发育学习在足球机器人基本动作技能中的应用

Application of Developmental learning to Basic Action of Soccer Robot

研究了发育学习算法及其在机器人足球比赛技术动作学习问题中的应用。结合发育学习算法的优点,选用合适的强化学习算法,并将其应用于足球机器人动作技能的学习中。无需任何先验知识和环境模型,通过不断与环境交互获得知识,自主地进行动作选择,具有自主学习能力,在自主机器人行为学习中受到广泛重视。最后,给出了试验结果分析,并验证了该算法的优越性和有效性,并且能够满足高水准机器人足球比赛的需要。

The Developmental learning algorithm and its apphcation to technical action learning of soccer robot are discusses. According to the advantages of the developmental learning, the proper reinforcement learning algorithm is choosed, and applid it to design technical actions learning of soccer robot without any prior knowledge and environment model involved. It can autonomously improve its behavior policy with the knowledge obtained by continuously interacting with the environ.ment. Finally, the experiment results are also shown to prove the presented method is effective and superior, and it can meet the demands of the high level robot soccer match.