表題番号:2018A-047 日付:2019/03/13
研究課題歩道通行や交差点横断を特徴とする屋外における歩行者誘導ロボティック・システム
研究者所属(当時) 資格 氏名
(代表者) 理工学術院 大学院情報生産システム研究科 教授 松丸 隆文
研究成果概要
Outdoor pedestrian-navigation robotic-system characterized by passing sidewalk and crossing intersection: We studied path planning for pedestrians in outdoor environment using two-dimensional digital map. The two-dimensional digital map is obtained on the network like Google Maps and OpenStreetMaps. However, it doesn't record all the data about crosswalks and pedestrian paths, for example outside urban areas. Therefore, a path planning which does not depend on preliminarily-recorded map data for pedestrians or large amount of data, for example, to execute SLAM (simultaneous localization and mapping), should be realized. Once given the departure and the destination, it gets the map data around and between them. First, it performs an image processing (contour detection) and visually recognizes city blocks. Next, a graph theory is applied to deduce the pedestrian path from the departure to the destination. In the trial using actual map data, it was possible to plan a reasonable path with 70 to 80% of success rate, including which side of the road to go through. In the future, we plan to detect pedestrian crossings and footbridges from satellite images and merge them into graph data, and to guide pedestrians using a mobile robot in actual environments.