Design and operations of on-demand multimodal transit systems

On-Demand Multimodal Transit Systems combine dynamically dispatched shuttles in low-density areas with high-occupancy vehicles to serve high-density corridors. They simultaneously address the infamous rst/last mile problem by pickingup/dropping-o riders close to their origins and destinations, congestion by aggregating rides in crowded arteries, and parking issues by removing cars from downtown. They ll a sweet spot between ridesourcing services and traditional transit systems. This paper reviews a series of optimization problems behind the design and operation of these systems: networkdesign, trip splitting, zone allocation, eet sizing, rostering, and real-time dial-a-ride optimization. Experimental results on Ann Arbor demonstrates the potential benefits.