画像からの迷路情報取得に関する検討

概要

論文の詳細を見る
When conceptualizing a maze-escaping robot "micro mouse", a more rapid way of finding the shortest passage to get out of a maze than using a usual optical sensor might be possible if we were able to get the maze information correctly from sensor images. In a maze, however, many similar image patterns exist, so it would be more effective to do so if we used a unique technique that was different from that of ordinary image processing for normal landscapes. Herein, we present some techniques to grasp a maze structure quickly by getting various information out of the maze image data. Additionally, we examine the robot balance control, which is necessary for practical use of a robot. Furthermore, we discuss the structure of a robot control system that helps a robot get out of a maze by obtaining correct information from the associated image data.