Evolutionary Acquisition of Robot Behavior by Genetic Programming Using Semantic Crossover
Semantic crossover has been proposed as the method for controlling global and local search in Genetic Programming. In the method, the pair of subtrees, which are used for crossover, are selected based on their semantic distance. However, in the robot control problems, it is difficult to consider the similarity of their behavior because they take actions under different situations. The purpose of this study is to propose the method to apply the semantic crossover to the robot control problems. In our method, multiple situations have been prepared beforehand, and the set of actions for the situations is used for the semantics of each subtree. Experimental results show that our proposed method can get optimal solutions faster than the conventional method.
2012 IEEE SMC Hiroshima Chapter Young Researchers' Workshop Proceedings = 2012 IEEE SMC Hiroshima Chapter 若手研究会講演論文集
IEEE SMC Hiroshima Chapter
©Copyright by IEEE SMC Hiroshima Chapter. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting
republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works.
2012 IEEE SMC Hiroshima Chapter若手研究会, 2012年7月14日, 広島市立大学