Blending User Intent and Obstacle Avoidance Through Artificial Potential Fields: An Experimental Validation on Smart Walker

Assistive robotics plays an integral role in reshaping the standard of living and offering mobility autonomy to elderly and impaired individuals. Smart Walkers have gained popularity in the recent past as a tool to support independent mobility. However, room for improvement remains. In particular, achieving a robust shared control authority between the walker and the user presents a complex engineering challenge. This paper presents an Artificial Potential Field based shared control system that provides smooth obstacle avoidance within indoor environments. Unlike typical systems that rely on expensive handle-mounted force sensors, our approach utilizes a sensor-less force estimation technique based on motor telemetry to infer user intent. We employ a 2D LiDAR to detect obstacles and generate a virtual repulsive field. This field considers repulsion from obstacles as a negative force and user intention as a positive force. A novel steering mechanism based on obstacle distribution asymmetry guides the user away from hazards. We fuse these inputs through a virtual admittance controller to ensure smooth locomotion. Experimental results demonstrate the efficacy of the proposed methodology. The walker successfully avoids obstacles while maintaining a natural feeling of control for the user. It achieves true shared autonomy where the robot handles safety and the human handles high-level planning.