2 February 2021 in IEEE Access
The surface electromyogram (sEMG) contains a wealth of motion information, which can reflect user's muscle motion intentions. The decoding based on sEMG has been widely used to provide a safe and effective human-computer interaction (HCI) method for neural prosthesis and exoskeleton robot control. The motor intention decoding based on low sampling frequency sEMG may promote the application of wearable low-cost EMG sensors in HCI. Therefore, a motor intention decoding scheme suitable for low frequency EMG signal is proposed in this paper, that is, transfer learning based on Alexnet. Moreover, the effects of different feature extraction methods and data augmentation with Gaussian white noise are fully analyzed. The proposed algorithm is evaluated with the NinaPro database 5. The highest accuracy can reach 70.4%±4.36% in 53 gestures identification of 10 subjects. Some classical machine learning algorithms such as support vector machine (SVM), linear discriminant analysis (LDA) and K Nearest Neighbor (KNN) are chosen to make comparison, where the SVM with Gaussian kernel function reaches to the maximum accuracy of 67.98%±4.56%. Two-way variance results show significant differences between each other. The experiment results show that the transfer learning is effective for decoding low-frequency sEMG for a large number of gestures.
24 November 2020 in IEEE Access
Joint torque estimation is of great significance for the research and clinical application of intelligent rehabilitation technology. This paper proposes a closed-loop model for joint torque estimation based on surface electromyography (sEMG). Combined with the physiological characteristics of muscle activation, a nonlinear autoregressive with eXogenous inputs(NARX) neural network model for joint torque estimation based on sEMG signals is established. In order to solve the drift phenomenon of torque estimation, a state-space framework is constructed by regarding NARX neural network based torque estimation model as state model and developing a measurement model by the easily measured joint angle signals. With the built state-space model, the extended Kalman filter (EKF) is used to realize the closed-loop filtering of the estimated torque. In order to test the accuracy of the proposed closed-loop joint torque estimation, 8 volunteers were recruited to perform elbow joint isotonic motion experiments under four kinds of loads. The test results show that the average normalized root mean square error (NRMSE) between the estimated values of closed-loop model and measurement values of all subjects under load-dependent, multi-load and load-independent tests are 0.1080±0.0411, 0.1326±0.0494 and 0.1674±0.0661 respectively, which is significant better than the results of the open-loop model (0.2694±0.1584 (p < 10-6), 0.2499±0.1326 (p < 10-6) and 0.3435±0.2061 (p < 10-4)). The presented closed-loop model combines offline modeling and online filtering to achieve online estimation of joint torque, which ameliorates the problem that the estimated torque in the open-loop model deviates greatly from the actual values and improves the accuracy of joint torque estimation.