Research On Control Of Visual Servoing System For Inertially Stabilized Platform

The inertially stabilized platform based visual servoing system can effectively isolates the movement of carrier,and then use the visual information to identify and track a moving target in the complex environment.It has been widely applied to many military and civil fields such as battlefield situation awareness,weapon aiming system,fire safety monitoring,astronomical and geographical observation and so on.How to improve the rapidity,accuracy and reliability is always the research focus of the target tracking system.However,the multi-source disturbance,visual measurement delay,large-period sampling,and target loss in the system bring great challenges to the design of visual servoing controller.In order to realize the high-performance control of visual servoing system of inertially stabilized platform,this study focuses on corresponding control algorithm designs for the problems in the system.Firstly,the physical structure and working principle of the system are introduced,the system model with carrier and target motions is established,and the generation and influence of control problems in the system are analyzed and summarized.Then,the high-performance control algorithm designs are researched from four aspects: anti-disturbance control,measurement delay compensation,multirate sampled control and output constraint control.1.In order to improve the control accuracy of the system with multi-source disturbances,a model predictive control method based on disturbance observer is proposed.Firstly,in the system modeling,the block method is used to obtain a kinematics matrix independent of the target depth information.Then,the disturbances such as unknown depth information,tracking error of angular velocity and target motion are concentrated as the lumped disturbance,and a discrete-time disturbance observer is designed to estimate it.Then,to suppress the influence of disturbance,the lumped disturbance estimate is introduced into the design of model predictive controller.Finally,the principle of parameters selection is explored through the experiment,and the effectiveness of the proposed controller is verified.In this work,the kinematic uncertainty caused by the unknown depth information of the target is regarded as the lumped disturbance,and the disturbance observer is designed to estimate and compensate it,which provides a simple and efficient controller design scheme for the visual servoing system based on monocular uncalibrated camera.2.Aiming to handle the visual measurement delay and disturbance,a measurement delay compensation control method based on time-delay disturbance observer is proposed to improve the rapidity and accuracy of the system.Firstly,a system model with measurement delay and disturbance is established.Then,a time-delay disturbance observer is constructed to estimate the past-instant disturbance and its differences.Then,the current state and disturbance are predicted by using the disturbance-related estimates and model information,and the sampled-data controller is designed based on the predictions of disturbance and state.Finally,the effectiveness of the proposed control approach for disturbance rejection and measurement delay compensation is verified by target tracking experiments.In this work,a time-delay disturbance observer is proposed with the delay-independent estimate error dynamic,which improves the prediction accuracy of current state and disturbance,and significantly improves the fast-tracking performance of the system.3.Aiming at the problem of large-period sampling of visual sensor,an extended state observer based on output predictor is proposed,and a multi-rate sampled controller is designed.Firstly,based on the discretetime output with measurement delay,an extended state observer based on output predictor is designed to obtain the estimates of current state and disturbance of the system.Then,based on the continuous-time extended state observer,virtual sampling points are added between the real measurement points,and a multi-rate sampled controller is proposed to improve the control frequency of the system.Finally,the superiority of the proposed control approach over the single-rate control method is verified by simulation and experiment.In this work,the stability of hybrid-time systems caused by the digital controller is strictly analyzed,which is more in line with the actual situation than direct discrete-time design and analysis.Besides,in engineering,the output prediction between samples improves the frequency of digital controller.4.In order to prevent the target losing from the camera field of view and improve the reliability of the system in the target tracking task,a control approach based on gain function and disturbance observer is developed.Firstly,the target loss problem is transformed into the output constraint problem,and the system model with disturbance and output constraint is established.Then,a feedback control method is designed based on the gain function to ensure that the target position is constrained in the camera field of view.Besides,to improve the anti-disturbance ability of the system,a disturbance observer is introduced into the controller design.Finally,the effectiveness of the proposed control method is verified by simulation and experiment.In this work,the gain function is introduced to ensure that the system output is constrained by the camera field of view,and the tracking performance of the system is improved by using the disturbance observer.