehicle/trip status indicators) and a user profile can be established by expanding existing rideshare apps.For example, existing 3 rd party voice assistant software (such as Amazon Echo Auto) can be integrated into the vehicle interface to aid with human-AV communication.Additionally, a personalized profile on existing rideshare apps that details the accessibility requirements for each user can be stated so that the correct vehicle type arrives at the user.These requirements could simply state whether a person requires a wheelchairaccessible vehicle or specific sensory accessibility features to aid those who have hearing or visual impairments.
Previous research on ride-hailing apps found that people who used white canes and crutches used rideshare services more than people who used other mobility aids such as wheelchairs or service animals [34].This was due to anecdotal reports of driver discrimination against these users and most vehicles are not accessible to powered wheelchairs or scooters.AVs could help to reduce discrimination if the algorithms and artificial intelligence (AI) were not biased and expanded accessibility to powered mobility device users.As stated in previous articles [8,10,34], policymakers should address making app-based ride-hailing services more broadly accessible.There have been attempts to address these accessibility challenges [16], but they did not account for transportation affordability, smartphone app accessibility, algorithm, or AI bias.End-users and providers have stated that accessible transportation features must be available in different formats (e.g., audio/visual) and compatible with voice assistant software on smartphones, which could assist with considering and planning the trip or locating the vehicle.These results are supported by the focus group study in [12].

Accessible AV interior
Internal to the vehicle itself, the vehicle should have a similar smartphone/device setup compatible with audio/visual/tactile as well as being multi-modal and multi-lingual.Wheelchair and service animal accessibility were among the greatest concerns raised by the end-user participants and providers.A universal automated wheelchair loading, unloading, vehicle docking, and securement system was noted to be needed for wheelchair users, especially for bariatric wheelchair users.It was also noted that because wheelchairs have increased in size and weight over the years, AV designers will be challenged by the constraints of electric AVs.As there is no internal combustion engine for electric AVs housed in the front of the vehicle, the batteries that will power the vehicle are built into the base [35].Physical modifications for accessibility will involve ramp installation and lowering the vehicle floor height to accommodate electric-powered wheelchairs [36].Therefore, its batteries will have to be relocated and affect the internal vehicle weight distribution as well as accommodating additional personal mobility devices such as electric-powered wheelchairs.

Accessible AV exterior
A universal design platform for accessible AVs, wheelchair on-off vehicle boarding, and in-vehicle communication systems are examples of technologies that were recommended by all interviewed stakeholders.Accessible door openers and transfer handles need to be considered so they are easy to use, especially for those with cognitive and/or physical impairments.

Safety and liability
Safety and accident liability were concerns that all interviewees expressed, questioning what the possible accident protocol might be for an accessible AV.The primary concerns were escaping the vehicle without human intervention, methods of requesting human assistance, and alternative accessible transportation in the event of an emergency.The protocols for safety and liability have not been addressed in ISO standards and regulatory bodies, and there needs to be additional attention given to accommodate those with disabilities where they would normally require caregiver assistance.On the other hand, this has been identified across a AV providers only focus group in Europe, and another with just older adults [11,17].

Limitations
Although detailed explanations were given to all stakeholder group participants for the definitions of autonomy and AVs, there were a lot of assumptions by accessible transportation providers about the limitations of the potentials AV automation levels.The study did not document the participants' direct experience with the various AVs of different automation levels.This could lead to a study to identify and interview people who have tried AVs and different services and then compare the viewpoints.The interviewer was able to provide clarifications to the AV automation levels to keep the conversation on track without hindering discussion.However, it diverted the focus from the accessibility aspect of AVs towards the performance of AV automation level.
The focus group itself did not include every type of disability, however, an attempt to mitigate this was to include 3 different types of stakeholder groups.The results of this study may have greater impact by including a larger more representative sample of people with disabilities.Therefore, future work should increase the sample size of people with disabilities to include a wider range of conditions (e.g., intellectual disability).Although both people with ambulatory and visual difficulties were included in the focus group, most of them were Caucasian.Their transportation experience and needs might be different from other groups.Future studies with a more diverse sample will be considered to further explore and compare their perspectives regarding AVs.
Providers could have been limiting their feedback to avoid violating any non-disclosure agreements (NDAs) with the respective employer, but the questions were structured (Appendix C) to mitigate this concern.

Conclusion
Our analysis highlights the importance of stakeholder communication to better design and develops accessible AVs, and to develop better policy and legislative guidance.This focus group has found that infrastructure changes to public transportation, private transportation, and appropriate protocols for emergencies need to be immediately addressed.Future work will involve conducting a national survey amongst end-users, transportation caregivers/companions, and accessible transportation providers to explore viewpoints from larger broader demographic cohorts.Future work will involve creating and conducting a national survey amongst end-users, transportation caregivers/companions, and accessible transportation providers to explore viewpoints from larger broader demographic cohorts.

Figure 1 .
Figure 1.Conceptual model of a complete trip of the journey without accident or system failure.

Table 1 .
Focus group theme coding Table.
CostCost of AVs, equipment, and their respective services Cost, expensive, 3 rd party modifications, variety of equipment Infrastructure The infrastructure required for accessible AVs Spontaneous travel, existing travel, remote regions Safety & Liability Multi-stakeholder perspective on safety and liability concerns for accessible AVs Emergencies, capacity, fatigue, reliability, risk, passenger safety Design Challenges Accessible AV Design Challenges from the perspective of transportation providers Universal designs, weight, standards, secure Potential Impact Multistakeholder potential impact Convenience, and spontaneous travel, open the possibility managers who have the expertise or training/knowledge in this area).Three were designers, developers, modifiers, or manufacturers of accessible or adaptive driving equipment or vehicles.