The impact of autonomous vehicles on commute ridesharing with uncertain work end time

•Work schedule uncertainty prevents ridesharing between commuters with identical home and work locations.•Autonomous vehicles (AV) alleviate the risk of ridesharing because additional AVs can be easily repositioned if needed.•We develop evening commute bottleneck models with uncertain work end time.•We establish mode choice (ridesharing vs. travelling alone) equilibrium for round-trip commute.•Autonomous vehicles encourage commute ridesharing and reduce expected round-trip travel costs. The uncertainty in work end time can prevent ridesharing between two commuters with identical home and work locations. This effect can be alleviated by autonomous vehicles (AV): when two commuters share a ride from home to work, if the realized work end times result in a long wait between the two, an additional AV can be easily repositioned from home to work, whereas with regular human-driven vehicles one has to either bear the long wait or request expensive taxi service. This observation implies that AVs have great potential for increasing commute ridesharing. To study this effect, we develop evening commute bottleneck models with uncertain work end time and establish mode choice equilibrium for the round-trip commute. We find that, with regular vehicles, because of the expected waiting cost caused by work end time uncertainty, commuters typically do not share rides even though they have the same home and work locations. With AVs, commuters always share rides for the morning commute, and they also share rides for the evening commute if their work end times turn out to be close; otherwise, they reposition an additional AV to pick one of them up. Our numerical examples show that AVs can encourage commute ridesharing and significantly reduce the expected round-trip travel costs for commuters.