Underwater Robot Based On Visual Aided Positioning Target Grab Control

In the past few decades,underwater vehicle technology has made considerable progress.Underwater robots are important equipment for performing sea creature capture,deep-sea surveys,military applications and other tasks.A large number of applications require not only the observation function of the underwater robot,but also the ability to operate.Therefore,the underwater vehivle-manipulator system(UVMS)as an important part of underwater robots,it has aroused people’s research interest.The target grasping ability is an important manifestation of the underwater robot’s operating ability,which can usually directly reflect the performance difference of UVMS.In order to achieve the purpose of target capture,traditional UVMS usually adopts remote control operation technology,which can be used by human pilots on surface ships to achieve remote target capture tasks through handles.However,remote operation has disadvantages that cannot be ignored.UVMS operators need to work in a stressful environment,and the success of the operation depends largely on their skills and attention.Compared with the traditional UVMS remote control operation technology,the target grasping autonomous operation technology has greater challenges,involving underwater visual perception,navigation and positioning,kinematics redundant solution,motion control and many other fields.Therefore,this article starts from the successful realization of the goal of UVMS to capture autonomous operations.The main research contents are as follows:1)Aiming at the positioning problem of underwater robots in the process of target grasping,the delay compensation EKF-SLAM technology based on the normal distribution transformation(NDT)algorithm is proposed.First,based on the NDT algorithm,use the result of ship position estimation to delay the binocular range measurement value,and finally use the EKF-SLAM technology for data fusion.Experiments prove that the proposed algorithm will meet the positioning accuracy and frequency requirements of the underwater robot in the grasping process.2)Aiming at the kinematic redundancy problem of UVMS target capture,this paper uses two different methods to solve it,one is the solution method based on Jacobian pseudo-inverse,and the other is the redundant capture of UVMS target The problem is transformed into a quadratic programming(QP)problem and solved by a recurrent neural network algorithm.The above two methods are verified by simulation experiments.It shows that if the quadratic programming method is adopted,it can successfully avoid the complicated and time-consuming process of calculating the pseudo-inverse of the Jacobian matrix in UVMS traditional kinematics.3)Aiming at the problem of UVMS motion control,based on quasi-barrier function adaptive non-singular terminal sliding mode controller(QBF-ANTSMC)is proposed.First,the adaptive non-singular terminal sliding mode controller(ANTSMC)is used as the basis for the uncertain parameters of the underwater robot and the unknown upper bound of the unknown disturbance.A quasi-blocking function is introduced to suppress the chattering problem,and the coupling of the manipulator The forces are decoupled using a feedforward strategy.The simulation experiment shows that if the QBF-ANTSMC is used,the control performance will be significantly improved.