Traffic management for last-mile public transportation systems using autonomous vehicles

We introduce our research project aimed at developing a last-mile public transportation system with autonomous vehicles. In particular, we focus attention on traffic control techniques. Our vehicles can form fleets of autonomous vehicles, each of which consists of a lead vehicle driven manually and some driverless vehicles running behind another vehicle. In our system, the transport requests of all passengers are aggregated in real time to a central management server, and according to the information received, an optimized operation schedule is dynamically determined. Based on this system, we present a technique that by reorganizing vehicles does not require passengers to transfer to other fleets. In addition, to ensure fault tolerance of the system against failure of the central server, we propose a method to automatically switch to a provisional operation schedule for continuing transportation service. We also describe a scheme to implement our vehicle operation management adopting techniques of parallel computing and virtualization.