비전기반 문 추출과 격자지도를 이용한 이동로봇의 영역단위 위상지도 생성 기법

Abstract

This paper proposes a method to generate a topological map using visual door detection and grid-map for mobile robot. The candidates of door detected from ceiling view image are defined as Image door, and the candidates of door detected from occupancy grid-map are defined as virtual door. The image doors are detected by using Canny edge detector and Hough transform, and the virtual doors are detected by extracting convex corner features as the posts of door from grid-map. Final doors are detected from image doors and virtual doors and the posts of final doors are registered as post feature. Initial one node topological map is generated, and the node is defined as a polygon that consists of corner features. The final topological map is generated by dividing the node of initial topological map into several nodes. As a result, the generated topological map consists of nodes and edges

each node consists of corner features and post features, and each edge is located on the position of real door. The proposed method provides a topological map for the user interaction and the mobile robot, and the topological map can be used to plan more efficient motion including room-to-room path planning and covering each room. Experimental results verify the performance of proposed method in real home environment.