Adaptive estimation and control of uncertain series elastic actuator systems

For series elastic actuator (SEA) used for robotics, there are many uncertain parameters in their models, which are difficult to obtain based on the design configurations. In this paper, to solve the problem of estimating the unknown model parameters of the SEAs online, we suggest a new adaptive law. Parameter estimation error obtained by filtering input and output measurements to drive the adaptive law, so that can rigorously prove the performance of the exponential convergence of the parameter estimation error. Moreover, in order to synchronously parameter estimation and motion tracking, the idea of using the estimation errors in the adaptive law is further adopted by an adaptive control integration. Finally, the feasibility of the proposed methods is verified by numerical simulations.