Iterative Learning Control of a Parallel Delta Robot Chems Eddine Boudjedir 1( ✉ ) , Djamel Boukhetala 1 , and Mohamed Bouri 2 1 Ecole Nationale Polytechnique (ENP), 10 Rue des Freres OUDEK, El-Harrach, 16200 Algiers, Algeria {chemseddine.boudjedir,djamel.boukhetala}@g.enp.edu.dz 2 Ecole Polytechnique Fédérale de Lausanne (EPFL), Station 9, 1015 Lausanne, Switzerland Mohamed.bouri@epfl.ch Abstract. In this work, an iterative learning control (ILC) is applied to a Delta robot in order to improve the tracking precision at high dynamic movement. Delta robot is a parallel manipulator designed for high-speed pick and place operations. Since the dynamic of Delta robot is highly coupled and nonlinear, the conven‐ tional controller like the proportional derivative (PD) failed to satisfy the required performances. To solve this problem a suitable controller for repetitive pick and place operations represents in the ILC has been introduced. The learning controller is combined with a PD controller to improve the tracking error through the iterations. Numerous simulations and robustness tests are carried out to demonstrate the effectiveness of the proposed scheme. Keywords: Iterative learning control · Delta robot · High dynamic movement PD control 1 Introduction The architecture design of Parallel Kinematic Manipulators (PKM) knew a huge progress over the years, starting from the 5 degrees of freedom (dof) of Pollard [1] to the 6 dof of Stewart platform [2] up to the 4 dof of Delta robot [3]. PKM plays a very important role in the industry due to the advantage they present compared to the serial arms, such as higher load capacity and more rigidity and accuracy. The Delta robot was invented by Reymond. Clavel in 1985 [4] from the EPFL (Ecole Polytechnique Fédérale de Lausanne), where it dedicates to execute pick and place operations at a high dynamic movement. The first prototype has three dof for translation and one dof for rotation, after that, various designs with Delta link architecture were developed, for instance, the Linear Delta [5] and the Inverted Delta [6]. For more designs of the Delta robot the reader may refer to the survey [7]. The traditional controllers like PD/PID are usually used to control the robot manip‐ ulators due to its easy synthesis and implementation. However, PD/PID failed to track the desired trajectory for high-speed pick and place tasks, because the gains of the controller are selected without considering the coupling effects. To overcome this © Springer Nature Switzerland AG 2019 M. Chadli et al. (Eds.): ICEECA 2017, LNEE 522, pp. 72–83, 2019. https://doi.org/10.1007/978-3-319-97816-1_6