Skill matters: Dynamic skill learning for multi-agent cooperative reinforcement learning

Tong Li, Chenjia Bai, Kang Xu, Chen Chu, Peican Zhu, Zhen Wang

Abstract: Highlights•Proposed a novel MARL method, called dynamic skill learning.•Skill discovery enables agents to learn diverse abilities effectively.•Solves agent behavior homogeneity from shared parameters, enhancing cooperation.