Research On Uncalibrated Object Servoing Control

As society develops,and science and technology advance,studies in robotics are flourishing and still in the ascendant.The control of robotic visual servo,which has long been the hot spot of the day,has laid the basis for robotics intelligence,and thus is being widely used for industrial production and in ordinary life.Visual servo,which is a multidisciplinary area,comprises of image processing,control algorithm,learning algorithm,etc.Coincident with the elevation of intelligence,it has been increasingly important to study visual servo.This thesis first looks back to the developmental procedure of robotic visual servoing,which puts the traditional three systems and uncalibrated control top priority,summarize their pros and cons,and points out their limitations.Concerning Position-based and image-based visual servoing systems,they have the problems of low precision of object detection and positioning,and lack of transplantability and robustness.Attempting to resolve these issues,this these propose a new method,namely,object detection and localization based on models,which,in general,is divided into 2-D and 3-D algorithms,and brings the advantage of high speed of detection,high precision of localization,and proven transplantability.This thesis,as for uncalibrated visual servo,details the Newton algorithm,and concludes its image noise may substantially affect the Jacobian estimation.In order to find out its resolution,this thesis introduces the uncalibrated systems based on Kalman filtering to reduce the influence of image noise on the Jacobian estimation and improve its precision of control.Finally,it uses PID method and adaptive method to realize the control of the manipulators.As to the two problems,this thesis first introduces the basics of robotic visual technology and servoing control based upon it,on the basis of which,this thesis give top priority to the object detection and localization algorithm based on models,and uncalibrated visual servo based on the Kalman filtering.At the end of the thesis,it designs relevant experiments and simulations to prove the efficacy of the proposed method,and propound the deficiency of this study,and outline the future of improvements.