Abstract: In the multi-robot cooperative carrying process, traditional reinforcement learning only uses numerical analysis and ignored reasoning approach. To solve this problem, independence reinforcement learning for multi-robot combines with Belief-Desire-Intention（BDI） model, which makes reinforcement learning link logical reasoning capabilities. And the distance nearest principle is employed which means that the nearest robot ranged from obstacles is the leader robot to control other robots move. Evaluation function which changes with the location of multi-robot and the barriers is proposed, and it combines with the behavior weight based on reinforcement learning which becomes more and more optimized through constantly interacting with the environment. Simulation results show that this method is feasible, and the cooperative carrying process can be successfully achieved.

Key words: multi-robot, reinforcement learning, cooperative carrying, obstacle avoidance

摘要： 在多机器人协同搬运过程中，针对传统的强化学习算法仅使用数值分析却忽略了推理环节的问题，将多机器人的独立强化学习与“信念-愿望-意向”（BDI）模型相结合，使得多机器人系统拥有了逻辑推理能力，并且，采用距离最近原则将离障碍物最近的机器人作为主机器人，并指挥从机器人运动，提出随多机器人系统位置及最近障碍物位置变化的评价函数，同时将其与基于强化学习的行为权重结合运用，在多机器人通过与环境不断交互中，使行为权重逐渐趋向最佳。仿真实验表明，该方法可行，能够成功实现协同搬运过程。

关键词: 多机器人, 强化学习, 协同搬运, 避障