Towards Neural-controlled Artificial Legs using High-Performance Embedded Computers

Abstract

The objective of this study is to develop a trustworthy and high-performance neural- machine interface (NMI) that accurately identifies user intent in real-time for neural control of artificial legs. We propose novel techniques to decipher non-stationary neuromuscular control signals collected from transfemoral amputees and accurately process such signals using a new high performance embedded system. Facing the challenges in integrating human neuromuscular control (physical) system with cyber systems, in this project we have addressed two CPS themes: (1) engineering methods and algorithms that permit tight integration of cyber and physical systems and expand the system's capability to deal with uncertainty; (2) embedded NMI system development for artificial legs. Our NMI design consists of a neuromuscular-mechanical fusion algorithm for user intent recognition and a trust management module to ensure the NMI's robustness and trustworthiness. The developed NMI algorithms have been specifically tailored to high-performance embedded systems (e.g. GPU and FPGA) for real time implementation. In order to evaluate this CPS, the NMI has been connected to a virtual reality cyber system or a powered transfemoral prosthesis worn by a lower limb amputee. Our prototype implementation demonstrates the feasibility of using neuromuscular-mechanical fusion to reliably control a virtual limb or powered prosthetic leg in real time. The developed NMI enables the user of powered prosthesis to perform various tasks intuitively, smoothly, and safely, which has not been achieved by the existing prosthesis control approaches. Therefore, our breakthrough in lower limb prosthetics has a potential to significantly improve the quality of life of patients with lower limb amputations. In addition, this project can benefit the CPS research community by providing novel conceptual models and engineering frameworks for design of powerful and reliable CPSs with human-in-the-loop.

(Real-time performance of our designed NMI can be found at our website

http://www.ele.uri.edu/faculty/huang/index_files/videos.htm)

Award ID: 0931820