A Collaborative Robotic Approach to Gaze-Based Upper-Limb Assisted Reaching

Abstract

Robotic treatment has been recognized as a powerful tool to face the consequences of neurological impairments such as stroke. Among several potential benefits, the possibility to program high-intensity and repetitive movements increased the popularity of this type of rehabilitation technologies for upper limb disabilities. Nevertheless, although the results are encouraging, we are still far from guaranteeing a complete recovery to all patients, and some degree of external assistance is required to enable them to execute their activities of daily living. Hence, we aim to build a system which may serve not only as a rehabilitative tool, but also as a complement to assist the users in everyday tasks. In this direction, we present a novel Human-Robot Interface (HRI), which can detect and learn repetitive user movements through gaze, and provide assistance via a collaborative robot. The robot, which provides support to the user's wrist using a comfortable interface, executes the learnt trajectories, enabling the user to perform repetitive movements with reduced effort. The experiments are carried out on ten healthy subjects. We demonstrate that the proposed HRI is not only flexible and intuitive, but also accurate in guiding the users' limb to the desired positions.