Improving Mapping and Selection of Low-Speed Autonomous Vehicle Shuttle Routes: A Data-Driven Framework

Low-speed automated vehicle (LSAV) shuttle technology has attracted attention for its potential to address first/last-mile links in public transit, improve connectivity in campus/office park environments, reduce human-induced crashes, reduce operating costs, and improve overall urban mobility. LSAV shuttle technology has seen an array of pilot tests and demonstrations around the world ranging from closed test sites to operating in mixed vehicular traffic. These demonstrations continue to provide evidence that LSAV shuttles can safely operate in these environments. The potential for these systems to fill some mobility and connectivity needs is clear and their capability to transport patrons safely has been documented. What has not been discussed in depth is a methodology for systematically and objectively identifying potential route locations using available data. To improve upon these past practices, this paper focuses on objectively identifying LSAV shuttle routes throughout the University of South Florida's campus utilizing existing pedestrian paths, or mixed pedestrian environment, with readily available campus demand and mobility data. The study's objective is to present a data-driven route selection methodology that accounts for the possible operational environment of these vehicles. The nature of LSAV shuttle people movers being able to operate outside of a typical roadway network creates a necessity to consider both trip distribution and route choice set models from the perspective of transit and pedestrian travel. Components of both modes were considered in the methodology and applied to generate routes utilizing readily available campus mobility and population data. The results demonstrate the most optimal routes are oriented on a line with the least deviation from higher density trip production locations (origins) to higher density trip attraction locations (destinations).