Research On Target Recognition Method Based On RoboCup Standard Soccer Robot

With the improvement of human science and technology,intelligent robot technology has been an unprecedented development.RoboCup(Robot World Cup)as a major international robotics competition,for researchers in the field of robotic technology research around the world to provide a broad platform,to promote the progress of robotics.In RoboCup,vision is an important source of the robot’s perception of environmental information,which plays an important role in the whole system.In this paper,we study the vision system of soccer robot in the competition environment with the background of RoboCup football match.First of all,the basic function of the visual system is to realize the detection and recognition of the target.This paper analyzes the image of NAO robot access,select the most efficient way for image acquisition.HSV color space model is selected for special game environment.After filtering off-scene background,the target image can be segmented quickly.And the noise reduction processing is performed on the divided image.A fast field line recognition algorithm based on scan line seed generation is proposed in this paper.On this basis,it realizes the fast and accurate detection and identification of special markers in the fields such as the intersection of the field lines and the goal posts.Secondly,the robot should also be able to obtain the depth information of the observed object,to provide support for the robot movement and localization.NAO robots have only monocular vision function.To this end,this paper firstly analyzes several commonly used monocular ranging models.Then,based on the ground plane constraint model,the monocular vision ranging for NAO robots is derived.The calibration experiment of the camera is carried out,and the parameters of the camera are obtained,and the monocular ranging experiment is carried out based on the ranging model.The experimental results show the validity and accuracy of the model.In the football match,the identification of the players is the basis of the strategy of route planning,passing the ball and so on.In order to meet these requirements,a recognition algorithm based on Haar-like features and cascade AdaBoost classifier is proposed.The candidate target regions are identified by using the color histogram matching for second times,in the removal of false detection area at the same time,NAO also identified other robots being friend or foe.The experimental results show that this method can effectively identify the NAO robot,and meet the requirements of the robustness and real-time of the recognition algorithm.