A Wireless, Smart EEG System for Volitional Control of Lower-Limb Prosthesis

Abstract:

Objective: A powered prosthesis is one of typical cyber-physical systems (CPS) with a human-in-the-loop. The human and prosthesis interaction is highly complicated; although the human user could learn to manipulate the prosthesis, increased effort from the user would be required. The prosthesis requires tuning to minimize the user’s energy expenditure such that the user can use of and interact with the prosthesis effortlessly and with comfort. Moreover, a smooth cooperation between the amputee user and the prosthesis is needed in order to adapt to altered situations and environments. If the prosthesis can ‘intelligently’ understand the user’s volition and subsequently provide adaptive support to the user, the user will use and interact with the prosthesis more comfortably with less effort owing to the reduced effort and load for the prosthesis manipulation and control. This brings a unique CPS challenge on the intelligent sensor technology to enable a robotic sense of the user’s volition. This project will develop CPS technology for the prosthesis optimization to minimize the user’s energy expenditure and for extending the capacity of prosthesis to adapt to dynamic situations and environments.

Impact: More than one million people are living with lower-limb amputation in the United States. The prosthesis optimization as well as the user’s control of prosthesis will promote a natural gait and minimize an amputee’s energy expenditure in prosthetics use. An optimized prosthesis with user control capability will increase equal force distribution and decrease the risk of damage to the intact limb from the musculoskeletal imbalance or pathologies. Maintenance of health in these areas is essential for the amputee’s quality of life and well-being.

Project Preliminary Results: Smooth and convenient user control of prostheses is critical to restore motor and cognitive functions in amputees. This study aims to develop a smart system supporting volitional control of prostheses directly from user’ thought without any interface manipulation. An ultra-portable, wireless-enabled, and low powered device prototype was developed to amplify and process multi-channel electroencephalography (EEG) signals in real-time. Signal processing and machine learning algorithms were designed to decode user’s volition from EEG signals and subsequently to command prosthetic devices. A feasibility test was conducted on an amputee user with right leg amputation. Preliminary results showed that the amputee user was able to volitionally control a knee-locker installed on the prosthetic leg at a sensitive rate of 83.5% with zero false positive detections in real-time. Future study will be directed to support multi-functional, volitional control of robotic prostheses.