Study on real-time obstacle avoidance of mobile robot based on vision sensor

Abstract

Aiming at the problem of obstacles avoidance from a single camera mounted on a robot in an unstructured indoor environment, the paper proposed an algorithm based on template matching. Firstly, the scenepsilas free space and obstructed space can be obtained by means of template matching. Secondly, we can get the robotpsilas running orientation and speed from the free space based on the method of visual attractive force (VAF). Considering the variation of the lighting, we designed a template updating algorithm based on particle filter. In order to test the validity of the method, a lot of experiments were conducted in various lighting condition. The results show that the method could detect and avoid stationary obstacles and moving obstacles reliably. Meanwhile, the real- time property and robustness of the system have been validated.