Fast Obstacle Detection for Monocular Autonomous Mobile Robots

Abstract

This paper proposes a monocular vision based obstacle detection algorithm for autonomous mobile robots. Our main algorithm consists of two stages. In the first stage, we use an inverse perspective mapping (IPM) based method for detecting small portions of an obstacle in the input image. In the second stage, we perform image abstraction and geodesic distance computation for segmenting the obstacle. We use the simple linear iterative clustering (SLIC) superpixel algorithm for decomposing the image into basic elements that preserve relevant structure, but abstract undesirable detail. The source superpixel for geodesic distance computation is selected according to semi-local texture features. Then we compute the obstacle score for accurate segmentation. Experimental results have shown that our proposed method achieves accuracy comparable to the state-of-the-art method, with more than 7 times faster computation.

Keywords

• segmentation
• obstacle detection
• mobile robots
• inverse perspective mapping
• geodesic distance