Motion intent recognition research is one of the key challenges in achieving human-robot collaboration in rehabilitation robots. In the traditional method, intention recognition is performed based on the complete sEMG, however, due to the muscle atrophy of stroke patients, the complete sequences are not captured at the early stage of rehabilitation, so in this paper, three sEMG segments of 1/2, 1/4, and 1/8 of three selected activity of daily living (ADL) movements of the upper limb are investigated, respectively, and comparing with the complete sEMG sequences, a novel method of motor intention prediction is proposed. In order to achieve optimal recognition accuracy and speed, the Kernel Extreme Learning Machine (KELM) algorithm optimized by the Sparrow Search Algorithm (SSA) algorithm was used for prediction. It was found that the SSA-KELM algorithm based on segmented sEMG has better recognition accuracy and recognition speed in each segment compared to other algorithms. The recognition accuracy in 1/8 sEMG segments is 98.4%, and the recognition time is 0.0102s, which shows how well the method works and what it means for rehabilitation robots working together with people.

In the rehabilitation training process for stroke patients, the level of excitement in the patient’s physiological state has a positive impact on the efficacy of the training. In order to improve patients’ initiative during training and prevent dependence on assistive systems, this study proposes an assist-as-needed control strategy based on a subjective intention decline model. The strategy primarily consists of two modules: a subjective intention decline control module and a limb movement assessment module. The subjective intention decline module collects surface electromyography (sEMG) data during patient training and optimizes support vector machine (SVM) using quantum particle swarm optimization (QPSO) algorithms to establish a subjective intention decline model. The limb movement assessment module collects information such as interaction force and position error during training and proposes a method for evaluating the motion state of the affected limb. This model combines traditional impedance control with a method for assessing limb movement and subjective status, automatically adjusting the level of assistive force on the affected limb in real time to enhance its active participation in tasks. Finally, we performed two verification experiments to assess the patient’s initiative in participating in the training. The experimental results show that the proposed method effectively reduced the average assist force by 65.66% for the traditional impedance control training system and effectively the average assist force by 35.2% for the control training system using only the assist force module based on force position information. At the same time, the accuracy of the subjective intention attenuation module established in the experiment to identify the fatigue level of the subjects reached 93.41%. Therefore, the proposed method effectively improves the initiative of trainers and also prevents patients from relying on the assist-as-needed control training system.

In recent years, the increasing number of patients with spinal cord injuries, strokes, and lower limb disabilities has led to the gradual development of rehabilitation-assisted exoskeleton robots. A critical aspect of these robots is their ability to accurately sense human movement intentions to achieve smooth and natural control. This paper describes research carried out on predicting the motion angles of human lower limb joints. Based on the design of a signal acquisition system for physiological muscle signals and inertial measurement unit (IMU) data, a hybrid neural network prediction model (QRTCN-BiLSTM) and a single neural network prediction model (QRBiLSTM) were constructed using quantile regression, temporal convolution network (TCN) and bidirectional long short-term memory network (BiLSTM), respectively. At the same time, 7-channel surface electromyographic signals (sEMG) and 12-channel IMU data from hip and knee joints were collected and input into the QRBiLSTM and QRTCN-BiLSTM models to unfold the training and analyze the comparison. The results show that the QRTCN-BiLSTM model can more accurately infer human movement intention and provide a more reliable and accurate prediction tool for human–robot interaction research in rehabilitation robotics.

The increasing utilization of upper limb rehabilitation robots in rehabilitation therapy has brought to light significant safety concerns regarding their mechanical structures and control systems. This study focuses on a six degrees of freedom (DOF) upper limb rehabilitation robot, which has been designed with an emphasis on safety through careful consideration of its mechanical structure and trajectory planning. Various parameter schemes for the shoulder joint angles were proposed, and the robotic arm’s structure was developed by analyzing the spatial motion trajectories of the shoulder joint motor. This design successfully achieves the objective of minimizing the installation space while maximizing the range of motion. Additionally, an enhanced artificial field method is introduced to facilitate the planning of self-collision avoidance trajectories for dual-arm movements. This approach effectively mitigates the risk of collisions between the robotic arm and the human body, as well as between the two robotic arms, during movement. The efficacy of this method has been validated through experimental testing.

Rigid–flexible coupled robots have problems such as vibration and elastic deformation caused by the flexibility of the members during the motion process, which significantly impacts the system’s motion accuracy and dynamics performance. To address the above problems, a dynamic modeling method based on a vector bond graph is proposed, and a multi-energy domain global dynamic model of Delta-type rigid–flexible coupled parallel robot considering rod flexibility is established. The coupled vibration of the control part and mechanical part of the system is analyzed, and model simulation is verified by 20-sim 4.4 software and ADAMS software (instructional version), which verifies the validity and reasonableness of the modeling method and provides a reference for the modeling of other rigid–flexible coupled systems with parallel systems in space.

To improve the accuracy of fatigue assessment in patients undergoing upper limb rehabilitation, this study proposes a lightweight fatigue detection algorithm based on ensemble learning and single-channel surface electromyography (sEMG) signals. Thirty healthy subjects without upper limb injuries or severe chronic diseases were recruited, and dynamic sEMG signals of the biceps brachii were collected during dumbbell bicep curls at a sampling frequency of 2048 Hz, yielding 6650 valid experimental samples. To address the class imbalance in the sEMG dataset, the SMOTETomek hybrid sampling algorithm was employed for data balancing. Three ensemble strategies—voting, stacking, and mean fusion—were integrated and combined with four different sets of base classifiers to construct a total of 12 fusion models, and the optimal model was identified through comparative screening. The experimental results demonstrated that, after SMOTETomek sample balancing, the LightGBM-LR-MLP soft voting ensemble model achieved the best overall performance, with the accuracy, recall, precision, and F1-score for both fatigue categories all exceeding 0.93. Multiple statistical analyses, including paired t-tests, effect sizes, and 95% confidence intervals, verified the reliability of the model selection and confirmed that the soft voting algorithm significantly outperformed the other comparison schemes. Machine learning can efficiently interpret dynamic biceps brachii sEMG signals; the proposed method improves the accuracy of fatigue recognition and provides an objective, quantitative fatigue reference index for upper limb rehabilitation training. It holds promise for assisting the dynamic adjustment of rehabilitation training intensity, thereby potentially reducing the risk of overtraining injuries and enhancing the safety of rehabilitation training.