Hands-Free Wheelchair Control Using Vision-Based Intention Estimation

Abstract

This paper presents a hands-free interface for controlling an omnidirectional electric wheelchair, designed to enhance mobility for individuals with motor impairments. The system utilizes multiple cameras to track face and torso movements, enabling natural and non-invasive driving system. A vision-based Intention Estimator integrates input from both the rider and caregiver, allowing for shared control. The proposed method ensures smooth navigation in confined indoor spaces, improving user experience. Experimental results demonstrate the feasibility of the approach, though challenges remain, particularly in addressing lighting conditions that affect facial landmark detection. Future work will focus on enhancing drivability and solving illumination issues to ensure reliable operation.