Design of an Auto-Tuning Feedback Controller Based on the Stiffness of Nonlinear Stiffness Actuators

Abstract

Variable stiffness actuators (VSAs) realize a compromise between bandwidth and safety through stiffness tuning. Many VSAs often apply classical PD controller to realize a fast and accurate output response in human robotic interaction (HRI). But for variable stiffness actuators, there exists some limitations like that the PID parameters called proportional gain, integration gain and derivation gain have to be tuned manually, and the parameters are fixed in a certain response which contains different stiffness values. Therefore, the PID parameters we set are may not a very appropriate choice for all stiffness values which lead to a weakening of the control performance. One of the solutions is tuning the PID parameters along with the variable stiffness. So, an auto-tuning feedback controller is proposed to adjust the PID parameters along with the change of a load-dependent nonlinear stiffness actuators (LDNSA) stiffness. At the same time, a complete simulation was then implemented on MATLAB/Simulink to make a comparison with the classical PID feedback controller. And experiments are also conducted to evaluate the feasibility of the proposed feedback controller. At last, simulated and experimental results present the proposed controller can perform well in the control of nonlinear stiffness actuators.