A Broad Learning System Based Method for Estimating Force in Human-Robot Interaction

Thanks to its significance in the physical human-robot interaction (pHRI), the force estimation has aroused increasing research interest in recent years. In this paper, a novel method based on broad learning system (BLS) is proposed to estimate the interaction force by using surface electromyography (sEMG). Attempting to extract useful sEMG information, there exists necessity to extract the features of sEMG and utilize the features as the input vector of the proposed model. The measured arm force is set as the reference input of the sEMG-force model. The proposed method is verified via an experiment test in the case where the robot cooperates with the subject to accomplish the drawing task. During the experimental test, the raw sEMG signal of the subject and human-robot interaction (HRI) force is collected. Moreover, the root mean squared error (RMSE) and mean absolute percentage (MAPE) are implemented to evaluate the performance of our proposed method. The experimental results confirm that the proposed method with hybrid feature dominates the one with the single feature such as root mean square (RMS), mean absolute average (MAV), and integrated electromyography (IEMG). Thus, it could be conclusive that our proposed method would be considered as a vital alternative in the field of HRI force estimation.