Experimental evaluation of vision-based obstacle avoidance function for power assisted wheelchair

This paper presents a power assisting control with driving assistance of avoiding obstacle for electric wheelchair. The authors have been developing several vision-based power assist control in past researches. In the paper, the simplified driving assist using visual information is evaluated experimentally. The stereo camera system is mounted on the wheelchair, and distance information in surroundings is obtained through image processing. The admittance controller is implemented for power assist functions, and then virtual rotational force is generated from vision-based information in order to assist the driver's operation for avoiding the obstacle. The proposed function allows for easy driving in complicated environments. The validity of our approach is evaluated by several experimental results.