Automated planning for long-term missions with collaborative heterogeneous systems

Abstract

With the recent development of cooperative systems consisting of unmanned vehicles and intelligent robots, the demand for automated planning for large-scale and long-term mission performance is growing. As a generalized problem of such missions, this study addresses the mission planning problem for numerous tasks distributed over a wide area using heterogeneous cooperative systems. The objective of this mission planning problem is to minimize the makespan, considering concurrent actions for the efficient operation of long-term missions. Reflecting these conditions, the mission planning problem is defined and plans are generated using the traditional planning method, PDDL (Planning Domain Definition Language). While PDDL offers flexibility in handling various constraints, it faces a significant limitation in optimization performance as the problem size increases. Therefore, this paper proposes a hierarchical mission planning method that reduces the complexity of the problem using mission decomposition techniques and enhances the optimization performance for concurrent actions, ensuring efficient mission planning even as problem size grows. Furthermore, by extending its application to multi-fleets, the proposed mission planning methods demonstrates high scalability. To evaluate the generated plans, a simulator is developed and utilized for visualization, thereby verifying the performance of the proposed mission planning method.