LAUSR

The connectivity of urban road networks plays an important role in ensuring the functions of the infrastructure and the transportation of emergency supplies of a postdisaster city. When a disaster happens, clearing vehicles that are able to open blocked roads are precious resources in connectivity restoration and their work should be planned systematically and promptly. In this article, we proposed a novel reinforcement learning approach to swiftly providing restoration plans for clearing vehicles according to damage situations. This is a cluster-first, route-second type of method that includes an event-triggering algorithm to deal with the time-space model of vehicles and a specially designed cognitive module to decide the vehicle behaviors, which is represented by three kinds of preference in the behavior probability: the guidance, the memory, and the rules. Our approach has been tested on damaged Istanbul and synthetic road networks, confirming its superior capacity to generate near-optimal solutions in an extraordinarily short time, which is more practical and efficient in dealing with the case of multivehicle systems operating on complex large-scale urban road networks.