Neuromusculoskeletal model self-calibration for on-line sequential bayesian moment estimation

Abstract

The results show the feasibility of neuromusculoskeletal model self-calibration. This study will contribute to a better understanding of the generalization of muscle models for subject-specific rehabilitation therapies. Moreover, this work is very promising for rehabilitation devices such as electromyography-driven exoskeletons or prostheses.

Targeted Achilles Tendon Training and Rehabilitation Using Personalized and Real-Time Multiscale Models of the Neuromusculoskeletal System

“…Model calibration is fundamental to obtain physiologically plausible muscle-tendon forces (Hoang et al, 2018) but at the current stage this process is still computationally intensive and must be performed offline. Online calibration procedures, such as Bueno and Montano (2017), should be implemented in the future to minimize idle times. Finally, as the FE method is computationally intensive, a surrogate FE model was used to enable real-time estimation of Achilles tendon strain.…”

Feature Extraction and Classification of Lower Limb Motion Based on sEMG Signals

“…However, the sEMG signal of the lower limb has complex nonlinearity, strong coupling, and dynamic time-variation, resulting in the lack of stability of the neural network model. In recent years, the hot research direction to solve nonlinear estimation is to use the unscented Kalman filter (UKF) algorithm [30,31]. Therefore, we used the UKF algorithm to dynamically optimize the weights of the neural network, enhancing the adaptive ability and improving the accuracy of the model.…”

“…Therefore, the accuracy of the model needs to be improved. In recent years, the unscented Kalman filter is a new filter estimation algorithm, which has been applied by researchers to adaptive control of exoskeleton [30,31]. Therefore, we applied the UKF algorithm to lower limb motion classification and used the scale correction to improve the UKF filtering accuracy.…”