The Studying Of The Pattern Recognition And Self-localization Based On Humanoid Robot NAO

Exploring the unknown environment was one of the frontier subjects in the field of humanoid robot study, The precise pattern recognition and self-localization were key technologies for exploring the unknown environment. How to overcome the uncertainty of external environment and the binding of humanoid robot sensors was the emphasis and difficulty for the pattern recognition and self-localization of humanoid robot. Based on humanoid robot NAO, the pattern recognition and self-localization algorithm were studied under the unknown environment, and were verified and applied in the Robo Cup standard platform.By comparison experiment to ge a better robustness HSI color space, according to the needs of the RoboCup competition, modeled three coordinates. According to the principle of pinhole imaging, it propose a new method of distance.Aiming at identifying the field information(ball, gate, white line) in the humanoid robot football playground, the regional creation, line creation of the basic pattern recognition algorithm were put forward based on the basic theory of computer vision and pattern recognition. It was implemented to accurately identify the field information for the humanoid robot NAO in the RoboCup platform.For the self-localization in the process of the humanoid robot soccer game, it was proposed that using the information of the intersection point of the white line to perform the humanoid robot's self-localization. While the limitations for each robot, it was very difficult for a single robot to acquire the global environment information. The competition field was meshed to realize the whole information modeling in this paper.In the real world, considering the limitations of a single robot in receiving environmental information and other restrains, the self-positioning technology among multi-robot was developed by using multi-communication principle in this paper. The visual sharing was achieved between multiple robots. It was verified effectively in the RoboCup standard platform.