Estimating Elbow Torque from Electrical Stimulation using a Particle Filter

This study analyses the relationship between functional electrical stimulation (FES) and the induced torque for elbow flexion. The aim is to develop an FES-torque model that is simple to implement and understand, and is easily invertible so that the required FES for a desired assistive torque can be determined to enable control by FES. For accurate control, the FES-torque model must also be adaptable to time-varying behaviour of the muscle such as fatigue. The proposed FES-torque model is a sigmoid function, and a particle filter is implemented to estimate the change in parameters of the sigmoid function over time. The results show that the particle filter is successfully able to adapt to changes in the FES-induced torque and can be used to improve the estimate of FES-induced torque, with an overall average RMS error of 0.24 N.m or 7.85%. The improved FES-torque estimate allows for simple and more effective control of FES assistance and better fatigue management. Copyright (C) 2020 The Authors.