High-Precision Tracking Control of a Soft Dielectric Elastomer Actuator With Inverse Viscoelastic Hysteresis Compensation

In this paper, we present a new control approach for high-precision tracking control of a soft dielectric elastomer actuator (DEA) with inverse viscoelastic hysteresis compensation. To this end, we first investigate the viscoelastic response of the DEA and divide it into transition region and stable region. Then, the viscoelastic response is characterized by creep and hysteresis effects according to the different features of the two regions. Finally, a two-level tracking control approach is developed as follows: a direct inverse hysteresis compensation controller with a phenomenological hysteresis model is designed for the viscoelastic hysteresis description and compensation, and a conventional proportional-integral feedback controller is combined to compensate for the model uncertainty and creep effect. To verify the effectiveness of the developed tracking control approach, several experiments are conducted with various reference sinusoidal trajectories. Experimental results show that: when the frequency of the trajectory is within the range of 0.1 to 1 Hz, the maximum tracking error and the root-mean-square error decrease from 40.63% to 3.95% and 28.38% to 1.86%, respectively. This paper is the first attempt to achieve high-precision tracking control of soft DEAs by combining a phenomenological-model feedforward compensator and a feedback control law for the viscoelastic compensation, which may accelerate the practical applications of DEAs to soft robots.

I. Introduction

Due to the capability of large deformation and shape change, dielectric elastomer actuators (DEAs) show promising applications in the field of soft robotics [1]–[3]. In general, a DEA consists of an electroactive polymer membrane sandwiched by compliance electrodes [4]. When a high voltage is applied to the electrodes, the Maxwell stress squeezes the membrane, so that it will expand in area and decrease in thickness to keep a constant volume [5]. On the basis of this working principle, different DEAs have been invented, including rectilinear [6], [7], rotation [8], bending [9], and ballooning [10], [11] motions. During the past decade, there are diverse achievements on design of DEA-driven soft robots, for instance, a soft fish [12] and a soft printable hexapod robot [13]. As one type of actuators, DEAs not only need to satisfy the functional movements of soft robots, but also acquire to achieve accurate positioning control for practical applications.