Opportunistic planning for a fleet of transportation robots

Abstract

The Dynamic Transportation-Planning Problem (DTPP) embodies a class of real-world applications that involve the reactive routing and scheduling of a fleet of vehicles in response to dynamically changing transportation demands. Examples include mobile robots in a warehouse, taxis in an urban road network, or aeroplanes for medical evacuation. In contrast with the Vehicle Routing Problem, for which a plethora of techniques is available, few approaches exist that permit the efficient deployment of a large number of vehicles in a changing environment. This paper highlights the characteristic features of the problem, reviews possible approaches and existing techniques, and proposes a heuristic solution to the DTPP using a Blackboard-based approach. The resulting application is an intelligent transportation planning system (ITPS) for a fleet of automated robot taxis, based on the generic assumption-based truth maintained Blackboard shell (GATMBS) and comprising traffic simulation models as well as various monitoring and problem-solving strategies for assignment and routing. The Blackboard architecture supports the dynamic alteration of planned routes in response to changes in traffic conditions and passenger requests. A prototype of the ITPS has been validated in simulation using a small fleet of robot taxis with randomly generated road networks, background traffic load, and passenger requests.