Dynamic Task Allocation and Pathfinding under Precedence and Temporal Constraints

Abstract

Under precedence and temporal constraints, a novel framework is proposed to optimize both dynamic task allocation and pathfinding for a multi-robot system. Firstly, the dynamic task allocation is solved by a greedy-based algorithm combined with a metaheuristic algorithm that satisfies precedence and temporal constraints, which dynamic update and re-planning of the tasks also handle during the task execution. Next, a conflict-based search algorithm is exploited to accomplish the multi-robot pathfinding. With the priority-based objective function, the proposed scheme initially maximizes the utility of the tasks, and accordingly minimizes the makespan of all robots. Simulation and experimental results are eventually demonstrated to verify the feasibility of the proposed framework.