Controlling an Exoskeleton with EMG Signal to Assist Load Carrying: A Personalized Calibration

Abstract

Implementing an intuitive control law for an upper-limb exoskeleton to perform force augmentation is a challenging issue in the field of human-robot collaboration. The aim of this study is to design an innovative approach to calibrate electromyography (EMG) data in order to detect the intention to lift or put down a charge while wearing an upper-limb exoskeleton. Based on a low-cost EMG sensor bracelet placed around the arm (Myo armband, Thalmics Lab, Ontario), a subject-specific mapping procedure is implemented to discriminate motion intentions during lifting tasks with a 1-DoF upper-limb exoskeleton. The processing is divided into two main parts: (i) Direction estimation with an artificial neural network, and (ii) A model-based intensity prediction. The mapping procedure has been tested on 7 healthy participants with a precision of 96.9 ± 3.1% for the classification and a RMS Error of 3.8 ± 0.8 N at the end effector. This study opens up the way for fast-deployment applications involving exoskeletons or cobots.