Ball tracking with a 2-DOF spherical parallel robot based on visual servoing controllers

Abstract

In this paper, an image based control architecture is developed f o r a 2-degree-of-freedom (DOF) spherical parallel robot. A camera, with digital output, is installed on the end-effector to follow the moving ball f o r a given color which is set by the user in the developed GUI. Some common image processing methods are applied on the obtained frames coming from eye-in-hand visual mechanism. These methods are used to filter out the background colors and noises from the image of the object in the image plane (screen). Smoothing the image of the object by using Median and Gaussian filter are used to precisely compute the centroid of the object in the screen. A PID and a sliding mode controller are introduced and designed to command actuators in such a way that the line of sight of the center of screen and the centroid of the image of the object are kept fixed. Implementation, testing and validation of the control algorithms are realized on the visual servoing architecture. Eventually, a comparison between the image based PID controller and the sliding mode controller is conducted and performances are fully discussed.