Abstract
RoboCup is an international project to promote AI, robotics, and related field. In RoboCup Soccer Middle Size League, the robots should recognize color and detect objects under the natural light condition. This paper presents an auto color calibration system using SVM (Support Vector Machine) for the autonomous omni-directional mobile robot. Our approach defines 5 classes of color (yellow, green, white, black, others). In our experiments, we compare the image binarized by human and the image binarized by SVM. This results more than 90% of recognition rate. We consider our approach can be useful in the RoboCup environment and we will improve it to run in real time.
