Smart Wheelchair Maneuvering Among People

Abstract

The advancement of technology is increasing with several applications on robotics like smart wheelchairs providing autonomous functions for severely impaired users with less need of caregiver support. One of the main issues in robotic wheelchair research is autonomous pedestrian avoidance for safety and smooth maneuvering. However, this is difficult because most of the pedestrians change their motion abruptly. Thus we need a fully autonomous smart wheelchair that can avoid collisions with individual or multiple pedestrians safely and with user comfort in mind for crowded environments like train stations. This paper presents a method for our smart wheelchair to maneuver through individual and multiple pedestrians by detecting and analyzing their interactions and predicted intentions with the wheelchair. Our smart wheelchair can obtain head and body orientations of pedestrians by using OpenPose. Using a single camera, we infer the walking directions or next movements of pedestrians by combining face pose and body posture estimation with our collision avoidance strategy in real-time. For an added layer of safety, we also use a LiDAR sensor for detection of any obstacles in the surrounding area to avoid collisions in advance.