No abstract

The surface electromyography (sEMG) technique is proposed for muscle activation detection and intuitive control of prostheses or robot arms. Motion recognition is widely used to map sEMG signals to the target motions. One of the main factors preventing the implementation of this kind of method for real-time applications is the unsatisfactory motion recognition rate and time consumption. The purpose of this paper is to compare eight combinations of four feature extraction methods (Root Mean Square (RMS), Detrended Fluctuation Analysis (DFA), Weight Peaks (WP), and Muscular Model (MM)) and two classifiers (Neural Networks (NN) and Support Vector Machine (SVM)), for the task of mapping sEMG signals to eight upper-limb motions, to find out the relation between these methods and propose a proper combination to solve this issue. Seven subjects participated in the experiment and six muscles of the upper-limb were selected to record sEMG signals. The experimental results showed that NN classifier obtained the highest recognition accuracy rate (88.7%) during the training process while SVM performed better in real-time experiments (85.9%). For time consumption, SVM took less time than NN during the training process but needed more time for real-time computation. Among the four feature extraction methods, WP had the highest recognition rate for the training process (97.7%) while MM performed the best during real-time tests (94.3%). The combination of MM and NN is recommended for strict real-time applications while a combination of MM and SVM will be more suitable when time consumption is not a key requirement.

Most studies about deep learning are based on neural network models, where many layers of parameterized nonlinear differentiable modules are trained by backpropagation. Recently, it has been shown that deep learning can also be realized by non-differentiable modules without backpropagation training called deep forest. The developed representation learning process is based on a cascade of cascades of decision tree forests, where the high memory requirement and the high time cost inhibit the training of large models. In this paper, we propose a simple yet effective approach to improve the efficiency of deep forest. The key idea is to pass the instances with high confidence directly to the final stage rather than passing through all the levels. We also provide a theoretical analysis suggesting a means to vary the model complexity from low to high as the level increases in the cascade, which further reduces the memory requirement and time cost. Our experiments show that the proposed approach achieves highly competitive predictive performance with significantly reduced time cost and memory requirement by up to one order of magnitude.

This paper presents a quantitative representation method for the upper-limb elbow joint angle using only electromyography (EMG) signals for continuous elbow joint voluntary flexion and extension in the sagittal plane. The dynamics relation between the musculotendon force exerted by the biceps brachii muscle and the elbow joint angle is developed for a modified musculoskeletal model. Based on the dynamics model, a quadratic-like quantitative relationship between EMG signals and the elbow joint angle is built using a Hill-type-based muscular model. Furthermore, a state switching model is designed to stabilize the transition of EMG signals between different muscle contraction motions during the whole movement. To evaluate the efficiency of the method, ten subjects performed continuous experiments during a 4-day period and five of them performed a subsequent consecutive stepping test. The results were calculated in real-time and used as control reference to drive an exoskeleton device bilaterally. The experimental results indicate that the proposed method can provide suitable prediction results with root-mean-square (RMS) errors of below 10° in continuous motion and RMS errors of below 10° in stepping motion with 20° and 30° increments. It is also easier to calibrate and implement.

The natural neuromuscular model has greatly inspired the development of control mechanisms in addressing the uncertainty challenges in robotic systems. Although the underpinning neural reaction of posture control remains unknown, recent studies suggest that muscle activation driven by the nervous system plays a key role in human postural responses to environmental disturbance. Given that the human calf is mainly formed by two muscles, this paper presents an integrated calf control model with the two comprising components representing the activations of the two calf muscles. The contributions of each component towards the artificial control of the calf are determined by their weights, which are carefully designed to simulate the natural biological calf. The proposed calf modelling has also been applied to robotic ankle exoskeleton control. The proposed work was validated and evaluated by both biological and engineering simulation approaches, and the experimental results revealed that the proposed model successfully performed over 92% of the muscle activation naturally made by human participants, and the actions led by the simulated ankle exoskeleton wearers were overall consistent with that by the natural biological response.INDEX TERMS Muscle stretch reflex, calf muscle activation, standing control, exoskeleton control.

No abstract

Robotic exoskeletons have emerged as effective rehabilitation and ability-enhancement tools, by mimicking or supporting natural body movements. The control schemes of exoskeletons are conventionally developed based on fixed torque-ankle state relationship or various human models, which are often lack of flexibility and adaptability to accurately address personalized movement assistance needs. This paper presents an adaptive control strategy for personalized robotic ankle exoskeleton in an effort to address this limitation. The adaptation was implemented by applying the experience-based fuzzy rule interpolation approach with the support of a muscle-tendon complex model. In particular, this control system is initialized based on the most common requirements of a ''standard human model,'' which is then evolved during its performance by effectively using the feedback collected from the wearer to support different body shapes and assistance needs. The experimental results based on different human models with various support demands demonstrate the power of the proposed control system in improving the adaptability, and thus applicability, of robotic ankle exoskeletons.INDEX TERMS Robotic ankle exoskeleton, muscle-tendon complex model, adaptive fuzzy rule interpolation, rehabilitation support.

This study presents an implementation of a continuous upper limb motion recognition method based on surface electromyography (sEMG) into control of an Upper Limb Exoskeleton Rehabilitation Device (ULERD). The raw sEMG hardly can be used directly as a reference control signals due to various influences. A feature extraction method, namely, an autoregressive algorithm, was thus applied to extract features of sEMG. The features of sEMG are usually used as switching signals to indicate whether activation happens. In this study, a continuous recognition is implemented using the coefficients of an AR model because of the coefficients’ characteristics of fitting the trend in sEMG signals. Because of the low signal-to-noise ratio of sEMG, the optimal order of the AR model was calculated based on the Akaike Information Criterion for a good fit to raw signals. Back-propagation neural networks were then trained using coefficients to recognize motion. Recognition results were used as the control source for the rehabilitation device. Experimental results showed that this method is effective for obtaining a control source through raw sEMG signals derived fromthe unaffected arm for motor control of a ULERD equipped on the affected armduring bilateral rehabilitation in real-time.