Proceedings of the 8th World Congress on Intelligent Control and Automation July 6-9 2010, Jinan, China 978-1-4244-6712-9/10/$26.00 ©2010 IEEE Hysteresis Compensation of a Piezoactuated XY Micropositioning System Based on Disturbance Observer ∗ Qingsong Xu and Yangmin Li Department of Electromechanical Engineering, Faculty of Science and Technology University of Macau Av. Padre Tom´ as Pereira S.J., Taipa, Macao SAR, China {qsxu, ymli}@umac.mo Abstract—In this paper, a two-loop controller is proposed to suppress the hysteresis and to achieve a precise motion tracking control for a piezo-driven XY parallel micropositioning system. Specifically, an inner-loop disturbance observer (DOB) is employed to tolerate the unmodeled hysteresis which is treated as a disturbance to the nominal plant model of the system, and an outer-loop feedback controller is used to compensate for the remaining nonlinearities and uncertainties. The effectiveness of the proposed controller over the sole PID controller is demonstrated through experimental studies. Results show that the DOB-based two-loop control can reduce the hysteresis to a negligible level and lead to a motion tracking with submicron accuracy, which provides a sound base of practical control of the micropositioning system for micro/nano scale manipulation. Index Terms—Micro/nano-positioning; piezoelectric actuation; hysteresis compensation; disturbance observer. I. I NTRODUCTION Robotic microhandling and self-assembly are two ap- proaches to handle objects in the micro world. Both methods have their advantages and disadvantages. Based on the prin- ciple of minimum potential energy (using gravity, capillary forces, or electrostatic forces, etc., as the driving potential), self-assembly has a good throughput due to a parallel process and is appropriate to mass production of simple micro struc- tures. Even so, self-assembly has poor flexibility and dexterity when compared to robotic microhandling. Robotic microhan- dling is capable of performing very complex handling tasks, such as MEMS assembly, various biological applications, and so on. The handling can be conducted in closed loop with position and force sensing feedback. As core component of robotic microhandling systems, micropositioning devices are dedicated to high accuracy positioning in microhandling appli- cations. A great number of micropositioning stages have been designed and proposed to deliver various types of motions in the literature. Most of the stages employ a flexure-based * This work was supported partially by Macao Science and Technology Development Fund under Grant 016/2008/A1 and the Research Committee of University of Macau under Grant UL016/08-Y2/EME/LYM01/FST to Y. M. Li (Corresponding author) . mechanism and piezoelectric actuators (PZTs) to produce a submicron or subnanometer resolution positioning. In addi- tion to vacuum compatibility, flexure mechanism eliminates clearance, backlash, friction, and lubrication requirement for the device. Moreover, PZT is capable of positioning with (sub)nanometer resolution, large blocking force, high stiff- ness, and rapid response characteristics. Nevertheless, PZT introduces nonlinearity into the system mainly due to its hys- teresis property occurring at the voltage-driven strategy, which will attenuate the positioning accuracy of the manipulator if it is not carefully treated. An intuitive approach to overcome the nonlinearity is to model the hysteresis and construct an inverse-based feed- forward compensator. Several hysteresis models are avail- able in the literature, such as the Preisach model, Maxwell model, Duhem model, Prandtl-Ishlinskii model, and Bouc- Wen model, etc. However, the nonlinearities can not be totally eliminated by the feedforward compensation solely in most practical cases. Thus, another further approach is to combine the feedforward with an additional feedback control to suppress the remaining nonlinearities as well as the creep and drift effects induced by the PZT [1]. In view of the fact that the establishment and identification of a hysteresis model will be a complicated procedure which results in a time consuming work for a controller design process, an alternative method to deal with the nonlinearity is to treat the hysteresis as an unmodeled uncertainty or a disturbance and then construct a robust controller to compensate it. For instance, the H ∞ robust control [2], sliding mode control [3], and neural network intelligent control [4] have been presented. In the current paper, a 2-DOF (two degree-of-freedom) con- trol scheme is proposed to suppress the nonlinear hysteresis and achieve a precise motion control. Specially, an inner- loop disturbance observer (DOB) is employed to tolerate the unmodeled hysteresis which is treated as a disturbance to the nominal plant model of the system, and an outer-loop feedback controller is used to compensate for the remaining nonlinearities and uncertainties. Although DOB is extensively adopted in the friction compensation, seldom application 2014