デッドレコニングおよび視覚認識の不確かさを考慮して移動ロボットを最適に誘導する手法

概要

論文の詳細を見る
This paper describes an method to efficiently navigate a vision guided mobile robot with an odometry system. First, the vehicle is driven by a human through a corridor. During this movement, a vision system extracts features such as doors or staircases. The categories and locations of the features are recorded in a map. This location data includes uncertainties of dead reckoning and vision. Then, given a destination, the vehicle examines the map and selects a sequence of useful landmarks so that the vehicle can arrive at the destination in the minimum time. In this selection, the vehicle considers the uncertainty of the map and that of dead reckoning. The vehicle passes through unselected features with dead reckoning and then, verifies the selected landmark by vision. Experiments are performed for our autonomous vehicle which moves in our building.
一般社団法人日本機械学会の論文
1995-01-25