Title :
Gene creation for a mobile robot by virtual training
Author :
Kobayashi, Hisato ; Yokota, Sho ; Blazevic, Pierre
Author_Institution :
Dept. of Art & Technol., Hosei Univ., Tokyo, Japan
Abstract :
This paper shows a method to generate an autonomous motor coordination for mobile robots. The method is to let robots learn their own walking strategy by trials and errors, namely by training in virtual environments. By using a dynamics engine, the robot training is done in various virtual environments under genetic algorithm with a proper natural selection.
Keywords :
genetic algorithms; learning (artificial intelligence); legged locomotion; motion control; robot dynamics; autonomous motor coordination generation; dynamics engine; gene creation; genetic algorithm; mobile robot; motion control; robot training; trial and error learning; virtual training; walking robots; DNA; Engines; Humans; Legged locomotion; Mobile robots; Motion control; Robot control; Robot kinematics; Robot sensing systems; Virtual environment;
Conference_Titel :
Advanced Robotics and its Social Impacts, 2005. IEEE Workshop on
Print_ISBN :
0-7803-8947-6
DOI :
10.1109/ARSO.2005.1511651
