Distributed Task Allocation for a Multi-UAV System with Time Window Constraints

Abstract

This paper develops a distributed method, namely, distributed allocation with time windows (DATW) for managing the multi-UAV task assignment problem (MTAP) with complex time window constraints. By aiming directly to minimize the average task completion time, the pro-posed DATW intends to achieve a conflict-free result that allocates all tasks within the validity time windows. Based on the decentralized PI (Performance Impact) framework, the proposed algorithm addresses the MTAP in a three-phase task assignment strategy, which includes task inclusion, conflict resolution, and task reallocation. The newly introduced task allocation phase achieves a noteworthy increase in an average number of allocated tasks. Unlike the traditional PI methods, the start time of each task is broadcasted among agents via communication typology, and the significance value of each task is directly related to its validity time window, such that the vast majority of tasks are able to be assigned properly without imposing any extra communication burdens. In the obtained conflict-free allocation solution by DATW, each task is allocated to a proper UAV with all given constraints satisfied. Finally, the simulation results demonstrate the effectiveness and superiority of the proposed DATW. Compared with existing (CBBA-based) solutions, results show up to an 18% increase in success rate (SR) using the proposed method.