A Novel Obstacle Detection Method Based on Monocular Camera and Laser Radar
Siyang Mao, Lunbo Li, Jianhui Guo, Chunxia Zhao · 2015
Obstacle detection is an important task in the field of environmental perception for Intelligent Ground Robots. This paper presents a new obstacle detection method which uses visual saliency detection to detect multiple obstacles in the scene. By analyzing the saliency map and the marker-based watershed algorithm, we combine their characteristics creatively to obtain accurate contours of obstacles, and propose a fast method to detect obstacles based on monocular camera and laser radar. The laser radar points representing obstacles are clustered, and projected onto the saliency map as markers of watershed algorithm. The experimental results indicate the effectiveness, robustness and performance of real-time of our method.