STMI: Stiffness Estimation Method Based on Semg-Driven Model for Elbow Joint.
Cheng Shen,Zhongcai Pei,Weihai Chen,Zhongyi Li,Jianhua Wang,Jianbin Zhang,Jianer Chen
DOI: https://doi.org/10.1109/tim.2023.3300424
IF: 5.6
2023-01-01
IEEE Transactions on Instrumentation and Measurement
Abstract:Human arm exhibits superb maneuverability in performing various tasks by using its ability to actively adjust impedance parameters and interaction forces; therefore, identifying the joint stiffness of the arm during motion and transmitting them to the exoskeleton robots is the key to achieving flexible rehabilitation motion. In this article, we propose a stiffness estimation method for elbow joint, termed stiffness and torque mapping index (STMI), which establishes the relationship between joint torque, joint angle, and joint stiffness based on surface electromyography (sEMG). To improve joint angle identification accuracy, we propose a regression algorithm (RA), termed improved ant colony optimization generalized regression neural network (IACO-GRNN), and propose a time-domain descriptor (TDD) sEMG feature. Feature comparison experiments show that the TDD features (RMSE: 7.9792 +/- 0.0721; R2: 0.9257 +/- 0.0020) are superior among the popular features, such as mean absolute value (MAV), root mean square (rms), and waveform length (WL). The combination of TDD and IACO-GRNN achieves more impressive regression performance (RMSE: 8.6839 +/- 0.0084, R2: 0.9109 +/- 0.0002) than RA in joint angle recognition experiments, such as support vector regression (SVR), random forest regression (RFR), particle swarm optimization-GRNN (PSO-GRNN) and genetic algorithm-GRNN (GA-GRNN). The average RMSE of the joint stiffness estimated using STMI was 1.0146 Nm/rad. The proposed method can offer satisfactory dynamic elbow joint stiffness estimation only using sEMG signals, avoiding complex calibration of physiological data and additional sensing devices.