Characterization Analysis and Stiffness Estimation of Variable Stiffness Actuator for Elbow Exoskeleton

Exoskeleton actuator with variable stiffness will hold compliant characteristic, which is more ergonomic as well as a safe interaction between robot and human. Therefore, a variable stiffness actuator was proposed. It can change the structural configuration and preload to achieve variable stiffness. Meanwhile, it can be integrated into the elbow exoskeleton to meet the needs of miniaturization and lightweight. Then, the relationship between the FMG signal and the elbow joint stiffness of human-robot interaction was investigated. Machine learning methods were conducted to obtain a stiffness prediction model for real-time estimation. Finally, an elbow exoskeleton was developed. Furthermore, elbow flexion and extension trajectory tracking tests were carried out to evaluate the characterization of the elbow exoskeleton.