Study On Visual Servoing And Fusion Control Under Dynamic Network Delay For Robot

The rapid development of Internet technologies has led to huge change in our lives and the society structure. The information and the media applications such as video-on-demand, distance education become more and more based on Internet. The new Internet-based studies are a popular topic in research including network robot. Different with local robot and telerobot, network robot system can get more information and has more uncertain factors. So the solution of network robot control is a challenge problem. This paper has considered the two questions: 1. mission dynamic configuration, 2. visual servoing. Focusing on the two questions, the present work is devoted to four issues as follows: the information perception and search strategy in dynamic network for robots, uncalibrated visual servoing control between mobile robots and distributed intelligent sensors, estimation of image Jacobian matrix with time-delay compensation, stable switch control for multiple cameras.1. Information perception and search strategy in dynamic network: The telerobot, the early network robot and some distributed intelligent robot systems adopt fixed link. Comparing with these systems, the distributed intelligent robot systems without fixed link has much more flexibility and expansibility and can fulfil more kind of missions. But the Information resource and the control relationship are determined hardly. So this paper implements the robot system based on the Robot Autonomous Perception Model (RAPM) can get the information on the Internet through autonomous perception against the problem"mission dynamic configuration", replace the dynamic network with a temporary static network. RAPM can abstract the entities in Internet, classify real-time information and get those on the Internet through autonomous perception to build paths. Genetic Algorithms optimize paths for multiple robots and avoid resource competition in robot workspace. The method breaks the limitation that robots only use the information of own or fixing sensors on the network and fulfill the mission out of the robots'field range. The experiment shows RAPM possesses greatly extended capability and proper structural frame, and the optimizing way based on GA is effective. This has significant purpose for increasing Internet-based robots intelligence and behavior capacity.2. Uncalibrated visual servoing for mobile robots: In such distributed intelligent robot systems the control relationship is dynamic. Usually we can not get all system calibrated information in almost situation and the offline calibration process is complicated and error-prone to be realized in practice. Thus uncalibrated visual...