[Research on mode adjustment control strategy of upper limb rehabilitation robot based on fuzzy recognition of interaction force].

In the process of robot-assisted training for upper limb rehabilitation, a passive training strategy is usually used for stroke patients with flaccid paralysis. In order to stimulate the patient's active rehabilitation willingness, the rehabilitation therapist will use the robot-assisted training strategy for patients who gradually have the ability to generate active force. This study proposed a motor function assessment technology for human upper-limb based on fuzzy recognition on interaction force and human-robot interaction control strategy based on assistance-as-needed. A passive training mode based on the calculated torque controller and an assisted training mode combined with the potential energy field were designed, and then the interactive force information collected by the three-dimensional force sensor during the training process was imported into the fuzzy inference system, the degree of active participation σ was proposed, and the corresponding assisted strategy algorithms were designed to realize the adaptive adjustment of the two modes. The significant correlation between the degree of active participation σ and the surface electromyography signals (sEMG) was found through the experiments, and the method had a shorter response time compared to a control strategy that only adjusted the mode through the magnitude of interaction force, making the robot safer during the training process.

Using different matrix factorization approaches to identify muscle synergy in stroke survivors.

Over the past several decades, many scholars have investigated muscle synergy as a promising tool for evaluating motor function. However, it is challenging to obtain favorable robustness using the general muscle synergy identification algorithms, namely non-negative matrix factorization (NMF), independent component analysis (ICA), and factor analysis (FA). Some scholars have proposed improved muscle synergy identification algorithms to overcome the shortcomings of these approaches, such as singular value decomposition NMF (SVD-NMF), sparse NMF (S-NMF), and multivariate curve resolution-alternating least squares (MCR-ALS). However, performance comparisons of these algorithms are seldom conducted. In this study, experimental electromyography (EMG) data collected from healthy individuals and stroke survivors were applied to assess the repeatability and intra-subject consistency of NMF, SVD-NMF, S-NMF, ICA, FA, and MCR-ALS. MCR-ALS presented higher repeatability and intra-subject consistencies than the other algorithms. More synergies and lower intra-subject consistencies were observed in stroke survivors than in healthy individuals. Thus, MCR-ALS is considered a favorable muscle synergy identification algorithm for patients with neural system disorders.

MCR-ALS-based muscle synergy extraction method combined with LSTM neural network for motion intention detection.

Introduction: The time-varying and individual variability of surface electromyographic signals (sEMG) can lead to poorer motor intention detection results from different subjects and longer temporal intervals between training and testing datasets. The consistency of using muscle synergy between the same tasks may be beneficial to improve the detection accuracy over long time ranges. However, the conventional muscle synergy extraction methods, such as non-negative matrix factorization (NMF) and principal component analysis (PCA) have some limitations in the field of motor intention detection, especially in the continuous estimation of upper limb joint angles. Methods: In this study, we proposed a reliable multivariate curve-resolved-alternating least squares (MCR-ALS) muscle synergy extraction method combined with long-short term memory neural network (LSTM) to estimate continuous elbow joint motion by using the sEMG datasets from different subjects and different days. The pre-processed sEMG signals were then decomposed into muscle synergies by MCR-ALS, NMF and PCA methods, and the decomposed muscle activation matrices were used as sEMG features. The sEMG features and elbow joint angular signals were input to LSTM to establish a neural network model. Finally, the established neural network models were tested by using sEMG dataset from different subjects and different days, and the detection accuracy was measured by correlation coefficient. Results: The detection accuracy of elbow joint angle was more than 85% by using the proposed method. This result was significantly higher than the detection accuracies obtained by using NMF and PCA methods. The results showed that the proposed method can improve the accuracy of motor intention detection results from different subjects and different acquisition timepoints. Discussion: This study successfully improves the robustness of sEMG signals in neural network applications using an innovative muscle synergy extraction method. It contributes to the application of human physiological signals in human-machine interaction.

Fuzzy Adaptive Passive Control Strategy Design for Upper-Limb End-Effector Rehabilitation Robot.

Robot-assisted rehabilitation therapy has been proven to effectively improve upper-limb motor function in stroke patients. However, most current rehabilitation robotic controllers will provide too much assistance force and focus only on the patient's position tracking performance while ignoring the patient's interactive force situation, resulting in the inability to accurately assess the patient's true motor intention and difficulty stimulating the patient's initiative, thus negatively affecting the patient's rehabilitation outcome. Therefore, this paper proposes a fuzzy adaptive passive (FAP) control strategy based on subjects' task performance and impulse. To ensure the safety of subjects, a passive controller based on the potential field is designed to guide and assist patients in their movements, and the stability of the controller is demonstrated in a passive formalism. Then, using the subject's task performance and impulse as evaluation indicators, fuzzy logic rules were designed and used as an evaluation algorithm to quantitively assess the subject's motor ability and to adaptively modify the stiffness coefficient of the potential field and thus change the magnitude of the assistance force to stimulate the subject's initiative. Through experiments, this control strategy has been shown to not only improve the subject's initiative during the training process and ensure their safety during training but also enhance the subject's motor learning ability.