Research Of Hardware Design Implementation And Vision System Of Soccer Robots In Robocup Middle-Size Matches

RoboCup^? is an international joint project to promote AI, robotics, and related field. It is an attempt to foster AI and intelligent robotics research by providing a standard problem where wide range of technologies can be integrated and examined. RoboCup chose to use soccer game as a central topic of research, aiming at innovations to be applied for socially significant problems and industries. Especially for middle-size soccer robot, each robot is autonomous, it raises more requirement for its transport mechanism, the performance of the movement, the stability, real-time processing, and accuracy of vision system. So developing the autonomous robot system and taking part in the matches is a challenging topic. Robot system contains four parts, which are decision-making subsystem, vision subsystem, wireless communication subsystem and motion control subsystem, and vision subsystem plays an important role in the whole performance.This thesis analyzed and researched current situation of the project home and abroad, then made some contribution on hardware platform construction and improvement, vision system design and implementation, field lines extraction and classification. Specific descriptions are as follows:(1) As the main participant, designed and implemented a hardware system as robot research platform, then improved omni-vision mirror, brace of mirror, and goal keeper. Omni-vision system can see 360°direction of the environment information; Omni-directional motion structure can travel in a confused dynamic circumstance and achieve an accurate positioning in a competition freely. After many experiments, we found the defect in the hardware architecture, then chose another omni-vision mirror based on the resolution fixity in 6.5m. Re-designed the front-vision hardware model, enhanced the mirror brace, and changed the ball-keeper mechanism, which made the whole hardware platform more steady and efficient, so that provided us a better experimental platform. (2) Based on hardware platform, designed and implemented vision system which combined omni-vision and front vision. Choose YUV color model which is affected little by lighting condition, adopted look-up table and RLE coding technology which reduced time complexity. Enhanced object recognition algorithm, made use of the context between two images, and increased the reorganization accuracy and speed. Because there was known defects in hardware manufacture and installation, so we used a mechanism which combined the omni-vision and front vision processing based on the multi-thread programming method. In this way, the front vision can help to recognize the ball when omni-vision fails to recognize the ball in image.(3) Enhanced the radius scan field lines extraction method, and analyzed the robot self-location issue using field lines. Based 2008 Middle Size RoboCup rules, the colored landmark will be removed gradually, using field lines for robot self-location becomes more and more important. In this thesis, researched existing field lines extraction methods, and enhanced radius scan to extract field lines method. Defined field lines features according to its non-structured and structured characters, classified its pattern, and analyzed to apply it to robot self-location.All in all, the hardware platform, vision software system, and using extracted field lines can adapt the matches, and get good grade in national and provincial level matches.