Development of a computer vision system applied to robot soccer

• Vitor da Silva Dias UFRB
• João Carlos Nunes Bittencourt UEFS

DOI: https://doi.org/10.5753/erbase.2022.228873

Abstract

In the IEEE Very Small Size category of robot soccer, one of the challenges is to determine the position coordinates of both the robots and the ball throughout a match. For this, in general, it is necessary to use computer vision techniques. In a match, robots adopt custom markers from standard colors. The solution proposed by this work uses two models of Convolutional Neural Networks based on the real-time object detection system You only look once (YOLO), which works according to time markers, to efficiently track the robots and the ball. The system operates at a rate of 2 FPS in player detection and 30 FPS in time and ball identification, returning the position of each detected object.

References

Braga, A. d. P., Ludermir, T. B., and Carvalho, A. C. P. d. L. F. (2000). Redes neurais artificiais: teoria e aplicações. LTC.

HAN, S. A Guide to Convolutional Neural Networks for Computer Vision. Crawley: Morgan Claypool, 2018.

FLECK, L. Redes Neurais Artificiais: Princípios Básicos. Revista Eletrônica Científica Inovação e Tecnologia - Universidade Tecnológica Federal do Paraná, 2016.

REDMON J.; DIVVALA, S. You Only Look Once: Unified, Real-Time ObjectDetection.arXiv:1506.02640, 2015.

RUDER, S. An overview of gradient descent optimization algorithms. arXiv:1609.04747,2017.

REDMON J.; FARHADI, A. YOLO9000: Better, Faster, Stronger. arXiv:1612.08242, 2016.

KHAN, S. A Guide to Convolutional Neural Networks for Computer Vision. Crawley: Morgan Claypool, 2018.

MILITÃO G.; COLOMBINI, E. RoboCup Soccer Ball Depth Detection using Convolutional Neural Networks. Universidade Estadual de Campinas - Instituto de Computação, 2017.

OLIVEIRA, M.; BOWEN, B.; MCKENNA, R.; CHANG, Y.-S. Fast digital image inpaintingIn: Proceedings of the International Conference on Visualization, Imaging and Image Processing.

Hoopes, D., Davis, T., Norman, K., and Helps, R. (2003). An autonomousmobile robot development platform for teaching a graduate levelmechatronics course. In33rd Annual Frontiers in Education, 2003.FIE 2003., volume 2, F4E–17. IEEE.

Kitano, H., Asada, M., Kuniyoshi, Y., Noda, I., Osawa, E., and Matsu-bara, H. (1997). Robocup: A challenge problem for ai.AI Magazine,18(1), 73–85.

Kitano, H., Asada, M., Noda, I., and Matsubara, H. (1998). Robocup:robot world cup.IEEE Robotics Automation Magazine, 5(3), 30–36.

Tadokoro, S., Kitano, H., Takahashi, T., Noda, I., Matsubara, H., Shin-joh, A., Koto, T., Takeuchi, I., Takahashi, H., Matsuno, F., et al.(2000). The robocup-rescue project: A robotic approach to the di-saster mitigation problem. InProceedings 2000 ICRA. MillenniumConference. IEEE International Conference on Robotics and Auto-mation., volume 4, 4089–4094. IEEE.