Wearable lower limb neuroprosthesis: system architecture and control tuning

Abstract

The use of functional electrical stimulation (FES) through neuroprosthesis is becoming a promising solution in lower limb neurorehabilitation. However, the wearability constraints and time-consuming tuning of stimulation parameters still limit the daily use of neuroprostheses. This work proposes two major contributions, namely: (i) a conceptual design and technical architecture of a fully wearable lower limb neuroprosthesis; and (ii) a Matlab-OpenSim framework that enables fast subject-and muscle-specific tuning of FES controllers based on OpenSim musculoskeletal models. The validation procedures for this study were divided into three phases: (i) Verification of the system architecture real-time requirements; (ii) evaluation of the reliability of the MATLAB-OpenSim framework for tuning PID controller; and (iii) its subsequent use in the neuroprosthesis control with a healthy subject. The obtained results demonstrated that the neuroprosthesis system was able to meet the real-time requirements, with control and data acquisition call periods below 10 ms. Further findings indicated reliable and stable behavior of the simulation-tuned PID controller with an overshoot of 9.82% and a rise time of 0.063 s. The trajectory tracking control results with the neuroprosthesis corroborated the robustness of the tuned PID controller in tracking the desired ankle trajectory (RMSE = 17.23 ± 2.97º and time delay = 0.21 ± 0.070 s).