Research On Optimal Stiffness Of Nonlinear Stiffness Actuators

Passive compliant actuators are widely used in human robot interaction(HRI)due to its energy storage,impact tolerance and safety performance,such as rehabilitation robots and medical robots et al.Classical series elastic actuators(SEAs)realize complaint human robot interaction through linear spring,however,its performance is limited by the constant stiffness of the spring.Actuators with high stiffness can realize high dynamic performance like control bandwidth,but high stiffness decreases the compliant performance like torque resolution and safety.Whereas actuators with low stiffness have high compliant performance.Variable stiffness actuators(VSAs)with decoupled stiffness and load can realize a tradeoff between the two contradictory performance in certain extent,but most VSAs adjust its stiffness through independent stiffness adjusting structure.The additional stiffness adjusting motors complicate the mechanical structure of the actuators.In HRI,Nonlinear stiffness actuators with coupled stiffness and load not only can realize a tradeoff between the dynamic performance and the compliant performance,but also can simplify and compact the mechanical structure of the actuators.This paper focus on the stiffness design and human robot interactive performance of the nonlinear stiffness actuators and firstly,introduce a load-dependent nonlinear stiffness actuator(LDNSA)which is designed according to the interaction strategy “large load corresponds to high stiffness and small load corresponds to low stiffness”.In addition,based on the system of the LDNSA,this paper researches the control performance of nonlinear stiffness actuators with different loads(stiffness)through simulation and experimental results.That provides a reference for stiffness selection of complaint actuators under different loads.Afterward,considering the limitation of proportion-differentiation(PD)controller which is usually used in the control of compliant actuators,this paper proposed an auto-tuning feedback controller based on the nonlinear stiffness of nonlinear stiffness actuators to adapt the variable stiffness.And the controller can guarantee that the nonlinear stiffness actuators are well damped in any stiffness situation which improves the control performance of the actuators including bandwidth and torque response stability et al.Finally,aiming the urgent problem needed to be solved of designing an optimal stiffness profile to realize an optimal tradeoff between the dynamic performance like control bandwidth and complaint performance like torque resolution,this paper proposed a quantized comprehensive performance index-“perceptivity and responsiveness index(PRI)” including control bandwidth and torque resolution to evaluate the comprehensive interactive performance.Minimizing the index enables compliant actuators achieve an optimal stiffness profile and corresponding feedback control gains.And simulations and experiments results perform that the LDNSA with optimal stiffness has best control bandwidth performance with high complaint performance like torque resolution in small load and higher control bandwidth and relative high torque resolution performance in large load.In addition,the LDNSA with optimal stiffness can realize optimal comprehensive performance under any load situation.