Optimizing the Speed and Accuracy of an EMG Interface in Practical Applications

Abstract

Due to the development of robotic rehabilitation technologies and modern electromyography (EMG) command-proportional control, the issues of muscle activity signal processing remain extremely relevant. In this paper we propose several options for optimizing the processing techniques for EMG signals. Our rather simple approaches improve the efficiency of the EMG interface in various parameters. In particular, the new method of signal averaging involves the calculation of two moving averages: the main “slow” average and an additional “fast” average. If the difference between the mean values exceeds a threshold, the slow mean is equated to the fast one. Another approach includes the use of a nonlinear (power) function for the proportional control of an object. The proposed approaches were tested in tasks on controlling virtual objects by 15 healthy subjects. It was found that the criteria such as the interface response time and the smoothness of movement of an object, as well as the accuracy and speed of control, were optimized. The results can be used in control systems of a wide range of robotic devices, including exoskeletons, prostheses, and wheelchairs.