An Auto-target-seeking Control Approach For UUV Based On Binocular Vision

Underwater visual system can provide high resolution, high precision and relatively complete external environment information for UUV in close range, hence, it has been widely used in a number of underwater missions such as underwater docking, docking,target tracking, target shooting, navigation and positioning, etc. When UUV assembles an underwater binocular vision system, it can capture and provide high precision visual measurement information of target object like the human eyes. Moreover,if we combine the binocular vision system with an effective visual controller to compose the visual control system, then the whole system can guide and control the UUV to approach to the underwater target object successfully.The paper is motivated by the practical operation requirement that UUV autonomously approaches to underwater target and reaches the operating range of operating tools. Based on such requirement, the study of auto-target-seeking approach for UUV based on binocular vision in close range is carried out. The main contents of this paper are as follows:Firstly, the six degrees of freedom kinematics and dynamics equations of motion of UUV are established. In addition, the added mass, hydrodynamic force (moment) and the restoring force (moment) are analyzed, and the thrust model and thrust allocation are given.Furthermore, the dynamic system properties are discussed.Secondly, the composition of UUV binocular vision system and its assembly on UUV are introduced. The three-dimensional coordinate of feature points of the detected target object is obtained through 3D reconstruction method. Considering that the practical operation position that UUV should reach is not the 3D position of the target object,according to the operation requirement of UUV as well as the constrained relationship between the target object position and the UUV practical operation position, the desired 3D operation position is gotten by coordinate transformation.Then, aiming at the position-given binocular visual control problem of UUV, three kinds of controllers are designed: 1) assuming that the UUV dynamics are precisely known and the external disturbance is absent, a PD plus restoring force compensation control method is proposed, and the stability property of the controller is proven; 2) with the assumption that the upper bounds of the UUV model uncertainties are known and the location of the points where the constant external disturbances act on UUV are known, a robust adaptive P ID control scheme is presented. Furthermore,a theorem which guarantees the stability property of the designed controller is also proposed and analytically proven; 3)in the case that the upper bounds of the UUV model uncertainties are known and the location of the points where the constant external disturbances act on UUV is the geometric center of UUV, a double closed-loop adaptive sliding mode controller is proposed. To completely avoid the chattering effects caused by the discontinuities in the controllers, a chattering-free double closed-loop adaptive sliding mode controller is developed. With regard to the new controller, it removes the need for a priori knowledge of upper bounds on uncertainties in the dynamic parameters of underwater vehicles. Stability and convergence properties of the two controllers are proven by means of Lyapunov stability theory and Barbalat's lemma.Finally, simulation studies are performed to verify the effectiveness of the three developed visual controllers. Simulation case is the visual control of UUV in three-dimensional space. In such case, an UUV autonomously approaches to the operation target using a binocular vision system in the presence of external disturbances and parameter uncertainties. Simulation results demonstrate the rationality and validity of the three visual controllers. Furthermore, the control performances are compared among the three controllers, and the advantages and disadvantages of each controller are analyzed.