Abstract—This paper proposes an alternate method to obtain position and orientation of multiple soccer robots in global vision system. The proposed method uses high-angle shot of camera as a substitute to bird-eye angle, and employs a specially designed robot markers that contain cues for the detection of the position, orientation, and IDs of the robots. Perspective transformation to normalize the perspective arena image obtained from high-angle shot is utilized in this method. Three testing phase, namely unit testing, integration testing, and system testing are conducted. The results of unit testing are: four arenas are detected from four different placements in the room, the robot detection has an error of 0.86%, and ball detection has an error of 0.4%. Integration testing obtains the global orientation of the robot, which has an error of 1.1%. From the result of system testing, the best camera resolution is 360p with 50% minimum camera brightness. The result of system testing also proves that the application system can be implemented in different rooms and with different camera positions.

Index Terms—robot soccer, global vision, image processing, border tracing, perspective transformation, high-angle shot

Cite: Hendra Tjahyadi, Giorgy Gunawan, Arnold Aribowo, and David Hareva, "Image Processing Based Robot Soccer: Obtaining Multiple Robots Position and Orientation Using High-Angle Shot of Camera," Journal of Image and Graphics, Vol. 4, No. 1, pp. 29-35, June 2016. doi: 10.18178/joig.4.1.29-35