Continuous angle estimation of lower limb multi-joints based on synergy activation model

The neural information decomposed from EMG signals provides a new way of thinking about human-robot interaction. Distinguished from the traditional EMG time-frequency domain features, muscle synergy analysis can capture synergistic motion information on multiple degrees of freedom, and this information provides more intuitive and natural information for intelligent rehabilitation. In this paper, we propose a neural interface for hip-knee-ankle joint angle prediction of human walking lower limbs based on muscle synergy advanced neural control signals, and compare the prediction results of different walking speeds and different features for the model to find out the best estimation method, which has great application prospects in the scenarios of exoskeleton robot control and rehabilitation training.