10205 センサ選択型融合システムを用いた移動ロボットの目標探索(OS15 知能ロボット)

概要

論文の詳細を見る
In recent years, robot technology has remarkably developed and many researchers have worked on mobile robot behavior control. We suggest a sensor seiected fusion method based on neural network for a mobile robot navigation. In this method, we evaluate sensor values by possibility measure and select useful ones among all sensors according to selection rule with fuzzy logic. Back propagation algorithm is applied to neural network learning. In some cases, it's hard to determine the teach data for learning, therefore the data is acquired by GA. In addition, self-organizing map is used as a constraint condition in evolution. Through a simulation of mobile robot navigation, the effectiveness of the proposed method is clarified.
社団法人日本機械学会の論文
2005-03-17