Automating Transportation Affordances for People Living with Disabilities Using a Machine Learning Approach

Award ID: 2236277

Community participation and employment is vital to health, well-being, self-determination, and quality of life. A major barrier to community participation and employment for persons with disabilities is the lack of access to transportation. Of the nearly 25% of the adult population in the United States living with a disability, roughly 60% are of prime working age. Thus, these barriers can significantly influence the economy and importantly limit upward socioeconomic mobility. One of the reasons for the lack of access to transportation is that persons with disabilities have different travel patterns compared with the general population. For instance, they tend to make fewer trips, are prone to utilizing slower means of transportation, travel shorter distances and rely more upon public transit. Despite decades of civil rights legislation for Americans with disabilities, access to safe and reliable routes to places of employment are hampered by inaccessible and poor-quality sidewalks, public rights-of-way, bus stops and general connectivity. Improving integration of these systems across rural and urban communities requires reliable data but these data are expensive to create and often unavailable. The lack of data inevitably hampers the ability for advocacy groups and local governments to understand, transparently discuss, and make informed planning decisions that improve transportation, and thus, employment access. This project will create a technique to rapidly generate data essential to building a more integrated transportation system for people living with disabilities. This project builds on the convergence of two distinct and active projects sponsored by the National Science Foundation (#2125087) and the National Institute on Disability, Independent Living and Rehabilitation Research (#90DPCP0004). This proposal aims to: 1) develop a prototype to automate the delineation and affordance assessment of bus stops to integrate with existing efforts on sidewalk quality automation, and 2) explore the role the built environment and first/last mile play on community participation to and from places of employment and households. This work is intended to make contributions to Human Computer Interaction and Computer Vision, particularly at the intersection with urban planning and disabilities studies. We also work to overcome challenges of machine learning bias toward minorities by working directly with persons with disabilities to improve empirical computational models. Technical team members will interact with disability experts and designers on the team to bridge disciplinary gaps that hamper efforts to translate experiential qualitative information (activities of daily living) into computational models that make interpretations and evaluations about the design of the built environment. This project will result in open-source accessibility analysis and visualization tools about first/last mile and the quality of sidewalks, bus stops and roadways. The work from this project can aid in community and state-level transportation master planning and spur additional partnerships with NGOs and governmental organizations. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

NSF Program Director: Alex Vadati