Optimal Adaptive Visual Servoing of Robot Manipulators

Abstract

The system proposed in this thesis is a part of object tracking system. The work done here is designed to perform two tasks. Firstly the direction of motion of the object is detected and given to the user in visual form. Secondly the speed with which the object is moving is calculated in terms of pixels per frame. In order to achieve this first a set of photographs of the moving object in various positions at regular time intervals is taken. Then the motion is detected by finding the difference of the consecutive images. In order to calculate speed, first the segmentation is done of the images so that the different objects can be separated. Then on the basis of the area the desired region in the image (i.e. the object to be tracked) is separated from the others. In this case a ball has been used as the object. After separating the region the centroid of the region of interest is calculated. This is done for all the images collected. Once the centroid position of the region is found in all the images the coordinates are subtracted to find out the number of pixels the object has moved in the next frame. In this manner it can be calculated that how much pixels the object has traveled in the complete set of images. Therefore the speed of the image is obtained in terms of pixels per frame.