제목

Mulpiple Top-view Synthesis for Humanoid Robot Navigation

저자명

Quing Lin, Sunhyoung Lee, Youngjoon Han, Hernsoo Hahn

영 문초록

 

This paper presents a navigation approach for humanoid robot based on monocular images only. To detect arbitrary obstacles around robot, the image captured at each single view is segmented using a region growing algorithm based on color pixel correlations. To observe a larger area for optimal path planning, the obstacle positions detected at each camera view are mapped and synthesized into a virtual plane of top-view, based on which a grid map is generated for path planning. Experiments with a humanoid robot in an indoor environment show that the proposed approach is effective in reducing the blind areas of robot during navigation. With a larger range of potential obstacles to be detected, the robot is able to reach the destination position more efficiently.

원문 수록처

6회 한국로봇종합학술대회, pp. 191~196

자료유형

학술대회자료