A Consensus-Based Joint Optimization Algorithm for Multi-agent Collaborative Dynamic Task Allocation.

In team-based task allocation, due to the full utilization of individual computing resources and getting rid of single node dependence, the advantages of distributed methods are always stronger than centralized ones. Therefore, in this paper, the multi-UAV system adopts a distributed architecture. We propose a consensus-based joint optimization algorithm. In the phase of task sequence construction, an improved differential evolution algorithm is used in the early stage, while the local adjustment is carried out in the later stage after converging to a stable solution. Then, for task sequence coordination among agents, the algorithm utilizes a conflict resolution strategy based on consensus principle. After the above-mentioned multiple rounds of task allocation, a conflict-free optimized solution is obtained. And tests confirm superior performance when compared with existing collaborative task allocation algorithms.