一种改进的多智能体Q学习算法

An improve multi-agent system Q-learning algorithm

针对多智能学习特点,提出了一种新的多智能体Q学习算法.算法中将多智能体转换为联合状态的单智能体来学习策略,同时利用改进的随机跳转搜索策略解决了Q算法易陷入局部最优解问题。仿真结果表明,将该算法应用在机械臂轨轨迹划中说明了算法的有效性与泛化能力。

Aiming at the study of MAS, we propose an improved MAS Q-learning algorithm, which convert the MAS into single-agent with the combination of the state, a new search strategy is introduced for the problem of local optimal solution in Q- learning. When applied to trajectory planning for manipulator, the simulation results show that the manipulator reaches the target position more quickly and to show the improve Q-learning algorithm is efficient and generalization.