Position sensing and control with FMG sensors for exoskeleton physical assistance

Abstract

Human intention decoding is a primary requirement to control an exoskeleton. In this work, a new method of decoding human intention by Forcemyography (FMG) is explored to estimate elbow joint angle during arm motion. The method utilizes an FSR-based sensor band to read muscle contraction and relaxation. The readings of the sensor band are mapped to the desired joint angle by using coarse Gaussian support vector machine (SVM) regression algorithm. The estimated joint angle is further used to control an elbow joint exoskeleton. Results show that the new method is able to estimate reliably the joint angle for controlling the exoskeleton.