基于遗传规划的机器人足球团体协作进化策略

Evolution Strategy of Teamwork in Robot Soccer Based on Genetic Programming

利用基于进化的方法,对机器人足球比赛中的团队协作策略进行设计。论文给出了一种新的强类型的遗传规划算法,它对传统遗传规划方法做出了较大的改进,主要是提出了两种新的计算方法:基于概率的树的生成算法1,和基于概率的树的生成算法2。这两种算法不仅算法简便,又可以让用户对树结构进行控制,从而使其能更好地适应机器人足球系统的实时性和动态要求。

Based on genetic programming（GP）, the evolution strategy of teamwork in robot soccer is described This paper introduces a new arithmetic of strongly - typed GP, which improves the traditional GP greatly. It concentrates on two new arithmetics：probabilistic tree - creation 1 and probabilistic tree - creation 2 . These arithmetics is simple , which can enable the users to control the tree size better. Then it can make the strategy to rather adapt the request of real time and dynamic requirement in the robot soccer.