Research On Servo Control Technology Based On Visual And Inertial Information

Control system as the core of machinery manufacturing equipment and industrial control system is widely used in many fields such as industry,agriculture,aviation,aerospace and navigation.With the development of computer technology and machine vision technology,the vision based control system has been extensively investigated and widely used in modern industrial automation due to its advantages of high flexibility and high effiency.The human visual function is simulated in the vision based control system by visual sensor.The useful information is extracted from the surroundings and identified for obtaining motion parameters to control actuator to complete the task.By using vision based control,the social productivity is improved and the human is liberated from extreme working conditions.A visual sensor is used to complete a non-contact measurement in work space.So the image processing and conversion process will be conducted to project the target from the three dimensional space to two dimensional space,the design of servo controller and working pattern of actuator are involved.Moreover,the nonlinear and complex structure of the actuator make visual servoing problem very challenging.Therefore,it is crucial that an effective visual servoing method is proposed to improve the control precision of the closed-loop system and the its applicable scope.In the mean time,it should be noticed that visual information is likely to be influenced by light condition from external environment or image processing technology,therefore its application is limited in some cases.To overcome this drawback,other type sensors can be used and combined with the visual sensor to develop a composite feedback system to improve the measurement accuracy.Inertial navigation developed with inertial sensor is used widely in military and civil field due to its advantages of independence,free from outside interference and real-time output information.Therefore,this thesis is based on the control System,to study vision technology and integrated visual-inertial navigation technology around the Visual Servo System Based on Image(Image-based Visual Servoing,IBVS)and stabilized Platform System(Stabilized Platform System,SPS).In order to ensure the stability and accuracy of the system.The thesis firstly introduces the history and the state-of-the-art of the visual based control system.Some basic concepts in visual based control,such as the classification of different control methods,their basic working principles,and the advantages and disadvantages under different modes are also presented.According to the current feasible solutions,combine with researching actuality of inertial navigation and summarizing application status and developing prospects of integrated visual-inertial information.Preliminary and fundamental description are finished to better research contents of this thesis.To resolve the problem of the irregular shape information in the image and difficult feature recognition which are caused by light or similar color gamut,a Max-Min image area extract and measurement method is presented to realize rapidly extract and measure process in regular scenes.The method is based on image region where target feature is extracted by large-area selection and smaller-area selection.Compared with other target extraction methods,this method is easy to implement,while the target feature can be extracted accurately in the meantime when graphical information owes clear.To address the problem that the traditional feature extraction method is easy to lose target information or fails to detect under fast moving target or noise background when single feature information or weak correlation information are used to describe target,the method proposed in this thesis uses multiple constraint in the target to overcomes the problem of target missing in detection.To handle the problem of unsatisfactory dynamic response when traditional IBVS are used to control a system with uncertainties,a PD-SMC method is presented to achieve fast convergence and ensure superior stability when the system is subjected to disturbances,this method is validated by an eye-in-hand of six degrees of freedom mechanical arm.The PD-SMC method integrates PD control(Proportional derivative control,PD)and Sliding mode control(SMC)which preserves the advantages of simple and easy implementation and fast convergence of PD control.Meanwhile,the sliding mode control method overcomes the uncertainty of the system and enhances the robustness to the disturbance,unmodeled dynamics,and system nonlinearities.When the PD-SMC method is used,the closed-loop stability of the IBVS system is guaranteed,and the requirement of rapid convergence,high control accuracy,and high disturbance rejection capability is achieved.Since the processed image information from visual sensor is used to develop the controller,the control precision depends on the accuracy of image information.However,in some cases,target missing is unavoidable due to external environment disturbances and inadequacy in image processing technology,which usually lead to the failure in control process and the limits the application of vision based control technology.To solve this problem,the visual sensor and inertial sensor for navigation are combined and an integrated control framework is presented.Kalman filtering method is used to integrate the visual information and the inertial information.As a result,the navigation deviation and image detection error are corrected by using this method.Finally,according to different tasks to choose suitable actuator in different forms.Six degrees of freedom mechanical arm is fixed with the bottom base,its moving space is limited.So dual-axis stabilized platform is chosen as the actuator.Aiming at the problem of tracking deviation and tracking/stable phase separation,an integrated visual/inertial information control method for dual-axis stabilized platform is presented to realize real-time tracking for target and stable control for platform.PID control method and switched PI-fuzzy control method are applied in the speed loop and position loop of the stabilized platform system respectively.After testing,target tracking task is accomplished efficiently and the attitude adjustment of the platform is ensured by the proposed method.