Control And Interactive Stiffness Identification Of Series Elastic Actuator

In the field of physical human-robot interaction, there is a tradeoff between human safety and actuator performance. How to design efficient and precise control algorithms to manipulate robots while protecting operators’ security has been one of the many difficulties. Recently, attention has been widespreadly paid on Series Elastic Actuator (SEA) due to its intrinsic safety features, but the introduction of elasticity increases the complexity of system dynamics, it is a challenge work to design reliable control algorithms of SEA.Based on 1DOF SEA experimental platform, this thesis carries out research on interactive security of human and SEA to make robots perform accurately and safely. The main contributions are as follows:1. Torque control method with inner velocity loop was improved. By adding noise observer, torque controller with velocity loop effectively decreased the influence of SEA time-varying parameters. The effectiveness, accuracy, rapidity and robustness of this method were verified by experiments.2. An adaptive compliance position control algorithm with variable learning rate was demonstrated. This algorithm adaptively estimated load dynamics of SEA and adjusted learning rate with Extended Kalman Filter (EKF). Experimental results shown that position tracking error was about 0.56°and it offered the desired safety.3. An stiffness identification approach of SEA interactive environment was proposed. By introducing slack variable, this approach estimated the interactive stiffness with parameter identification method of fading memory. Simulation results shown its great real-time performance and strong capability of disturbance rejection.