A rolling horizon procedure for dynamic pickup and delivery problem with time windows

In this paper, the dynamic Pickup and Delivery Problem with Time Windows (PDPTW) is studied, in which requests from customers arrive continuously, vehicles may breakdown, and time windows of customers may change. It is an NP-hard problem and impossible to find the optimal solution in a reasonable short time. Therefore, the exact methods are too limited to tackle this dynamic problem and the heuristic method is the only way. Inspired by the rolling horizon optimization theory from predictive control, a periodic and event-driven rolling horizon procedure is proposed to deal with the dynamic PDPTW. Simulation results show that our method is suitable for solving this class of dynamic problems. Finally, the conclusion is offered and some directions for further research are indicated.