Real time game field limits recognition for robot self-localization using collinearity in Middle-Size RoboCup Soccer Fernando Ribeiro (1) Gil Lopes (2) (1) Department of Industrial Electronics, Guimarães, University of Minho (2) Department of Computer Sciences, Porto, Portucalense University Keywords: Mobile Robotics, RoboCup; self-localization; collinearity; edge detection. Abstract Enabling a mobile robot to achieve its self-localization in real-time with vision only, demands for new approaches and new computer algorithms. An approach for giving game field self-localization to a Middle Size RoboCup Soccer robot can be based in two steps: finding the game field lines and evaluating the obtained coordinates calculating the robot coordinates. This paper describes a method to achieve the first step. This approach is based on an algorithm that combines three major features: edge detection, selection and collinearity search. The final target is to retrieve the line segments (defined by its two limits coordinates), which identify the game field boundary lines. These line coordinates will be used on the next step that is the process to calculate the robot position in the game field. Since this first step is to find lines in real time, it is an alternative method to the Hough Transform Method. 1. Introduction Computer vision in Middle-Size RoboCup Soccer is greatly responsible for success in this competition. Apart from other technologies that some teams use to achieve robot location/orientation (a description is made by [5], most of them are based on computer vision and recognition. But robot orientation does not mean robot localization. MINHO team [6] uses cameras for giving robot orientation even though the robots do not know where they are on the field. They simply follow a purpose of making