Objective: Humans are able to localize the source of a sound. This enables them to direct attention to a particular speaker in a cocktail party. Psycho-acoustic studies show that the sensory cortices of the human brain respond to the location of sound sources differently, and the auditory attention itself is a dynamic and temporally based brain activity. In this work, we seek to build a computational model which uses both spatial and temporal information manifested in EEG signals for auditory spatial attention detection (ASAD). Methods: We propose an end-to-end spatiotemporal attention network, denoted as STAnet, to detect auditory spatial attention from EEG. The STAnet is designed to assign differentiated weights dynamically to EEG channels through a spatial attention mechanism, and to temporal patterns in EEG signals through a temporal attention mechanism. Results: We report the ASAD experiments on two publicly available datasets. The STAnet outperforms other competitive models by a large margin under various experimental conditions. Its attention decision for 1-second decision window outperforms that of the state-of-the-art techniques for 10-second decision window. Experimental results also demonstrate that the STAnet achieves competitive performance on EEG signals ranging from 64 to as few as 16 channels. Conclusion: This study provides evidence suggesting that efficient low-density EEG online decoding is within reach. Significance: This study also marks an important step towards the practical implementation of ASAD in real life applications.

Robot-assisted rehabilitation is a growing field that can provide an intensity, quality, and quantity of treatment that exceed therapist-mediated rehabilitation. Several control algorithms have been implemented in rehabilitation robots to develop a patient-cooperative strategy with the capacity to understand the intention of the user and provide suitable rehabilitation training. In this paper, we present an upper-limb motion pattern recognition method using surface electromyography (sEMG) signals with a support vector machine (SVM) to control a rehabilitation robot, ReRobot, which was built to conduct upper-limb rehabilitation training for post-stroke patients. For poststroke rehabilitation training using the ReRobot, the upper-limb motion of the patient's healthy side is first recognized by detecting and processing the sEMG signals; then, the ReRobot assists the impaired arm in conducting mirror rehabilitation therapy. To train and test the SVM model, five healthy subjects participated in the experiments and performed five standard upper-limb motions, including shoulder flexion, abduction, internal rotation, external rotation, and elbow joint flexion. Good accuracy was demonstrated in experimental results from the five healthy subjects. By recognizing the model motion of the healthy side, the rehabilitation robot can provide mirror therapy to the affected side. This method can be used as a control strategy of upper-limb rehabilitation robots for self-rehabilitation training with stroke patients.

With the rapid development of deep learning, more and more deep learning-based motor imagery electroencephalograph (EEG) decoding methods have emerged in recent years. However, the existing deep learning-based methods usually only adopt the constraint of classification loss, which hardly obtains the features with high discrimination and limits the improvement of EEG decoding accuracy. In this paper, a discriminative feature learning strategy is proposed to improve the discrimination of features, which includes the central distance loss (CD-loss), the central vector shift strategy, and the central vector update process. First, the CD-loss is proposed to make the same class of samples converge to the corresponding central vector. Then, the central vector shift strategy extends the distance between different classes of samples in the feature space. Finally, the central vector update process is adopted to avoid the non-convergence of CD-loss and weaken the influence of the initial value of central vectors on the final results. In addition, overfitting is another severe challenge for deep learning-based EEG decoding methods. To deal with this problem, a data augmentation method based on circular translation strategy is proposed to expand the experimental datasets without introducing any extra noise or losing any information of the original data. To validate the effectiveness of the proposed method, we conduct some experiments on two public motor imagery EEG datasets (BCI competition IV 2a and 2b dataset), respectively. The comparison with current state-of-the-art methods indicates that our method achieves the highest average accuracy and good stability on the two experimental datasets.

Powered exoskeletons are used to assist human walking, but it is inconvenient for a user to carry a bulky energy-supply system. During walking, human lower limbs accelerate and decelerate alternatively, during which period the human body does positive and negative work, respectively. Therefore, if the negative work performed during deceleration can be harnessed using some assisting device to then assist the acceleration movement of the lower limb, the total metabolic cost of the human body during walking can be reduced. In this work, an unpowered flexible lower-limb exoskeleton is proposed; it is worn in parallel to the lower limbs to assist human walking without consuming external electrical power. The flexible exoskeleton imitates the physiological structure of a human lower limb, consisting of elastic and damping components. When worn on the lower limb, the exoskeleton can partly replace the function of the lower limb muscles, which can scavenge kinetic energy during lower limb deceleration to assist the acceleration movement, so that the biomechanical power consumption can be reduced during walking. In addition to the walking assistance, the generator in the exoskeleton, serving as a damping component, can harvest kinetic energy to produce electricity for powering wearable electronics. A prototype of the flexible exoskeleton was developed, and experiments were carried out to validate the effectiveness. The experiments showed that the lower-limb exoskeleton could reduce the metabolic cost by 3.12% at the walking speed of 4.5 km/h, and a maximum electrical power of 6.47 W was generated at the walking speed of 5.1 km/h.

Background: Compensatory movements are commonly employed by stroke survivors during seated reaching and may have negative effects on their long-term recovery. Detecting compensation is useful for coaching the patient to reduce compensatory trunk movements and improving the motor function of the paretic arm. Sensor-based and camera-based systems have been developed to detect compensatory movements, but they still have some limitations, such as causing object obstructions, requiring complex setups and raising privacy concerns. To overcome these drawbacks, this paper proposes a compensatory movement detection system based on pressure distribution data and is unobtrusive, simple and practical. Machine learning algorithms were applied to classify compensatory movements automatically. Therefore, the purpose of this study was to develop and test a pressure distribution-based system for the automatic detection of compensation movements of stroke survivors using machine learning algorithms. Methods: Eight stroke survivors performed three types of reaching tasks (back-and-forth, side-to-side, and up-and-down reaching tasks) with both the healthy side and the affected side. The pressure distribution data were recorded, and five features were extracted for classification. The k-nearest neighbor (k-NN) and support vector machine (SVM) algorithms were applied to detect and categorize the compensatory movements. The surface electromyography (sEMG) signals of nine trunk muscles were acquired to provide a detailed description and explanation of compensatory movements. Results: Cross-validation yielded high classification accuracies (F1-score>0.95) for both the k-NN and SVM classifiers in detecting compensation movements during all the reaching tasks. In detail, an excellent performance was achieved in discriminating between compensation and noncompensation (NC) movements, with an average F1-score of 0.993. For the multiclass classification of compensatory movement patterns, an average F1-score of 0.981 was achieved in recognizing the NC, trunk lean-forward (TLF), trunk rotation (TR) and shoulder elevation (SE) movements. Conclusions: Good classification performance in detecting and categorizing compensatory movements validated the feasibility of the proposed pressure distribution-based system. Reliable classification accuracy achieved by the machine learning algorithms indicated the potential to monitor compensation movements automatically by using the pressure distribution-based system when stroke survivors perform seated reaching tasks.

The continuous control of rehabilitation robots based on surface electromyography (sEMG) is a natural control strategy that can ensure human safety and ease the discomfort of human-machine coupling. However, current models for estimating movement of the upper limb focus on two dimensions movement, and models of three dimensions movement are too complex. In this paper, a simple-structure temporal information-based model for upper limb motion was proposed. An experiment of the multijoint motion of the upper limb was carried out. We studied the touching motion and the compound task motion of the upper limbs. The touch motor task consists of three tasks, namely, shoulder abduction, shoulder forward bend and finger-nose. The compound tasks include driving and fetching objects. Three-dimensional upper limb movement data and sEMG signals of seven upper limb muscles were recorded from seven healthy subjects. Model training was carried out after data preprocessing and feature extraction. Then, 120 s of upper limb motion data was used to verify the performance of the model proposed in this paper. The estimated accuracy of the model for the touch tasks was 0.9171, and it was 0.8109 for compound tasks. Compared to the multilayer perception (MLP) model, a 13.57% reduction in the root-mean-square error (RSME) was observed. The results show that the model has good accuracy for estimating the angular motion of the upper limb and that it has the potential to be applied for three-dimensional motion control in an upper limb mirror-image therapy rehabilitation robot.