To improve the life quality of forearm amputees, prosthetic hands with high accuracy, and robustness are necessary. The application of surface electromyography (sEMG) signals to control a prosthetic hand is challenging. In this study, we proposed a time-domain CNN model for the regression prediction of joint angles in three degrees of freedom (3-DOFs, include two wrist joint motion and one finger joint motion), and five-fold cross validation was used to evaluate the correlation coefficient (CC). The CC value results of wrist flexion/extension motion obtained from 10 participants was 0.87–0.92, pronation/supination motion was 0.72–0.95, and hand grip/open motion was 0.75–0.94. We backtracked the fully connected layer weights to create a geometry plot for analyzing the motion pattern to investigate the learning of the proposed model. In order to discuss the daily updateability of the model by transfer learning, we performed a second experiment on five of the participants in another day and conducted transfer learning based on smaller amount of dataset. The CC results improved (wrist flexion/extension was 0.90–0.97, pronation/supination was 0.84–0.96, hand grip/open was 0.85–0.92), suggesting the effectiveness of the transfer learning by incorporating the small amounts of sEMG data acquired in different days. We compared our CNN-based model with four conventional regression models, the result illustrates that proposed model significantly outperforms the four conventional models with and without transfer learning. The offline result suggests the reliability of the proposed model in real-time control in different days, it can be applied for real-time prosthetic control in the future.

Recently, many muscle synergy-based human motion prediction models and algorithms have been proposed. In this study, the muscle synergies extracted from electromyography (EMG) data were used to construct a musculoskeletal model (MSM) to predict the joint angles of the wrist, thumb, index finger, and middle finger. EMG signals were analyzed using independent component analysis to reduce signal noise and task-irrelevant artifacts. The weights of each independent component (IC) were converted into a heat map related to the motion pattern and compared with human anatomy to find a different number of ICs matching the motion pattern. Based on the properties of the MSM, non-negative matrix factorization was used to extract muscle synergies from selected ICs that represent the extensor and flexor muscle groups. The effects of these choices on the prediction accuracy was also evaluated. The performance of the model was evaluated using the correlation coefficient (CC) and normalized root-mean-square error (NRMSE). The proposed method has a higher prediction accuracy than those of traditional methods, with an average CC of 92.0% and an average NRMSE of 10.7%.

No abstract

In the past decade, multiple anthropomorphic prosthetic hands have been developed to replace the role of human hands. Prostheses should not only replace the functions of human hands in functionality but also replicate human hands in appearance and sense of bodybelonging intuitively. Human fingers have very delicate and complex structures, and it is these complex structures that make our hands dexterity. This study proposes a design based on the anatomical characteristics of the human hand. The proposed design replicates human fingers from bones, ligaments, extensor hoods, and extensor mechanism of tendon, intended to develop a prosthesis that has the same flexibility and appearance as human hand. To evaluate the performance of the proposed prosthetic in daily life, we conducted grasping experiments on common objects. It is successfully proved that the proposed design helps to improve the grasping performance of the artificial hand and has a natural appearance. In this paper, our design succeeds to improve the grasping performance of the artificial hand and gain natural appearance.