Departmental Bulletin Paper Multi-Agent Reinforcement Learning Collaborative Environment

李, 嘉誠

As in the multi-agent system, access to new knowledge and adapt to the new environment is one of the essential characteristics of agent.So Agent need to learn to adapt to the dynamic changes in the environment, in order to obtain optimal or sub-optimal behavior strategy. This paper studies multi-robot systems,proposes a further reinforcement learning algorithm and applied to multi-robot system. The multi-robot path trace planning issues were discussed.We designed a reinforcement function, the design method proposed reinforcement piecewise function. The simulation results show the effectiveness of this method.

Number of accesses :  

Other information