Fleet Sizing for Pooled (Automated) Vehicle Fleets

This paper proposes an (automated) on-demand public transport service using different vehicle capacities to serve current car demand in cities. The service relies on space and time aggregation of passengers that have similar origins and destinations. It provides a point-to-point service with predefined pick-up and drop-off locations. In this way, detours to pick-up en-route passengers is avoided. The optimization problem that minimizes the fleet size along with limiting rebalancing distances is defined as a mixed-integer linear programming problem. Solving the problem for Zurich, Switzerland, yields, in the best case, a fleet size equal to 3.7% of the current fleet that could serve current car demand. Vehicle kilometers traveled could also be reduced by nearly 10%. Results also show that the speed of automated vehicles has a substantial effect on the necessary fleet size and free-flow speeds generally produce over-optimistic results