Gaze Controlled Safe HRI for Users with SSMI

Abstract

Robots are already used for rehabilitation of people with disabilities mainly in the form of prosthetics or exoskeleton. This paper proposes a Markov Decision Process based safe distance maintenance algorithm to make a small robotic manipulator accessible to users with severe speech and motor impairment (SSMI). We have developed a video see through eye gaze controlled graphical user interface for operating the robotic arm while an intelligent agent monitors presence of users' hand in the vicinity of the robot and plans a navigation path using value iteration algorithm. A user study involving 5 users with SSMI and 7 able-bodied users found the system safe and usable by all users undertaking tasks within 3 minutes and significant decrease in task completion time with practice.