Feiyan Min, Xianzhe Chen, Yujie Lin, Yingying She, Gao Wang, Qing Guo
A new$L_{1}$adaptive control (L1AC) method with bound-gain-forgetting (BGF) estimator is proposed in floating-base series elastic actuators (SEAs) to enhance both transient and steady-state force control performance. The transient performance is evaluated by the output force overshoot, the convergence speed of tracking error, and the chatters effect. Meanwhile, the steady-state performance is evaluated based on static error. First, the controller stability together with the transient performance are analyzed in the controller design. Then, the BGF-based estimator is adopted to achieve a time-varying adaptation for environmental stiffness, and improve the robustness and tracking speed of the SEA. Finally, the proposed controller performances are compared with the resonance ratio control and the model reference adaptive control in experimental verification. It indicates that the L1AC results has not any overshoot and less steady-state error than the other two methods. Furthermore, the chatter amplitude of the spring deformation is only in 0.2$^\circ$during force control.