Abstract—This work demonstrates performance improvement in motion control under a particular set of machine system constraints. A high performance industrial magnetic micro-manipulation system, the Minimag, is introduced and modeled with both first principles and system identification. The form of the closed loop controller is constrained by operational bounds and the system software, resulting in limits to achievable performance. A model-based motion control enhancement is developed and implemented using tools from Iterative Learning Control. The resulting performance improvements indicate the benefits of motion control even when the closed loop controller is fixed. Experimental results at two different size scales (500 m and 4.5m) are given. I. INTRODUCTION AND SYSTEM OVERVIEW The use and impact of microscale machines have grown tremendously in the past 20 years [1,2]. With increased utilization of microscale machines, across diverse fields [3], comes the demand for increasingly stringent manipulation at commensurate length scales. Automation is a key enabler for this microscale manipulation and there are emergent applications in fields such as robotic microassembly [2] and also cell mechanics [4]. Throughout the various fields of use for microscale machines, significant opportunities exist for control tools to contribute in their development and performance [3]. This article focuses on one such specific example. Several different techniques have been proposed that can servo microscale objects to localized targets; these include both tethered actuators [5] and untethered systems. Untethered actuation can be based on dielectrophoretic [6], magnetic [7], fluidic [8], acoustic [9] or optically induced forces [10]. Depending on the application, one approach might be advantageous over the other. Magnetic forces are favorable especially in several biological applications that require a biocompatible, non-contact approach. In many of these biological applications, magnetic fields are used to manipulate small spherical objects; these objects are available in range of sizes and are functionalized as the task requires. The key to the object manipulation is the ability to precisely (<1m) transport these systems over distances of a few to hundreds of micrometers. A magnetic manipulation system that is specifically designed for such a task is the MiniMag [11,12]. The system consists of 8 electromagnets A. Alleyne is with the Univiersity of Illinois, Urbana-Champaign, 1206 West Green Street, Urbana, IL, 61801 (phone: +1-217-244-9993; e-mail: alleyne@illinois.edu) S. Schurle and B. Nelson are with ETH Zurich. (email: schuesim@ethz.ch and bradley.nelson@iris.mavt.ethz.ch)). A. Meo is with the University of Pisa, Italy (email: alessandro.meo@gmail.com) arranged in a single hemisphere and can be incorporated into either an upright or inverted microscope (see Fig. 1.a). Magnetic samples are placed in a custom-made sample holder on the x-y-stage with the lens is located underneath. Figure 1: (a) Picture of microscope mounted MiniMag and schematic of an inverted coil configuration; (b) screenshot of the control panel Graphical User Interface (GUI). Based on visual feedback the magnetic object can be servoed to a target location in open or closed-loop fashion. A typical scenario is depicted in Fig. 1.b. The position of the object, a 500 µm large magnetic sphere, is detected based on a visual servoing algorithm. The region of interest is indicated by the blue box. Desired trajectories that specify the task can then be loaded as system input or entered by the user through the GUI. Here, two trajectory waypoints (grey) are given as the input target signal. The interested reader is referred to [11,12, 13], and the references therein, for further details on the MiniMag system construction, use, and applications. Motion Control for Magnetic Micro-scale Manipulation Andrew G. Alleyne, Simone Schurle, Alessandro Meo, Bradley J. Nelson (b) (a) 2013 European Control Conference (ECC) July 17-19, 2013, Zürich, Switzerland. 978-3-952-41734-8/©2013 EUCA 784