Abstract: Variable Stiffness Actuation (VSA) is an efficient, safe, and robust actuation technology for bionic robotic joints that have emerged in recent decades. By introducing a variable stiffness elastomer in the actuation system, the mechanical-electric energy conversion between the motor and the load could be adjusted on-demand, thereby improving the performance of the actuator, such as the peak power reduction, energy saving, bionic actuation, etc. At present, the VSA technology has achieved fruitful research results in designing the actuator mechanism and the stiffness adjustment servo, which has been widely applied in articulated robots, exoskeletons, prostheses, etc. However, how to optimally control the stiffness of VSAs in different application scenarios for better actuator performance is still challenging, where there is still a lack of unified cognition and viewpoints. Therefore, from the perspective of optimal VSA performance, this paper first introduces some typical structural design and servo control techniques of common VSAs and then explains the methods and applications of the Optimal Variable Stiffness Control (OVSC) approaches by theoretically introducing different types of OVSC mathematical models and summarizing OVSC methods with varying optimization goals and application scenarios or cases. In addition, the current research challenges of OVSC methods and possible innovative insights are also presented and discussed in-depth to facilitate the future development of VSA control.

Key words: Bionic robotic joint , · Variable stiffness actuator , · Optimal control , · Robotics , · Controller design