Highly Cooperative Task Planning Based on Automated Coalition Formulation and Hierarchical Task Specification

Cooperative multi-robot system has been playing a significant role for its high flexibility and robustness in task implementation. However, without proper modeling and solving method, the rigorous cooperativity and orderliness would not be guaranteed especially for large scale multi-robot systems. To further tackle these difficulties, this paper proposes a novel task planning method based on automated coalition formulation and hierarchical task specification. The main innovations include: 1) an automatically formulated coalition with Nash equilibrium is introduced for higher cooperativity between robots with different ability-types; 2) a dual-level task hierarchical specification tactic is proposed to decompose the complex task into concise fragments and improve the planning performance; 3) a post-earthquake rescuing task planning simulator is constructed with multiple temporally ordered heterogeneous tasks and execution time constraint. Extensive experiments with other 3 state-of-the-art (SOTA) methods imply that the adopted strategies in our method (Ours) can indeed boost the planning performance on both cooperativity and task implementation efficiency.