Comparing Parking Strategies of Autonomous Transit On Demand with Varying Transport Demand

Abstract

Autonomous transit on demand are increasingly considered to become a viable substitute for taxi services. AVs can be managed through a centralized controlling system, targeting system optimization rather than user optimality. This centralized control can enable a more efficient, strictly-adhered-to parking strategy to reduce inefficient empty traveling. In this project, four different parking strategies are implemented in the AV extension of MATSim (Multi-agent transport simulation), namely demand-based roaming, parking on the street, parking in depots and a mixed strategy of parking on the street and in depots. The influence of different PT demand levels on the different parking strategies was explored, showing that the shared system is robust to varying levels of demand, and that the different parking strategies trade off user convenience for operational cost. The road parking strategy appears to be the best for consolidating rides into larger vehicles, especially for the increased demand scenario.