Learning self-localization with teaching system
Takeshi Fukase, Hayato Yuasa, Tamio Arai · 2002
Learning approach is effective to design a agent that takes action autonomously. We propose to adopt teaching approach to design autonomous agents for RoboCup soccer tasks. Our teaching system consists of the ceiling CCD camera, SONY legged robots, and a PC for the teaching terminal. As a first step of the teaching process, we realized the self-localization of the robot using BP algorithm of feed forward neural networks.