Research On Visual Servo Target Grasping Control Of Underwater Vehicle Based On Line Feature

Entering the 21 st century,rapid progress has been achieved in the field of underwater robots.As the main approaches of deep-sea exploration and underwater operation in various countries,underwater robots play an important role in the process of human utilization and development of marine resources.With increasingly sophisticated and complex applications of robots in underwater environment,a lot of work requires not only the ability to observe the environment,but also to interact with surrounding objects in unstructured environments.Therefore,the Underwater Vehicle Manipulator System(UVMS)composed of a traditional underwater vehicle and multi-functional manipulator gradually becomes the research hotspot in the field of marine intelligent equipment.Stable and reliable grasping of underwater targets is an important manifestation of UVMS’ ability to interact with the surrounding environment.In order to achieve this task,the current UVMS capturing operation is often completed by remote control with professional operators on the surface with console.This remote operation method requires professional operators to perform long-term high-pressure operations at sea,and emphasizes the operator’s role in the closed loop,which is not conducive to long-term high precision operations and the improvement of system intelligence.Compared with the remote control scheme,the autonomous visual servo grasping scheme,in which the visual feedback of the camera is introduced into the closed loop to guide the UVMS to complete the grasping task,has the reaction speed,work efficiency and intelligence advantages.Hence,this paper starts from UVMS autonomous visual servo grasping tasks,and the main research contents are as follows:1)Model establishment of the UVMS visual grasping system is completed.According to different installation position of the camera mounted on the visual grasping system,the relationship between the objects observed by the camera and their corresponding locations in the image space is determined by analyzing the forward kinematics of UVMS and the perspective projection model of the camera.The general framework of the visual grasping system under arbitrary camera configuration is derived which lays the foundation for the subsequent research on visual servo grasping methods.2)Aiming at the kinematics control problems of autonomous visual servo grasping tasks,the line feature in the image space is represented by feature parameters,and the image Jacobian matrix for line feature is further derived to establish the relationship between the feature motion and the desired motion of UVMS.According to different camera configurations and the characteristics of the UVMS kinematic redundancy,two autonomous visual servo grasping schemes based on line feature are proposed.For camera "eye-to-hand" configuration,the mapping between the image error and the desired velocity of the underwater vehicle is established to realize the autonomous grasping tasks with the underwater vehicle motion only.For camera "eye-in-hand" configuration,the mapping between the image error and the desired velocity of UVMS is established to realize the autonomous grasping tasks under the coordinated motion of the underwater vehicle and the manipulator simultaneously.Meanwhile,a strategy is proposed to ensure the porper performance of the system under the temporyry disappearance of the line feature considering the sudden interruption of line feature extraction during the grasping tasks.3)Aiming at the dynamic control problem in the autonomous visual servo grasping task,a UVMS motion control algorithm based on time delay estimation is proposed.The time delay estimation term is utilized to estimate and compensate the lumped disturbance on the UVMS in real time.The control scheme has low dependence with modelling accuracy.This paper uses Lyapunov theory to analyze stability of the proposed method and the control performance is verified through simulations.Finally,the feasibility of the proposed visual servo target grasping control schemes are demonstrated through simulation results.