Feedforward Control of Twisted and Coiled Polymer Actuator Based on a Macroscopic Nonlinear Model Focusing on Energy

Abstract

This letter proposes a feedforward (FF) controller for the twisted and coiled polymer actuator (TCPA), the so-called fishing line/sewing thread artificial muscle, based on its macroscopic nonlinear model. TCPA is an artificial muscle, and it contracts up to 50% in response to heating. Although several methods considering Joule heating as the input have been proposed to control the TCPA displacement, the FF controller based on the linear model shows the nonnegligible offset error. Therefore, the FF controller should be designed based on a nonlinear model, e.g., the model constructed by the authors. However, computing the inverse of the nonlinear model for the FF controller is difficult. Even if the inverse can be computed under static conditions, the problem of how to compensate the dynamics still remains. For this problem, we compensate the dynamics based on the idea of the Hammerstein model, namely we divide the inverse dynamics into the nonlinear transformation part and linear dynamics part. The former part consists of a converter from the displacement to temperature and a ratio adjuster for the squared voltage. Meanwhile, the latter is represented by the linear model of TCPA temperature. Through experiments, it is verified that the proposed FF controller can reduce the offset error of the displacement compared with the linear controller.