Fuzzy knowledge inference-based dynamic task allocation method for multi-agent systems

Mengke Wang, Xiaoming Wang, Xinzhi Wang, Xiangfeng Luo, Shaorong Xie

Abstract: Highlights•Fuzzy reasoning enhances task allocation, fills missing info, and boosts agent cooperation under local observation.•Fuzzy rules express task assignment knowledge, optimize search space and improve task matching accuracy.•Dual-layer framework updates task allocation and policy learning, ensuring stable training and effective task execution.•This method validated in StarCraft II, Google Football, and particle environments.