Performance-Impact Based Distributed Task Rescheduling for Multi-UAV Multi-Task Scenarios

With the rapid development of UAV technology, optimizing task allocation is a promising approach to improving the efficiency and application breadth of multi-UAV systems. A major bottleneck in task allocation for multi-UAV systems is how to optimize multiple targets simultaneously. This paper considers the problem of maximizing the number of task assignments in a distributed multi-UAV system under strict time constraints while minimizing the waiting time for tasks. The proposed method is based on the well-known PI-MaxAss algorithm and improves it with the full-shuffle task sequence allocation (FSTSA) strategy. The basic idea is that if an unassigned task can replace the assigned task of a certain UAV, the unassigned task will be reassigned locally together with the tasks in the UAV's task sequence. Another serious issue is that the UAV may fall into an infinite loop for task allocation, which hinders task performance and significantly prolongs overall completed time. We address the deadlock problem by introducing a deletion record table and a flag for new task inclusion. Numerical simulations show that the proposed method can achieve more task allocation while reducing the average waiting time. Especially compared with PI-MaxAss in large-scale complex scenarios, its advantages are more obvious.