Font Size: a A A

Research On Robotic Visual Servoing And Networked Control System

Posted on:2010-12-24Degree:DoctorType:Dissertation
Country:ChinaCandidate:X H ZhangFull Text:PDF
GTID:1118360305470182Subject:Control theory and control engineering
Abstract/Summary:PDF Full Text Request
Robotic visual servoing system has the advantages of higher flexibility and precision, strong robustness to calibration error and so on compared with robot control system based on traditional sensor, which is a research focus in the field of robot control. Networked Control Systems (NCSs) developed in recent years are also research focuses in the field of control, which are the automation technology based on network, and the combination of computer network, communication and automatic control theory. But the Networked Robot System developed from telerobot is then the combination of the above two research fields, so it becomes more challenging. The dissertation develops the studies of some problems in the fields of robotic visual servoing and networked control by constructing a robotic visual servoing system based on network. The main work presented in the dissertation can be summarized as follows.1) A new method for image feature extraction is proposed according to the idea of template matching to perform the tasks of making a robot end effector track the mobile objects with various shapes on a plane, which is based on fast generation of matching template and image search by Genetic Algorithm (GA). In order to enable the image target tracking system fast transform among the various shapes objects, the template fast generation algorithm is proposed and the correlation degree function is defined. The index of target deviation degree and its computing algorithm is also put forward by experimental study of static images, which can improve the distinguish capacity for various shapes and the searching speed in an image. The template parameters are optimized by GA using the above index as fitness function. Real-time performance of the algorithm is guaranteed by adopting some techniques of Step GA, image classification searching and vision wave gate dynamic setting. The visual tracking experiments for the various shapes targets show that the image feature extraction algorithm proposed has the advantages of fast and flexible transformation of target, high distinguish capacity for various shapes, strong real-time performance for image search, strong anti-noise ability and so on. The algorithm can completely satisfy the requirements of the image feature extraction demanded by robotic visual servoing system studied in this paper.2) Transformation matrices of the joints coordinate system, jacobian matrix and image jacobian matrix are derived according to the motion mechanism and link parameters of robot RBT-6T/S04S. Mathematical model of the robotic visual servoing system is established. Simulation model of the robot RBT-6T/S04S, Camera system module and visual jacobian module are implemented by using the Robotics Toolbox (Release 7.1) in matlab environment, and then the image-based visual servoing system is simulated and studied by using proportion control law.3) A Linear Time Invariance (LTI) system with nonnegative system matrix elements and using all-state feedback control law is discreted in network environment, and then a corresponding networked control system is built. According to the above system and under the research hypotheses of single-packet transmission, no data packet dropout and network delays taking arbitrary values within [0, h] (h is the sampling period), it is proposed and proved that if the variable delay of a networked control system at any arbitrary moment is less than a certain constant value and the system with the constant value as delay is asymptotically stable, then the networked control system with variable delay is asymptotically stable too. Firstly, the effects of sampling period and network delay on system stability and dynamic performance are analysised in the paper. Secondly, according to the known achievements in research, the stability region of networked control system for sampling period and constant network delay is determined by using Hybrid System theory on the promise of invariant state feedback gain. Thirdly, the stability region of networked control system with the same state feedback gain for sampling period and variable network delay is obtained through the theory analysis. Finally, the validity of the conclusion is verified by simulation experiments.4) Considering the networked control system and the research hypotheses in 3) and following the design principle'Remote is complex and local is simple'of Agent Based Control (ABC), the delay compensation method of networked control system based on layer structure is proposed. In networked control system, the controller can't know the delay of the current signal when calculating the control signal and the network delay can get arbitrary value, so accurate prediction of network delay is impossible. In this paper, a simple local controller is used, the remote controller need not calculate the final control signals, but calculates some intermediate results demanding complex operation and sends them to the local controller, then the local controller calculates the final compensation control signals based on the true network delay of current control signals, so the question of accurate compensation for the variable network delay is solved. The method proposed is verified by some simulation experiments.5) Considering the networked control system and the research hypotheses in 3), a new method of system synthesis is proposed. The basic steps of the method are as follows. Firstly, the state feedback gain is designed based on the mathematical model of the continuous time system to meet the anticipated performance. Secondly, the optimum sampling period is obtained by solving a conditional extremum problem during the discretization procedure in network environment, which has the best capability of tolerating variable network delay on the promise of maintaining the stability of the system obtained. Finally, the system obtained is compensated by adopting the above delay compensation method of networked control system based on layer structure, and then the system has near the same performance compared with the continuous time system. In the paper, a continuous time LTI system of equivalence of the time-variant visual servoing system is found by analyzing the characteristics for control law of the robotic visual servoing system. The visual controller in network is designed by adopting the above method of system synthesis, and the simulation study of the image-based networked robotic visual servoing system is made.Finally, the image-based robotic visual servoing local and networked systems based on robot RBT-6T/S04S are built, the methods and conclusions proposed in the paper are verified by experiments.
Keywords/Search Tags:Visual servoing, Image feature extraction, Genetic algorithm, Template matching, Networked control systems, Stability, Variable delay compensation, Networked control system synthesis
PDF Full Text Request
Related items