Detection and object position measurement using computer vision on humanoid soccer

Bandung State Polytechnic (POLBAN) has participated twice in humanoid robot soccer competition. From those two participations, it was known that the weakness was in computer vision. Computer vision capability is constrained by robot hardware specifications so that it was impossible to embed our recent object recognition application. In this study, we propose a computer vision system that implemented the latest technology similar to that technology used in the humanoid soccer winner season 2011. The model uses a field where the object and size comply with the rules of humanoid soccer tournament 2011. Some previous methods use off the field camera which is cannot be used in humanoid soccer tournament because the sensor used has to be attached to the robot. While the approach in this paper emphasized to the fact that goalkeeper's position tend to be static relative to the object in a competition field. Goal keeper through its vision system recognizes objects and measures ball position using image processing technique. The process of ball position measurement was first carried out by recognizing three different objects in the competition field: ball, goal's bar, and field line. Recognition process utilizes back projection method based on HSV information. After the three objects were detected, the measurement of ball position on the field was carried out by ANN model by considering ball position in the image, position of goal's horizontal bar, and the middle field line point. After 10,000 training, the result is encouraging with the average error is less than 1 cm.

[1]  Jürgen Leitner,et al.  Autonomous learning of robust visual object detection and identification on a humanoid , 2012, 2012 IEEE International Conference on Development and Learning and Epigenetic Robotics (ICDL).

[2]  Chih-Hsien Hsia,et al.  An efficient object recognition and self-localization system for humanoid soccer robot , 2010, Proceedings of SICE Annual Conference 2010.

[3]  Chih-Hsien Hsia,et al.  Stereo vision-based self-localization system for RoboCup , 2011, 2011 IEEE International Conference on Fuzzy Systems (FUZZ-IEEE 2011).

[4]  Chih-Hsien Hsia,et al.  A Stereo Vision-Based Self-Localization System , 2013, IEEE Sensors Journal.

[5]  Chih-Hsien Hsia,et al.  An efficient object recognition system for humanoid robot vision , 2009, 2009 Joint Conferences on Pervasive Computing (JCPC).

[6]  Giorgio Metta,et al.  Active object recognition on a humanoid robot , 2012, 2012 IEEE International Conference on Robotics and Automation.

[7]  Yoshiaki Shirai,et al.  Tracking players and a ball in soccer games , 1999, Proceedings. 1999 IEEE/SICE/RSJ. International Conference on Multisensor Fusion and Integration for Intelligent Systems. MFI'99 (Cat. No.99TH8480).

[8]  Murray Shanahan,et al.  Object Detection for a Humanoid Robot Using a Probabilistic Global Workspace , 2008, ICVW.

[9]  Hubert Konik,et al.  CAMSHIFT improvement on multi-hue and multi-object tracking , 2011, Proceedings of the 2011 International Conference on Electrical Engineering and Informatics.

[10]  Chih-Hsien Hsia,et al.  Efficient neural network approach of self-localization for humanoid robot , 2009, 2009 Joint Conferences on Pervasive Computing (JCPC).