Research On Tracking Dynamic Object For Mobile Robot

For the research topic of mobile robot, the following main contributions about visual tracking are presented in the thesis: The mobile robot tracking algorithm based on the monocular vision is discussed; overall framework to track moving object is designed; based on the pan-tilt camera, a pure color object identification and tracking system is established. So that pure color object is tracked by the mobile robot successfully.Firstly, Kalman filter algorithms is introduced, and it is used to filter the historical position so that the dynamic parameters are estimated. Some experiment results show that the dynamic process of tracking moving object via pan-tilt cameras is more smoothly, also the tracking rate is more quickly due to use the Kalman filtering algorithm.Secondly, for improving the dynamic performance, from establishing the mathematical model of the camera, the relationship between the images target and the rotation angle of the pan-tilt can be described effectively so that the pan-tilt equipment is controlled to keep the goal in the field of the central region.Thirdly, HSI and RGB color model is introduced. Some experiments are done to compared the both of color models. In addition, a seed filling algorithm is used to pick-up the target from the image information.Finally, the flow chart of the presented and experimental results with their analysis are given. Under the VC++.NET2003 environment, a friendly interface is developed so that the control of pan-tilt and the robot's motion control function modules are combined organically.