Application of Optical Flow to Sonar Image for Mobile Robot Navigation

In this paper, a novel data processing algorithm for ultrasonic sensors is proposed using optical flow technique to get relative velocity of obstacles. Ultrasonic sensors are widely applied to robotic system, however, limitations exist for positioning object's location by angular uncertainty. Collision avoidance in dynamic environment requires the information of relative motion, such as velocity for obstacles in addition to distance. To assist operator's control skills, a force is generated for collision avoidance using ultrasonic sensors' data. Previous research on the generation algorithm of force reflection used relative distance between robot and obstacles. We applied optical flow technique to artificial image sequence which is generated from ultrasonic sensor's range data to obtain relative velocity. The experimental results show the effectiveness of the proposed algorithm for robot navigation in dynamic environment.