Pattern-based grasping force estimation from surface electromyography

Abstract

Aiming at maintaining the accuracy of grasping pattern recognition meanwhile evaluating the required force, this paper uses Linear discriminant analysis (LDA) to realize pattern recognition and artificial neural networks to establish the relationship between surface EMG signals and fingertip force in each grasping mode. Once the grasping pattern identified, the program calls the corresponding force model to estimate force value and achieve the combination force decoding and pattern recognition. The experiment shows that the force predicted with an average error of 10% meanwhile the average classification accuracy is about 83.21%.