Thumb-tip force prediction based on Hill`s muscle model using non-invasice electromyogram and ultrasound signals

Abstract

Restoring human limb that is lost during accident with prostheses is one challenging issue in engineering field. The loss of human limb, involving upper and lower limbs leaves great impact on human that it limits daily activities in many ways. There are many studies that have been done previously in making and improving prosthetic unit for amputees. Most of them are made to replicate the basic functionalities of the missing limbs and adopt more conventional passive controllers. As a result, the prosthetic unit range of motion is limited and the motion is unnatural. Hence, there is a need to develop model based controller for such system to address the problems. The goal of the research is to develop a semi-analytical model of a thumb that could be used to develop a model based controller. In this work, the information from the muscle characteristics is gathered. The electromyography signals from the four muscles responsible on thumb flexion are measured and recorded using biosignal measurement system. The thumb tip force is measured using thumb training system developed for the research. On top, information such as the length of the muscles and tendons is collected from the ultrasound probe and magnetic resonance imaging (MRI) machine to increase the data accuracy. These data are fed into Hill's muscle model and optimized by the particle swarm optimization (PSO) technique to map the relationship between thumb posture, thumb tip force and all the signals measured. The resulting thumb model developed using the method proposed in this research work has shown lower root mean square error (RMSE) as compared to previous method.