Detection And Tracking Of Moving Object Based On The Self-localization Robot

With the development of intelligent technology and computer vision, the application ofmobile robot is more and more widely. Mobile robot in the actual development and life playsan important role, and it cannot be ignored, such as in the field of power industry, industrialproduction, medical service, video surveillance, public service and so on. Research on thepositioning and navigation of mobile robots has become a new hot spot, it is a very valuableresearch to realize the position by itself. For a robot to walk in the real environment, toachieve the moving target detection and tracking is another important aspect of intelligentrobots, this technology has broad prospects in aerospace, video surveillance, intelligent robotfields.The main contents are as follows:1. Discussed and reviewed the research status and development of the mobile robot, theresearch content of machine vision and the development of robot localization, and introducedthe mobile robot on the direction of the moving target detection and tracking related researchpresent situation, and the main structure of this paper are introduced in this paper.2. The robot self-localization. In order to achieve its own position on the navigation path,designed by a "0","1" composed binary string of manual coding, radial ply in the intersectionand T-intersection. The electronic map reflected by the actual navigation path, Dijkstraalgorithm plan the best way to the destination, desk computer(decision control terminal, AIServer) and the PC(Intelligent control terminal, Control Server) through the wireless networktransfer the information.The processing results returned to the control server to control therobot’s behavior, it is the important basis of next behavior such as left, right or straight, brakesand others, then help the robot complete self localization.3. Detection and tracking of moving object based on the self-location mobile robot. Firstof all, find out a large gradient between the two images, to calculate the relative displacement,and with the relative displacement improve the image motion compensation, the dynamicbackground is converted to a static background and then processed. Second use vibealgorithm to identify and detect the moving target. Finally, the moving target trackingcompleted by the Kalman filter added into the CamShift. The next step is to further explore of the fast moving target detection and tracking basedon the robot, as well as robot avoidance the obstacle under complex circumstances, finallyachieve the effective convergence between obstacle detection and avoidance strategies. Atpresent, the problem of optimization and the solution strategy is being studied.