Sensor and Information Fusion Applied to a Robotic Soccer Team

This paper is focused on the sensor and information fusion techniques used by a robotic soccer team. Due to the fact that the sensor information is affected by noise, and taking into account the multi-agent environment, these techniques can significantly improve the accuracy of the robot world model. One of the most important elements of the world model is the robot self-localisation. Here, the team localisation algorithm is presented focusing on the integration of visual and compass information. To improve the ball position and velocity reliability, two different techniques have been developed. A study of the visual sensor noise is presented and, according to this analysis, the resulting noise variation depending on the distance is used to define a Kalman filter for ball position. Moreover, linear regression is used for velocity estimation purposes, both for the ball and the robot. This implementation of linear regression has an adaptive buffer size so that, on hard deviations from the path (detected using the Kalman filter), the regression converges more quickly. A team cooperation method based on sharing of the ball position is presented. Besides the ball, obstacle detection and identification is also an important challenge for cooperation purposes. Detecting the obstacles is ceasing to be enough and identifying which obstacles are team mates and opponents is becoming a need. An approach for this identification is presented, considering the visual information, the known characteristics of the team robots and shared localisation among team members. The same idea of distance dependent noise, studied before, is used to improve this identification. Some of the described work, already implemented before RoboCup2008, improved the team performance, allowing it to achieve the 1st place in the Portuguese robotics open Robótica2008 and in the RoboCup2008 world championship.