Shared autonomous vehicles: Model formulation, sub-problem definitions, implementation details, and anticipated impacts

• Main Author: Fagnant, Daniel J.
• Format: Conference Proceeding
• Language: English
• Subjects: Analytical models; Investment; Mobile robots; Simulation; Vehicle dynamics; Vehicles
• ISSN: 0743-1619; 2378-5861
• DOI: 10.1109/ACC.2015.7171124

The emergence of self-driving vehicles holds great promise for the future of transportation. While it will still be a number of years before fully self-driving vehicles can safely and legally drive unoccupied on U.S. streets, once this is possible, a new transportation mode for personal travel looks set to arrive. This new mode is the shared autonomous vehicle (SAV), combining features of short-term on-demand rentals with self-driving capabilities. This presentation seeks to demonstrate how SAVs' potential may be assessed through agent-based modeling, as applied in Austin, TX. The framework sheds SAVs' current speed limitations established in early pilot SAV demonstrations by CityMobil2 and Google. A 12-mile by 24-mile regional geofence is employed to limit service within Austin to the areas with the greatest demand intensity. The simulation uses a sample of trips from the region's planning model to generate demand across traffic analysis zones and a 32,272-link network. Trips call on the vehicles in 5-minute departure time windows, with link-level travel times varying by hour of day based on MATSim's dynamic traffic assignment simulation software. A sizable degree of market share is assumed, though not market dominance, with adoption levels ranging from 2.3-11.1 percent of regional personal trip-making within the geofenced area. This simulation work also assumes that individual travelers may share rides through dynamic ride-sharing (DRS), which may pool two or more travelers with similar origins, destinations and departure times in the same vehicle. The presentation focuses on problem formulation and solution implementation details regarding SAV-traveler assignment, unoccupied vehicle relocation, and dynamic ridesharing. Model objectives in these problems seek to balance competing goals of minimalized total miles driven, as well as minimalized traveler wait (particularly long waits) and in-vehicle travel times. Multiple scenario variations are also tested, as well as a fleet size optimization procedure that seeks to maximize return on investment by a private operator. Results show that each SAV is able to replace around 10 conventional vehicles within the 24 mi × 12 mi area while still maintaining a reasonable level of service (as proxied by user wait times, which average just 1.0 minutes), though up to 8 percent more vehicle-miles traveled (VMT) may be generated if DRS is not utilized, due to SAVs journeying unoccupied to the next traveler, or relocating