Research And Design Of Autonomous Tracking Robot For Football Match Broadcast Outside The Football Field Line

With the rapid development of robotics and the maturity of machine learning technology,tracking robots are widely used in various areas,and replacing humans in more and more jobs to complete complex tasks,achieved intelligence of production and life.Questions about the shooting and broadcast of sideline area in football matches,this paper researched and designed an autonomous tracking robot based on target detection and other technologies,the purpose is replace the sideline camera team to automate match broadcasts,reduce the work intensity of personnel while enhancing safety,at the same time,enhance shooting flexibility and reduce resource investment.The design form of the autonomous tracking robot is a two-wheeled selfbalancing robot,including three parts: moving platform,visual platform and detection and control platform,the system takes the target detection algorithm based on deep learning as the core technology,record and broadcast the event by tracking and shooting football or players in the sideline area.This paper has researched Mobile Net network and SSD algorithm,analyzed and designed MV3-SSD target detection algorithm,lightweight network facilitates building and application on embedded devices,and provides a strong guarantee for the autonomous tracking robot to achieve the expected goal.At the same time,this paper analyzes the limitations of the MV3-SSD target detection algorithm,taking the application environment of autonomous tracking robot as the starting point,analyzed the data distribution of tracking target aspect ratio with K-Means clustering algorithm.By optimizing the aspect ratio and quantity of the prior layer of the feature layer in the MV3-SSD algorithm and introducing a new loss function,this paper proposes the Specific MV3-SSD algorithm for specific target detection.In addition,this paper studies the target tracking strategy,by combining the design of positioning and tracking control,further improve the intelligent level of autonomous tracking robot,and ensure the efficiency and flexibility of shooting in the sideline area of the football field.This paper builds an experimental platform for autonomous tracking robot according to the design plan,debugged the overall operation of the system by some tests such as basic operation of prototype,positioning movement and image acquisition and transmission.In the test,this paper uses PASCAL VOC2007 and other data sets to train and perform performance testing on the MV3-SSD target detection algorithm and Specific MV3-SSD calculation,and evaluated by m AP and FPS indicators.Research shows that the target detection algorithm designed and optimized in this paper significantly improves the detection speed and the detection accuracy of specific targets,and the autonomous tracking robot basically has the ability to replace the sideline cameraman for shooting football matches,at the same time,the research results also provide new ideas for the detection of specific targets in other fields.