Operational transport planning with incidents

This research addresses robust incident management techniques in distributed environments with autonomous actors. Focusing on applications in operational pickup and delivery transportation planning, the distributed environment consists of a transport network, vehicles controlled by operational agents and pickup and delivery tasks with time-windows. The two main challenges of this research are to explore and develop context-aware routing methods (routing methods that are aware of other traffic) and methods that are robust even in uncertain environments. This uncertainty in the pickup and delivery domain arises from changes (or newly arriving) transportation requests, malfunctioning transportation means or traffic jams in the transport network (e.g. due to accidents). A multi-agent transport planning simulator (Traplas) has been developed for experimenting in the pickup and delivery domain. Where analyzing the problem does not lead to satisfying answers, Traplas is used to test and compare the different methods developed during this research.