Rapid improvements to public transit could occur if agencies elect to integrate shared, automated, electric vehicle technology in the existing public transportation system. This project develops a replicable, open, deployable model which can (1) identify unmet service need based on transit dependence and low income and minority status and (2) conduct a cost analysis on the feasibility of operating shared automated vehicles and shuttles as a part of a public transit system.
Public transportation systems provide people with access to economic and leisure opportunities. However, current U.S. transportation systems do not provide an equitable level of access for private vehicle owners and transit dependent travelers . Transit-dependent populations in areas with lower population densities are subjected to limited mobility that diminishes access to opportunities. This situation becomes more inequitable when the transit-dependent population disproportionately overlaps with low-income and minority populations. To increase transit access to these areas, transit agencies occasionally incorporate ridesharing or on-demand transit service, but costs are much higher than conventional bus or rail transit per passenger-mile . Automated vehicles and low-speed automated electric shuttles present an opportunity to address transit needs at lower cost and could enable greater access by providing first/last mile services for rural and lower population density areas. The aim of this project is to develop a replicable, open, deployable model which can:
• Identify the census block groups in a metropolitan area that have unmet transit need
• Compile a socio-demographic profile of these census block groups and
• Evaluate the cost-efficiency of automated vehicles and shuttles integrated into a public transit system.
The project will use Allegheny County, Pennsylvania as an initial case study and proof of concept. The project will then model the cost-efficiency of shared autonomous mobility solutions across a larger sampling of major cities in the United States. The project will use Python-GIS code in order to quickly evaluate each city.
To effectively identify areas with unmet transit access needs in the county, a transit coverage score is calculated. First, the transit-dependent population will be derived from data from the 2016 American Community Survey. The determinants for transit service are (1) number of bus and rail stops in each block group, (2) frequency of service for each stop per weekday in each block group, and (3) number of routes in each block group , . The transit service frequency was determined using general transit specification feed (GTSF) processed data from Carnegie Mellon University’s Mobility Data Analytics Center which provided the bus frequency by the hour for each road segment in Allegheny County and the result value was divided by the net acreage then normalized.
Census data on income and ethnicity are used to identify census blocks with high proportions of low-income and/or minority households, these census block groups are then labeled as priorities. Spatial analysis tools will be used to created street routes from the center of priority census block groups to the nearest transit stop. The route distance inputs will be used to derive a range of annual mileage into the cost model for each alternative. Initial work on a cost model has been completed to estimate the cost per mile and cost per passenger trip for one SAV, shuttle or bus providing the service, and this model will be expanded with additional detail and sensitivity analyses. Estimates from research literature and financial reports from transit agencies will be used to estimate the direct costs for each alternative.
 D. J. Forkenbrock and L. A. Schweitzer, “Environmental Justice in Transportation Planning,” Journal of the American Planning Association, vol. 65, no. 1, pp. 96–112, Mar. 1999, doi: 10.1080/01944369908976036.
 D. Gupta, H.-W. Chen, L. A. Miller, and F. Surya, “Improving the efficiency of demand-responsive paratransit services,” Transportation Research Part A: Policy and Practice, vol. 44, no. 4, pp. 201–217, May 2010, doi: 10.1016/j.tra.2010.01.003.
 J. Jiao and M. Dillivan, “Transit Deserts: The Gap between Demand and Supply,” Journal of Public Transportation, vol. 16, no. 3, Sep. 2013, doi: http://dx.doi.org/10.5038/2375-0901.16.3.2.
 S. Mamun and N. Lownes, “Measuring Service Gaps: Accessibility-Based Transit Need Index,” Journal of Planning Education and Research, 2011.
Task 1: Finalize model in Python with Allegheny County (July 2020-September 2020)
Task 2: Evaluate and select 10 U.S. cities to evaluate for cost-efficiency and public transit-automated vehicle services compatibility (September 2020-Januay 2021)
Task 3: Create final report, briefings and journal paper (December 2020-May 2021)
The project analyses support the planning and equitable deployment of shared automated vehicles in public transit agencies. We will meet with stakeholders in transit agencies and make the model freely available online.
Expected Accomplishments and Metrics
The code will be made publicly available for use by other researchers or policymakers. We expect to develop a report and several presentations to transit agency stakeholders over the course of the project, specifically through our involvement in the Pittsburgh Mobility Collective. We will also publish papers with our research findings and detailing the model creation in peer-reviewed journals and conferences. We will work with CMU and Traffic21 to hold a policy briefing for Congressional Staff members in Washington, D.C.
||Carnegie Mellon University
||Faculty - Untenured, Tenure Track
||Carnegie Mellon University
||Student - PhD
Amount of UTC Funds Awarded
Total Project Budget (from all funding sources)
No match sources!
|Transit agencies TBD
||Deployment Partner Deployment Partner