Model to Relationship the Speed of Hand Movements with the SEMG Signals from the Forearm

Abstract

Abstract Various previous works have sought to achieve hand prostheses with natural movements. It has been employed tools to recognize patterns of surface electromyographic signals associated with each move. Although many successful studies classify some types of hand movements with high performance, the results show that the speed and strength must be analyzed so that the resulting moves are like those of a natural hand. This study evaluates 23 healthy subjects at two different speeds and six types of movements (pronation, supination, ulnar deviation, radial deviation, flexion, and extension -276 records of SEMG and Velocity). The objective was to obtain a model (transfer function) that would allow the relation of the velocity profiles with the tone of the forearm SEMG signals. The results show models with an average RMSE of 18.55% for slow movements using low-order systems (2). The parameters of the models between subjects are very different, with high coefficients of variation and standard deviations, which implies that the fitting must be for each subject.