An embedded feature-based stereo vision system for autonomous mobile robots

Abstract

Stereo vision systems have recently become an essential part for most autonomous mobile robots. There are several commercially available stereo vision products, which have been used for autonomous functions in mobile robot platforms. However, most of them are not suitable for compact sized robots. This paper presents an embedded stereo vision system that provides flexible baseline for robots of compact size. In terms of hardware, it provides baseline flexibility, and can be easily fitted into any robot. In terms of software, a feature based stereo vision algorithm is proposed to improve the real-time performance of the stereo vision system. The proposed featured-based method is evaluated by comparing it with the area-based method using standard test images. The results show that the computational time is significantly reduced (reduced by 60% in a relatively complex image). The proposed method is also implemented in real world environments in order to test its effectiveness.