Joint Motion Estimation Method Based on EMG Model and Unscented Kalman Filter

Abstract

The invention relates to a joint motion estimation method based on an EMG model and an unscented Kalman filter. Firstly, the biceps femoris, quadriceps femoris, vastus lateralis, vastus medialis, half The EMG signals and real-time angles of the tendon and gracilis muscles are processed by band-pass filtering, and the wavelet coefficients and root mean square features are extracted, and then a method that combines muscle dynamics, joint dynamics, bone dynamics and The state-space myoelectric model of relevant myoelectric features, through the unscented Kalman filter algorithm, obtains the Sigma sampling set χ i and weight W i , and then make further predictions, calculate the system state variables and covariance matrix P(k+1|k), and realize the estimation of the continuous motion of the knee joint after an iterative cycle. Compared with the traditional angle estimation method, this method reduces the influence of system error, cumulative error and external interference, has high precision, good stability, and quick response to target maneuvering, which has been significantly improved.

Description

technical field [0001] The invention belongs to the field of pattern recognition, and relates to a method for pattern recognition of electromyographic signals, in particular to a method for estimating continuous motion of joints based on a state-space electromyography model and an unscented Kalman filter. Background technique [0002] Surface Electromyography (sEMG) is a popular input signal source for human-computer interaction in cutting-edge science and technology. Action potential sequence, which has obvious characteristics, rich information, and simple and non-invasive acquisition, has become a hot research field in human-computer interaction technology. The research on surface EMG mainly focuses on the two processes of feature extraction and pattern recognition. The corresponding research results are relatively mature and can identify multiple discrete action categories. However, in the fields of rehabilitation medical robots and other fields, it is more often necess...

AI Technical Summary

Problems solved by technology

The model involves multiple physiological parameters, which is difficult to calculate and limited in practical application

HMM is the most commonly used muscle model for estimating continuous joint motion, but there are two problems: one is that HMM involves many complex physiological parameters that are difficult to identify, and the calculation burden is relatively large; the other is that HMM can directly calculate joint torque from sEMG signals, But if you need continuous joint motion estimation, you also need to calculate the motion state from the torque

This often introduces cumulative errors, reducing forecast accuracy

Images