Development of a simulator for robots and its application to position estimation by self-organization

摘要

We have developed a simulator for robots to effectively build an environmental map and estimate position of a robot as the first step to robotics study. Firstly, we self-organize an environmental map based on sensory information, motor information and position information. Secondly, a robot runs autonomously based on the resulting self-organized map. Two kinds of SOMs, i.e., forward SOM and backward SOM, are developed and used. Bayesian position estimation is also developed.