Realization of Stable Free Motion on 1DOF Admittance Controlled Knee Exoskeleton Using Prediction of Extension-Flexsion by EMG signal

Abstract

Control system design of exoskeleton type robots is basically different to that of conventional or industrial robots which requires robust motion control. Exoskeleton type robots always interact with human during achieving its own goal, so designing control system of exoskeleton type robot must encounter important issues such as ensuring safety, adaptive and robust performance. Exoskeleton type robots can roughly classified into three categories according to its purpose or target performance, assistance, rehabilitation and human-power augmentation. If attention is narrowed on powered-exoskeleton which has at least one actuator, the common requirement of the each types of exoskeleton to achieving its target performance is to make the exoskeleton interacts with human as if the exoskeleton system acts like desired virtual environment. For example, exoskeleton for assistance of raising arm should have environment which push the human arm upward so that it can assist desired raising motion of human arm. The virtual environment which the exoskeleton simulates could be combination of realistic mechanical system, or any virtual characteristics which actually do not exist in real world. This requirement of rendering virtual environment is common issue on haptics technology which has purpose on transparently simulating virtual environment to human operator. Haptic display have a limitation about rendering virtual environment on two extremes, high impedance and low impedance environments. For given haptic device, when haptic display render hard surface or free motion, It is impossible to render complete hard surface or free motion. It is previously investigated by many researchers that its higher and lower boundary can be written in relatively simple formula which includes system characteristics. This range of rendering is important indicator of evaluation of haptic display performance. Exoskeleton type robots have advantage in gathering extra information from biosignal which enable the system to estimate or predict human movement because exoskeleton usually rigidly bind to human. This paper proposes triggering algorithm using surface electromyography(sEMG) signal for improving performance of haptic display during free motion on admittance-controlled 1DOF exoskeleton. Within the framework of haptics, The purpose of this research is expanding lower impedance boundary of virtual environment by implementing triggering algorithm using sEMG signal. For avoiding complexity caused by including human model, this triggering algorithm is driven by simple pattern-recognition of sEMG signal, not by quantitative evaluation of the signal. Two-port absolute stability criteria is considered for designing the exoskeleton control system so that it guarantees stability with arbitrary characteristics of human operator and virtual environment. The limitations of conventional haptic display to implement free motion and the concept of the triggering algorithm is illustrated. The performance of proposed algorithm is presented by simulation results and experimental results.