A performance improvement of industrial delta robot by using Iterative learning control

Abstract

Delta robots, which are widely used in industrial settings, are more complicated in shape and control than serial robots, but can move quickly within a narrow workspace, so they have advantage for picking and moving objects (Pick and Place Motion). Delta robot is a robot system that is suitable for precise work because its end effector is light and has an advantage in terms of inertia.

In general, only PID control is applied to control delta robots used in industrial sites. However, it is challenging to tune the PID gain so that the various motions performed by the delta robot can produce the best performance.

In this study, it is intended to improve the control performance of all motions performed by the delta robot by using the Iterative Learning Control (ILC) technique. ILC, one of the control methods, saves errors generated while performing motion and uses them to control the next same trial. Since ILC has the characteristic that the error converges to 0 when repetitive operations continue, it is effective in controlling systems that perform many identical operations. In addition, performance can be improved through learning each of the various motions.

This paper presents the process of improving control performance by applying iterative learning control to a delta robot that installed a general PID controller. In this paper, 1) the system was constructed based on the robotics theory of the delta robot, and 2) the process of applying the PID controller and the ILC controller to the delta robot was presented. 3) Then, the simulation results are analyzed, and the effect of applying ILC was evaluated. 4) Based on the research results, a learning scenario that can be used as a tuning guide in industrial sites using delta robots was proposed.

This thesis will serve as a reference guide for researchers who want to research the control of delta robots or engineers who want to use delta robots in the industrial field.