Obstacle Avoidance Based on Optical Flow for Mobile Robots in Unknown Environment

Abstract

A new method for mobile robot to avoiding obstacle in unknown environment is present in this paper. The robots trajectory is modified with obstacles distribution through obstacle map, which is computed using optical flow (OF) and Time to Contact (TTC) provide by a monocular vision sensor. In this paper, the field of view (FOV) is divided into several equal column, each column have a value which determined by OF, TTC and target bias is used to determine the preference heading direction. Take a comparison with use OF or TTC only, our method is more appropriate for avoid obstacles, because it can not only release the dependence on texture and light of the environment, but only it can reduce the impact on TTC which is caused by noise pixel. Simulation results verify that our method is effective.