Robust location tracking method for mixed reality robots using a rotation search method

In the RoboCup Mixed Reality Soccer League, the location and direction of each micro robot on the field are detected by a camera positioned above the field. However, setting and adjusting the camera to achieve greater accuracy takes a very long time. Besides, the detection process is influenced by the surrounding environment, including illuminance of the room and location of the field. Therefore, while the robots are in play, accuracy gradually decreases as the illuminance changes over time and eventually, the robots become invisible to the camera. In these situations, changes in illuminance and motion blur can cause serious issues. To solve this problem, we apply a rotation search method in the detection system, which is a “vision tracking system”. In addition, we implement a redesigned vision tracking system and evaluate the new system. Our results confirm that the new system is effective.