Object detection in changing environment of middle size RoboCup and some applications

Abstract

In this paper, we introduce several novel ideas on robot vision, its implementation on RoboCup and some of its applications. We discuss about a new color model which components are taken from different color models, a fast object detection method using the idea of searching on a few jump points in the perspective view of robot front view CCD camera, a reliable object shape estimation and a fast method to detect a few edge points on field borders for line estimation. Our practical experiments convinced us that in dynamically changing environment such as RoboCup, fast and almost correct solutions to the vision problems are good enough for detecting objects, finding their distance and angle from the robot, collision avoidance and position estimation. This vision system was tested on our robots in RoboCup competitions and very good results were obtained.