Unmanned ship target tracking control method based on sea-air cooperative visual perception
By using stereo vision and hierarchical collaborative control of the UAV-unmanned vessel cooperative system, the problems of unstable visual positioning and differences in dynamic characteristics in target tracking in complex marine environments have been solved, achieving stable and efficient target tracking results.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- DALIAN MARITIME UNIVERSITY
- Filing Date
- 2026-04-16
- Publication Date
- 2026-06-30
Smart Images
Figure CN122308449A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the fields of ship control engineering and unmanned aerial vehicle (UAV) navigation equipment technology, and in particular to a target tracking and control method for unmanned vessels based on sea-air collaborative visual perception. Background Technology
[0002] In dynamic target tracking tasks at sea, unmanned surface vessels (USVs) equipped with visual perception devices have become important carriers for this task due to their high autonomy and wide operating range. However, relying solely on the USV's onboard visual system is susceptible to the influence of complex sea conditions such as wave turbulence, sea-line interference, and changes in lighting, leading to unstable target detection or even target loss. Introducing unmanned aerial vehicles (UAVs) to form a sea-air collaborative platform can utilize their high-altitude, wide-view advantage to compensate for the USV's perception blind spots, significantly improving the reliability and coverage of target tracking. Currently, the academic community has conducted a series of studies on USV-UAV collaborative systems and made certain progress, laying the technical foundation for realizing collaborative operations of heterogeneous sea-air platforms.
[0003] However, most existing unmanned surface vessel-unmanned aerial vehicle (USV) collaborative systems rely on separating the perception module algorithm from the control module algorithm in terms of target tracking, and do not have an integrated design for the entire process from perception to decision-making to optimization. This is crucial for solving how to achieve stable, efficient and autonomous target tracking by unmanned systems in complex marine environments.
[0004] Based on the above analysis, existing unmanned surface vessel-unmanned aerial vehicle (USV) algorithms have the following two main shortcomings in autonomous target tracking tasks: Existing visual positioning methods mostly rely on image information at a single moment or preset reference points, lacking real-time estimation and noise suppression mechanisms for the continuous state of dynamic targets. Under conditions such as target maneuvering, image jitter, or sudden changes in illumination, the raw visual measurement data is difficult to directly generate a smooth and usable reference trajectory, leading to oscillations or even instability in the tracking process. Most existing collaborative control strategies adopt a unified controller, which fails to fully consider the differences in dynamic characteristics and mission requirements of unmanned ships and drones, resulting in vastly different requirements for convergence speed and disturbance resistance. At the same time, the compensation mechanism for model uncertainty and marine environmental disturbance is not yet perfect, which limits the robustness of the system under complex sea conditions. Summary of the Invention
[0005] This invention provides a target tracking and control method for unmanned vessels based on sea-air collaborative visual perception. It overcomes the problem that the differences in dynamic characteristics and mission requirements between unmanned vessels and drones are not fully considered, which leads to different requirements for convergence speed and anti-disturbance capability. Furthermore, the compensation mechanism for model uncertainty and marine environmental disturbance is not yet perfect, which limits the robustness of the system under complex sea conditions.
[0006] To achieve the above objectives, the technical solution of the present invention is as follows: A target tracking and control method for unmanned surface vessels based on sea-air cooperative visual perception includes: S1. Establish a nonlinear mathematical model of the UAV-unmanned vessel cooperative system; S2. The drone's binocular cameras acquire left images from the left camera and right images from the right camera in real time; a dynamic positioning framework for fusing stereo vision and platform pose is constructed, and target detection and coordinate processing are performed on the left and right images to obtain the position coordinates of the fused target. S3. Based on the location coordinates of the fused target, the reference tracking path of the UAV-unmanned vessel cooperative system is obtained through multiple optimization and path generation mechanisms. S4. Based on the reference tracking path of the UAV-UAV cooperative system and the nonlinear mathematical model of the UAV-UAV cooperative system, the control law of the UAV and the control law and attitude reference signal of the UAV are obtained through the hierarchical cooperative control strategy. S5. The UAV-UAV cooperative system is controlled by the control laws of the unmanned surface vessel, the control laws of the UAV, and the attitude reference signals to achieve the target tracking task.
[0007] Furthermore, the expression for the nonlinear mathematical model of the UAV-unmanned vessel cooperative system is as follows: (1) (2) (3) Among them, Equation (1) and Equation (2) are nonlinear mathematical models of the UAV-unmanned ship cooperative system, Equation (1) is the kinematic model, Equation (2) is the dynamic model, and Equation (3) is the expansion formula of some variables in Equation (2); In the formula, The positions of the unmanned vessel in the geodetic coordinate system are respectively Axis coordinates Axial coordinates and heading angle; Let be the velocity vector of the unmanned surface vessel, and ,in, The forward speed of the unmanned vessel. The drift speed of the unmanned boat. The bow roll rate of the unmanned vessel; The unmanned ship in three-dimensional coordinate space Axis coordinates Axis coordinates Axis coordinates, roll angle, pitch angle, and yaw angle; Let be the velocity vector of the drone, and ,in, For drones along The displacement velocity of the shaft, For drones along The displacement velocity of the shaft, For drones along The displacement velocity of the shaft, The pitch speed represents the attitude of the drone. The roll speed is the attitude of the drone. The yaw speed represents the attitude of the drone. For the unknown nonlinear damping term of the unmanned vessel; For the corresponding known actuator performance coefficients; The engine rotation speed of the unmanned vessel; The rudder angle for the unmanned vessel; These represent the environmental disturbance forces experienced by the unmanned vessel during forward motion, the environmental disturbance forces experienced by the unmanned vessel during lateral drift motion, and the environmental disturbance forces experienced by the unmanned vessel during bow turning motion. For the nonlinear damping term of the UAV; For the quality of the drone; In three-dimensional space axis, axis, Rotation matrix corresponding to the axial direction; It is the acceleration due to gravity; The total lift generated by the drone's rotor; These are the environmental interference forces experienced by the UAV during forward motion, during drift motion, during yaw motion, during pitch motion, during roll motion, and during yaw motion, respectively. Let be the diagonal moment of the UAV rotor; The rotational torque generated by the difference in rotor speed of the UAV; respectively drones orbit The moment of inertia of the axis, the drone's orbit The moment of inertia of the axis, the drone's orbit Moment of inertia of the shaft; The added mass of the unmanned vessel during its forward motion; The added mass during the drifting motion of the unmanned vessel; The added mass during the bow-turning motion of the unmanned vessel; This represents the moment of inertia of the drone's propeller blades about their own axis of rotation. This is the coefficient of rotational resistance. Let be the rotor angular velocity, and ; This represents the first-order hydrodynamic parameter in the forward velocity degree of freedom of the unmanned vessel. The second-order hydrodynamic parameters represent the forward velocity degree of freedom of the unmanned vessel. The third-order hydrodynamic parameters represent the forward velocity degree of freedom of the unmanned vessel. The first-order hydrodynamic parameter represents the degree of freedom of the unmanned vessel's drift velocity. The second-order hydrodynamic parameters represent the degree of freedom of the unmanned vessel's drift velocity. The third-order hydrodynamic parameters represent the degree of freedom of the unmanned vessel's drift velocity. The first-order hydrodynamic parameter represents the degree of freedom of the unmanned vessel's bow turning speed; The second-order hydrodynamic parameters represent the degree of freedom of the unmanned vessel's bow turning speed; The third-order hydrodynamic parameters represent the degree of freedom of the unmanned vessel's bow turning speed.
[0008] Furthermore, the execution steps of the dynamic positioning framework that fuses stereo vision with platform pose include: S21. Based on the left image obtained from the left camera and the right image obtained from the right camera, a target recognition algorithm is used to identify the target and obtain the target pixel coordinates in the left image and the target pixel coordinates in the right image. S22. Based on the target pixel coordinates in the left image and the target pixel coordinates in the right image, the precise disparity of the target center point in the left and right images is calculated using the SGBM algorithm. S23. Using the left image as the main image, based on the target pixel coordinates in the left image and the precise parallax of the target center points in the left and right images, obtain the camera coordinate system coordinates of the target through coordinate transformation. S24. Based on the target's camera coordinate system coordinates, combined with the UAV's own attitude and position information, the target's geodetic coordinate system coordinates are obtained through coordinate transformation equations, i.e., the fused target's position coordinates.
[0009] Furthermore, the execution steps of the multiple optimization and path generation mechanism include: S31. Simplify the position coordinates of the target to be fused into two-dimensional planar coordinates; S32. Based on the two-dimensional plane coordinates, noise reduction and state estimation are performed using an extended Kalman filter to obtain the preliminary accurate positioning coordinates of the target; and the heading angle of the target is calculated based on the preliminary accurate positioning coordinates of the target; the preliminary accurate positioning coordinates of the target and the heading angle of the target are combined to obtain the preliminary optimized target position; S33. Based on the preliminary optimized target position, the smoothed target position is obtained through a moving average filter, expressed as: (4) In the formula, For the first The target position after optimization by the moving average filter at any given time, i.e., the smoothed target position; This represents the window width of the moving average filter; This is the current sampling time; The target location is for preliminary optimization; S34. Based on the smoothed target position, continuous reference tracking trajectory points are obtained through a cubic spline interpolation function, expressed as: (5) In the formula, For reference tracking trajectory points; These are parameters to be determined. System uptime; For the first Time to the The interval of system runtime at any given moment; For the first System runtime at any given moment; For the first System runtime at any given moment; S35. Combine continuous reference tracking trajectory points to obtain a reference tracking trajectory; map the reference tracking trajectory using a virtual mapping method to obtain the reference tracking path of the UAV-unmanned vessel cooperative system.
[0010] Furthermore, the specific steps to obtain the control law of the unmanned ship are as follows: S411. Based on the UAV-UAV cooperative system's reference tracking path's UAV position information and the UAV's position in the geodetic coordinate system, define the UAV's kinematic error as follows: (6) In the formula, For unmanned ships Shaft error; For the location information of the unmanned vessel Axis coordinates; For unmanned ships Shaft error; For the location information of the unmanned vessel Axis coordinates; The Euclidean distance is the positional error of the unmanned vessel. This refers to the heading angle error of the unmanned vessel; The heading angle is the position information of the unmanned vessel. S412. Differentiate the Euclidean distance of the unmanned vessel's position error with the bow angle error of the unmanned vessel to obtain the following expression; (7) In the formula, for The derivative; As an auxiliary variable, and ; for The derivative; for The derivative; S413. Based on equations (6) and (7), a virtual control law for the forward motion and turning motion of the unmanned vessel is designed, and a first-order filter is used to process the virtual control law for the forward motion and turning motion of the unmanned vessel to obtain the output of the first-order filter; the expression of the virtual control law is: (8) In the formula, A virtual control law for forward motion; For the virtual control law of bow-rolling motion; and To design control parameters; The expression for processing the virtual control law of forward motion and turning motion using a first-order filter is as follows: (9) In the formula, This is the filtering time; This is the filtering error; The expected velocity reference of the unmanned vessel is the output of the first-order filter; for The derivative; It is an index variable, and , among which, when At time, the forward degree of freedom of the unmanned vessel is... At that time, the bow roll angle is the degree of freedom of the unmanned vessel; It is a virtual control law, and ; S414. Based on the output of the first-order filter and the velocities of the unmanned vessel in each direction, define the dynamic error of the unmanned vessel, and differentiate the dynamic error to obtain the following expression: (10) In the formula, For the dynamic error of the unmanned vessel, and ,in, The speed of the unmanned vessel in each direction; for The derivative; For the direct control law of the unmanned vessel, and when hour, ,when hour, ; for The derivative; The added mass of the unmanned vessel during its motion; The environmental disturbance forces experienced by the unmanned vessel during its movement; For the unknown nonlinear damping term of the unmanned vessel; S415. By approximating equation (10) using a fuzzy logic system, the following expression is obtained: (11) In the formula, The highest membership degree of the fuzzy logic system serving the unmanned vessel; The fuzzy basis function vector serving the fuzzy logic system of the unmanned vessel; To serve the fitting error of the fuzzy logic system for unmanned ships, and ,in, Unknown constants for the fuzzy logic system serving unmanned ships; Design parameters for the fuzzy logic system serving unmanned ships; The upper bound of the fitting error for the fuzzy logic system serving unmanned ships; S416. Introduce the following disturbance observer into the dynamics controller of the unmanned vessel: (12) In the formula, This is to provide the interference observer's observation of the environmental disturbance forces experienced by the unmanned vessel during its movement; for The derivative; As an auxiliary variable; for The derivative; Design variables; S417. Combining equations (10), (11), and (12), the final control law of the unmanned vessel is obtained through the backstepping framework, and its expression is: (13) In the formula, and For design parameters; These are the adaptive parameters for the unmanned surface vessel; The derivative of the adaptive parameters of the unmanned vessel is the adaptive law of the final control law of the unmanned vessel.
[0011] Furthermore, the specific steps for obtaining the control law and attitude reference signal of the UAV are as follows: S421. Based on the UAV's position information in the reference tracking path of the UAV-Unmanned Vessel Cooperative System and the UAV's position in the geodetic coordinate system, the kinematic error of the UAV is defined as follows: (14) In the formula, For the kinematic errors of the drone; This represents the position of the UAV in the geodetic coordinate system. The location information of the UAV in the reference tracking path of the UAV-unmanned vessel cooperative system; Let be the index variable of the UAV's spatial coordinate axes, and ,when At that time, for drones Degrees of freedom in the axial direction, when At that time, for drones Degrees of freedom in the axial direction, when At that time, for drones Degrees of freedom in the axial direction; S422. Differentiating equation (14), we obtain the following expression: (15) In the formula, for The derivative; The displacement velocity of the UAV along the spatial coordinate axes; for The derivative; S423. Based on equations (14) and (15), a virtual control law for the UAV's motion along the spatial coordinate axis is designed, and a first-order filter is used to process the virtual control law for the UAV's motion along the spatial coordinate axis to obtain the output of the first-order filter; the expression of the virtual control law is: (16) In the formula, The virtual control law for the motion of the UAV along the spatial coordinate axes; , , and For design parameters, and ; The expression for processing the virtual control law of the UAV's motion along the spatial coordinate axes using a first-order filter is as follows: (17) In the formula, This is the filtering time; This is the filtering error; The expected speed reference for the UAV is the output of the first-order filter; for The derivative; S424. Based on the output of the first-order filter and the displacement velocity of the UAV along the spatial coordinate axes, define the dynamic error of the UAV, and differentiate the dynamic error of the UAV to obtain the following expression: (18) In the formula, This represents the dynamic error of the drone, and ,in, The displacement velocity of the UAV along the spatial coordinate axes; for The derivative; For the unknown nonlinear damping term of the UAV; Rotation matrices corresponding to each axis direction in three-dimensional space; The environmental disturbance forces experienced by the unmanned vessel during its movement; S425. By approximating equation (18) using a fuzzy logic system, the following expression is obtained: (19) In the formula, The highest membership degree of the fuzzy logic system serving the drone; Fuzzy basis function vectors for fuzzy logic systems serving unmanned aerial vehicles; To address the fitting error of the fuzzy logic system serving the unmanned aerial vehicle (UAV), and ,in, Unknown constants for fuzzy logic systems serving unmanned aerial vehicles; Design parameters for fuzzy logic systems serving unmanned aerial vehicles; The upper bound of the fitting error for the fuzzy logic system serving unmanned aerial vehicles; S426. Introduce the following disturbance observer into the dynamics controller of the UAV: (20) In the formula, This refers to the observation value of the environmental interference force experienced by the UAV during its movement by the interference observer; for The derivative; As an auxiliary variable; for The derivative; , , Design variables; S427, combining equations (18), (19), and (20), the final control law of the UAV is obtained through the inverse stepping framework, and its expression is: (twenty one) In the formula, This is the final control law for the drone; The rate of change of the expected speed of the drone after smoothing; For design parameters; These are the adaptive parameters for the drone; The derivative of the adaptive parameters of the UAV is the adaptive law of the final control law of the UAV. S428. Based on the final control law of the UAV, the attitude reference signal of the UAV is obtained through a nonlinear decoupling method, and its expression is: (twenty two) In the formula, The desired pitch angle for the drone; The desired roll angle for the drone; The desired yaw angle for the drone; for The final control law in the axial direction; for The final control law in the axial direction; for The final control law in the axial direction.
[0012] Beneficial effects: This invention uses a binocular camera mounted on a drone to collect left and right images in real time, and combines stereo vision with a dynamic positioning framework that fuses platform pose, to accurately obtain the three-dimensional position coordinates of the target, effectively overcoming the problems of low positioning accuracy and susceptibility to interference in complex marine environments caused by single vision or single platform. By employing multiple optimization and path generation mechanisms to process target position coordinates, a smooth, feasible reference tracking path that satisfies motion constraints can be generated for the UAV-UAV cooperative system, significantly improving the system's path adaptability and tracking efficiency in dynamic environments. To address the heterogeneous dynamic characteristics of UAVs and unmanned surface vessels (USVs), a hierarchical collaborative control strategy is designed to generate control laws for USVs and control laws and attitude reference signals for UAVs, respectively. This enables efficient collaborative operations between air and sea platforms and ensures stability and consistency during target tracking. Attached Figure Description
[0013] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0014] Figure 1 This is a flowchart illustrating the target tracking control method of the present invention; Figure 2 This is a schematic diagram of the target three-dimensional visual positioning in an embodiment of the present invention; Figure 3 This is a schematic diagram illustrating the feasible guidance trajectory generation in an embodiment of the present invention; Figure 4 This is a schematic diagram of the three-dimensional path following trajectory of the UAV-unmanned vessel cooperative system in an embodiment of the present invention; Figure 5 This is a schematic diagram of the kinematic error analysis of the unmanned vessel in target tracking in an embodiment of the present invention; Figure 6 This is a schematic diagram of the UAV tracking error obtained by the UAV-unmanned vessel cooperative system in an embodiment of the present invention; Figure 7 This is a real-time 3D target localization map based on UAV stereo vision in an embodiment of the present invention; Figure 8 This is a diagram showing the flight trajectory of an unmanned surface vessel (USV) and an unmanned aerial vehicle (UAV) during a real-ship experiment in an embodiment of the present invention. Figure 9 This is a schematic diagram illustrating the tracking error of a UAV against an unmanned surface vessel in an embodiment of the present invention. Figure 10 This is a diagram showing the target localization result in an embodiment of the present invention. Detailed Implementation
[0015] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0016] This embodiment provides a target tracking and control method for unmanned surface vessels based on sea-air cooperative visual perception, such as... Figure 1 As shown, it includes: S1. Establish a nonlinear mathematical model of the UAV-unmanned vessel cooperative system; The nonlinear mathematical model of the UAV-unmanned vessel cooperative system serves as the controlled object of the controller designed in subsequent steps. Preferably, the expression for the nonlinear mathematical model of the UAV-unmanned vessel cooperative system is: (1) (2) (3) Among them, Equation (1) and Equation (2) are nonlinear mathematical models of the UAV-unmanned ship cooperative system, Equation (1) is the kinematic model, Equation (2) is the dynamic model, and Equation (3) is the expansion formula of some variables in Equation (2); In the formula, The positions of the unmanned vessel in the geodetic coordinate system are respectively Axis coordinates Axial coordinates and heading angle; Let be the velocity vector of the unmanned surface vessel, and ,in, The forward speed of the unmanned vessel. The drift speed of the unmanned boat. The bow roll rate of the unmanned vessel; The unmanned ship in three-dimensional coordinate space Axis coordinates Axis coordinates Axis coordinates, roll angle, pitch angle, and yaw angle; Let be the velocity vector of the drone, and ,in, For drones along The displacement velocity of the shaft, For drones along The displacement velocity of the shaft, For drones along The displacement velocity of the shaft, The pitch speed represents the attitude of the drone. The roll speed is the attitude of the drone. The yaw speed represents the attitude of the drone. For the unknown nonlinear damping term of the unmanned vessel; For the corresponding known actuator performance coefficients; The engine rotation speed of the unmanned vessel; The rudder angle for the unmanned vessel; These represent the environmental disturbance forces experienced by the unmanned vessel during forward motion, the environmental disturbance forces experienced by the unmanned vessel during lateral drift motion, and the environmental disturbance forces experienced by the unmanned vessel during bow turning motion. For the nonlinear damping term of the UAV; For the quality of the drone; In three-dimensional space axis, axis, Rotation matrix corresponding to the axial direction; It is the acceleration due to gravity; The total lift generated by the drone's rotor; These are the environmental interference forces experienced by the UAV during forward motion, during drift motion, during yaw motion, during pitch motion, during roll motion, and during yaw motion, respectively. Let be the diagonal moment of the UAV rotor; The rotational torque generated by the difference in rotor speed of the UAV; respectively drones orbit The moment of inertia of the axis, the drone's orbit The moment of inertia of the axis, the drone's orbit Moment of inertia of the shaft; The added mass of the unmanned vessel during its forward motion; The added mass during the drifting motion of the unmanned vessel; The added mass during the bow-turning motion of the unmanned vessel; This represents the moment of inertia of the drone's propeller blades about their own axis of rotation. This is the coefficient of rotational resistance. Let be the rotor angular velocity, and ; This represents the first-order hydrodynamic parameter in the forward velocity degree of freedom of the unmanned vessel. The second-order hydrodynamic parameters represent the forward velocity degree of freedom of the unmanned vessel. The third-order hydrodynamic parameters represent the forward velocity degree of freedom of the unmanned vessel. The first-order hydrodynamic parameter represents the degree of freedom of the unmanned vessel's drift velocity. The second-order hydrodynamic parameters represent the degree of freedom of the unmanned vessel's drift velocity. The third-order hydrodynamic parameters represent the degree of freedom of the unmanned vessel's drift velocity. The first-order hydrodynamic parameter represents the degree of freedom of the unmanned vessel's bow turning speed; The second-order hydrodynamic parameters represent the degree of freedom of the unmanned vessel's bow turning speed; The third-order hydrodynamic parameters represent the degree of freedom of the unmanned vessel's bow turning speed.
[0017] S2. The drone's binocular cameras acquire left images from the left camera and right images from the right camera in real time; a dynamic positioning framework for fusing stereo vision and platform pose is constructed, and target detection and coordinate processing are performed on the left and right images to obtain the position coordinates of the fused target. In this embodiment, the drone is equipped with two cameras: a left camera located on the left side of the drone and a right camera located on the right side of the drone.
[0018] Preferred, such as Figure 2 As shown, the execution steps of the dynamic positioning framework that fuses stereo vision with platform pose include: S21. Based on the left image obtained from the left camera and the right image obtained from the right camera, a target recognition algorithm is used to identify the target and obtain the target pixel coordinates in the left image and the target pixel coordinates in the right image. S22. Based on the target pixel coordinates in the left image and the target pixel coordinates in the right image, the precise disparity of the target center point in the left and right images is calculated using the SGBM algorithm. S23. Using the left image as the main image, based on the target pixel coordinates in the left image and the precise parallax of the target center points in the left and right images, obtain the camera coordinate system coordinates of the target through coordinate transformation. Specifically, the expression for the coordinate transformation is: (4) In the formula, The coordinates of the target in the camera coordinate system; This is the horizontal baseline distance between the left and right cameras after stereo correction; This corresponds to the focal length of the left camera axis; This corresponds to the focal length of the left camera axis; The pixel coordinates of the center of the left image; These are the pixel coordinates of the target in the left image; The precise parallax of the center points of the target in the left and right images.
[0019] S24. Based on the target's camera coordinate system coordinates, combined with the UAV's own attitude and position information, the target's geodetic coordinate system coordinates are obtained through coordinate transformation equations, i.e., the fused target's position coordinates. Specifically, the expression for the coordinate transformation equation is: (5) In the formula, The geodetic coordinates of the target; Let be the set of rotation matrices corresponding to 3D space, and ; The distance between the drone's current position and the origin of the coordinate system is given. ,in, The current position of the drone relative to the origin of the coordinate system. Axial distance; The current position of the drone relative to the origin of the coordinate system. Axial distance; The current position of the drone relative to the origin of the coordinate system. Axial distance.
[0020] In this embodiment, the target recognition algorithm is the YOLOv5 algorithm, which realizes real-time detection of dynamic targets by acquiring the left and right image streams from the stereo camera of the UAV; the SGBM algorithm is a stereo matching method that calculates the precise parallax of the target center point through correction and dense matching; the YOLOv5 algorithm and the SGBM algorithm are existing technologies and will not be described in detail here.
[0021] S3. Based on the location coordinates of the fused target, the reference tracking path of the UAV-unmanned vessel cooperative system is obtained through multiple optimization and path generation mechanisms. Preferably, the execution steps of the multiple optimization and path generation mechanism include: S31. Simplify the position coordinates of the target to be fused into two-dimensional planar coordinates; Specifically, remove the location coordinates of the target to be merged. The z-axis coordinates are used to obtain the two-dimensional plane coordinates. This is to ensure that the target and the unmanned vessel are on the same horizontal plane.
[0022] S32. Based on the two-dimensional plane coordinates, noise reduction and state estimation are performed using an extended Kalman filter to obtain the preliminary accurate positioning coordinates of the target; and the heading angle of the target is calculated based on the preliminary accurate positioning coordinates of the target; the preliminary accurate positioning coordinates of the target and the heading angle of the target are combined to obtain the preliminary optimized target position; Specifically, the expression for the extended Kalman filter is: (6) In the formula, The state variables of the extended Kalman filter; that is, the state variables of the motion model of the sea surface target established in the filter, their initial values will be set to 0, and their values will be updated in each iteration step; This is the state transition matrix; Let be the observation vector, and ,in, The target's two-dimensional plane coordinates at the current sampling time; The observation matrix; The state noise of the extended Kalman filter has the following covariance: ; The covariance of the observation noise of the extended Kalman filter is: ; This is the current sampling time; The expression for noise reduction and state estimation using the extended Kalman filter is: (7) In the formula, These are the covariance matrix of the state prediction, the covariance matrix of the prediction estimate, and the filter gain matrix, respectively. Preliminary and precise coordinates for the target; It is the identity matrix; The expression for calculating the target's heading angle based on its preliminary precise positioning coordinates is as follows: (8) In the formula, The heading angle of the target; The initial precise location coordinates of the target at the current sampling time; These are the preliminary, precise coordinates of the target at the previous sampling time.
[0023] S33. Based on the preliminary optimized target position, the smoothed target position is obtained through a moving average filter, expressed as: (9) In the formula, For the first The target position after optimization by the moving average filter at any given time, i.e., the smoothed target position; This represents the window width of the moving average filter; This is the current sampling time; The target location is for preliminary optimization; S34. Based on the smoothed target position, continuous reference tracking trajectory points are obtained through a cubic spline interpolation function, expressed as: (10) In the formula, For reference tracking trajectory points; These are parameters to be determined. System uptime; For the first Time to the The interval of system runtime at any given moment; For the first System runtime at any given moment; For the first System runtime at any given moment; S35. Combine continuous reference tracking trajectory points to obtain a reference tracking trajectory; map the reference tracking trajectory using a virtual mapping method to obtain the reference tracking path of the UAV-unmanned vessel cooperative system; Specifically, the mapping of the reference tracking trajectory using the virtual mapping method is expressed as follows: (11) In the formula, The position information of the unmanned vessel in the reference tracking path of the UAV-unmanned vessel cooperative system; and These are mapping parameters that can be adjusted appropriately based on the specific tracking target. and This refers to the positional error of the unmanned vessel. For reference tracking trajectory, the first One coordinate; Meanwhile, to achieve stable tracking, the target plane attitude of the UAV is defined based on the actual attitude of the UAV, thus obtaining the UAV's position information in the reference tracking path of the UAV-UAV cooperative system. And manually set the cruising altitude. ; The reference tracking path of the UAV-UAV cooperative system is obtained by combining the position information of the UAV in the reference tracking path of the UAV-UAV cooperative system with the position information of the UAV in the reference tracking path of the UAV-UAV cooperative system.
[0024] S4. Based on the reference tracking path of the UAV-UAV cooperative system and the nonlinear mathematical model of the UAV-UAV cooperative system, the control law of the UAV and the control law and attitude reference signal of the UAV are obtained through the hierarchical cooperative control strategy. Preferably, the specific steps to obtain the control law of the unmanned vessel are as follows: S411. Based on the UAV-UAV cooperative system's reference tracking path's UAV position information and the UAV's position in the geodetic coordinate system, define the UAV's kinematic error as follows: (12) In the formula, For unmanned ships Shaft error; For the location information of the unmanned vessel Axis coordinates; For unmanned ships Shaft error; For the location information of the unmanned vessel Axis coordinates; The Euclidean distance is the positional error of the unmanned vessel. This refers to the heading angle error of the unmanned vessel; The heading angle is the position information of the unmanned vessel. S412. Differentiate the Euclidean distance of the unmanned vessel's position error with the bow angle error of the unmanned vessel to obtain the following expression; (13) In the formula, for The derivative; As an auxiliary variable, and ; for The derivative; for The derivative; S413. Based on equations (12) and (13), a virtual control law for the forward motion and turning motion of the unmanned vessel is designed, and a first-order filter is used to process the virtual control law for the forward motion and turning motion of the unmanned vessel to obtain the output of the first-order filter; the expression of the virtual control law is: (14) In the formula, A virtual control law for forward motion; For the virtual control law of bow-rolling motion; and To design control parameters; The expression for processing the virtual control law of forward motion and turning motion using a first-order filter is as follows: (15) In the formula, This is the filtering time; This is the filtering error; The expected velocity reference of the unmanned vessel is the output of the first-order filter; for The derivative; It is an index variable, and , among which, when At time, the forward degree of freedom of the unmanned vessel is... At that time, the bow roll angle is the degree of freedom of the unmanned vessel; It is a virtual control law, and ; S414. Based on the output of the first-order filter and the velocities of the unmanned vessel in each direction, define the dynamic error of the unmanned vessel, and differentiate the dynamic error to obtain the following expression: (16) In the formula, For the dynamic error of the unmanned vessel, and ,in, The speed of the unmanned vessel in each direction; for The derivative; For the direct control law of the unmanned vessel, and when hour, ,when hour, ; for The derivative; The added mass of the unmanned vessel during its motion; The environmental disturbance forces experienced by the unmanned vessel during its movement; For the unknown nonlinear damping term of the unmanned vessel; S415. By approximating equation (16) using a fuzzy logic system, the following expression is obtained: (17) In the formula, The highest membership degree of the fuzzy logic system serving the unmanned vessel; The fuzzy basis function vector serving the fuzzy logic system of the unmanned vessel; To serve the fitting error of the fuzzy logic system for unmanned ships, and ,in, Unknown constants for the fuzzy logic system serving unmanned ships; Design parameters for the fuzzy logic system serving unmanned ships; The upper bound of the fitting error for the fuzzy logic system serving unmanned ships; S416. Introduce the following disturbance observer into the dynamics controller of the unmanned vessel: (18) In the formula, This is to provide the interference observer's observation of the environmental disturbance forces experienced by the unmanned vessel during its movement; for The derivative; As an auxiliary variable; for The derivative; Design variables; S417, combining equations (16), (17), and (18), the final control law of the unmanned vessel is obtained through the backstepping framework, and its expression is: (19) In the formula, and For design parameters; These are the adaptive parameters for the unmanned surface vessel; The derivative of the adaptive parameters of the unmanned vessel is the adaptive law of the final control law of the unmanned vessel. Based on the above unmanned vessel controller design, it can be proved through Lyapunov functions that all relevant error variables in the unmanned vessel control system have semi-globally uniform boundedness, which means that the relevant error signals in the system can achieve asymptotic stability. In the unmanned vessel path following system, the asymptotically stable controller can achieve reliable tracking of guidance signals under model uncertainty and environmental disturbance conditions, while avoiding the problem of excessive control force caused by overly complex controller structure. This is beneficial for continuous and stable tracking of dynamic targets on the sea surface.
[0025] Preferably, the specific steps for obtaining the control law and attitude reference signal of the UAV are as follows: S421. Based on the UAV's position information in the reference tracking path of the UAV-Unmanned Vessel Cooperative System and the UAV's position in the geodetic coordinate system, the kinematic error of the UAV is defined as follows: (20) In the formula, For the kinematic errors of the drone; This represents the position of the UAV in the geodetic coordinate system. The location information of the UAV in the reference tracking path of the UAV-unmanned vessel cooperative system; Let be the index variable of the UAV's spatial coordinate axes, and ,when At that time, for drones Degrees of freedom in the axial direction, when At that time, for drones Degrees of freedom in the axial direction, when At that time, for drones Degrees of freedom in the axial direction; S422. Differentiating equation (20), we obtain the following expression: (twenty one) In the formula, for The derivative; The displacement velocity of the UAV along the spatial coordinate axes; for The derivative; S423. Based on equations (20) and (21), a virtual control law for the UAV's motion along the spatial coordinate axes is designed, and a first-order filter is used to process the virtual control law for the UAV's motion along the spatial coordinate axes to obtain the output of the first-order filter; the expression of the virtual control law is: (twenty two) In the formula, The virtual control law for the motion of the UAV along the spatial coordinate axes; , , and For design parameters, and ; The expression for processing the virtual control law of the UAV's motion along the spatial coordinate axes using a first-order filter is as follows: (twenty three) In the formula, This is the filtering time; This is the filtering error; The expected speed reference for the UAV is the output of the first-order filter; for The derivative; S424. Based on the output of the first-order filter and the displacement velocity of the UAV along the spatial coordinate axes, define the dynamic error of the UAV, and differentiate the dynamic error of the UAV to obtain the following expression: (twenty four) In the formula, This represents the dynamic error of the drone, and ,in, The displacement velocity of the UAV along the spatial coordinate axes; for The derivative; For the unknown nonlinear damping term of the UAV; Rotation matrices corresponding to each axis direction in three-dimensional space; The environmental disturbance forces experienced by the unmanned vessel during its movement; S425. By approximating equation (24) using a fuzzy logic system, the following expression is obtained: (25) In the formula, The highest membership degree of the fuzzy logic system serving the drone; Fuzzy basis function vectors for fuzzy logic systems serving unmanned aerial vehicles; To address the fitting error of the fuzzy logic system serving the unmanned aerial vehicle (UAV), and ,in, Unknown constants for fuzzy logic systems serving unmanned aerial vehicles; Design parameters for fuzzy logic systems serving unmanned aerial vehicles; The upper bound of the fitting error for the fuzzy logic system serving unmanned aerial vehicles; S426. Introduce the following disturbance observer into the dynamics controller of the UAV: (26) In the formula, This refers to the observation value of the environmental interference force experienced by the UAV during its movement by the interference observer; for The derivative; As an auxiliary variable; for The derivative; , , Design variables; S427, combining equations (24), (25), and (26), the final control law of the UAV is obtained through the inverse stepping framework, and its expression is: (27) In the formula, This is the final control law for the drone; The rate of change of the expected speed of the drone after smoothing; For design parameters; These are the adaptive parameters for the drone; The derivative of the adaptive parameters of the UAV is the adaptive law of the final control law of the UAV. S428. Based on the final control law of the UAV, the attitude reference signal of the UAV is obtained through a nonlinear decoupling method, and its expression is: (28) In the formula, The desired pitch angle for the drone; The desired roll angle for the drone; The desired yaw angle for the drone; for The final control law in the axial direction; for The final control law in the axial direction; for The final control law in the axial direction; Based on the above UAV controller design, it can be proved through Lyapunov functions that the UAV control closed-loop system will achieve fixed-time stability, and the tracking error will converge to the residual set within a fixed time. For the UAV-UAV cooperative system, a fixed-time convergence controller is adopted to enable the UAV to reach and maintain the required relative position in a short time even after being disturbed and deviated. This can significantly improve the reliability and stability of the cooperative visual perception system.
[0026] S5. The UAV-UAV cooperative system is controlled by the control laws of the unmanned surface vessel, the control laws of the UAV, and the attitude reference signals to achieve the target tracking task.
[0027] In this embodiment, in order to evaluate the unmanned vessel target tracking and control method based on sea-air cooperative visual perception proposed in this embodiment, numerical simulation experiments and closed water area experiments will be conducted to verify its effectiveness. First, simulation experiments were conducted to model the environmental disturbances caused by sea winds and waves based on the NORSOK spectrum and JONSWAP spectrum, and these models were used to simulate the marine environmental disturbances of the UAV-unmanned vessel cooperative system in the simulation experiment. like Figure 3 As shown, the method is based on waypoints. Generate the target ship's trajectory and maintain a constant speed. To simulate perceptual error, Gaussian noise (variance) is added to the measured values of the visual signal. ); After obtaining the reference tracking trajectory and the unmanned vessel's heading, mapping parameters are set. cruising altitude and The reference path of the system is generated; to compare and evaluate the control law of the UAV, a sliding mode control-based unmanned surface vessel and a UAV cooperative motion control algorithm are used for comparative analysis. like Figure 4The simulation of the three-dimensional tracking trajectory is shown. During the process of the UAV following the unmanned vessel, the unmanned vessel's trembling trajectory due to marine environmental interference and control overshoot poses a challenge to the UAV's ability to maintain a stable relative position. Compared with the unmanned surface vessel and UAV cooperative motion control algorithm based on sliding mode control, the UAV control law proposed in this embodiment achieves faster response speed, higher accuracy, and smaller overshoot due to its nonlinear feedback characteristics and fixed-time convergence characteristics, thereby significantly improving the actual tracking stability. Figure 5 and Figure 6 The kinematic position and attitude errors of the UAV-Unmanned Ship cooperative system during target tracking are demonstrated. In the initial stage, the position control of the Unmanned Ship exhibits transient peaks due to acceleration and overshoot effects, which then tend to stabilize within an acceptable narrow range. In the kinematic control of the UAV, the error signals of each axis show small oscillations. These oscillations originate from the continuous influence of visual perception noise during the reference path generation process, and are further amplified during the movement of the Unmanned Ship. The designed final control law for the UAV and Unmanned Ship can suppress the resulting errors to a satisfactory level. Subsequently, a real-ship experiment was conducted in static waters to verify the effectiveness of the proposed sea-air collaborative visual perception method. The experiment deployed a collaborative platform consisting of an unmanned vessel and a drone; as the platform navigated along a preset route, it performed real-time visual perception of fixed navigation buoys in the water (with known RTK-calibrated ground coordinates); During the experiment, the system's real-time attitude data was transmitted to the shore-based collaborative information processing unit via visual data transmission, and after fusion processing, it was used for target localization. Simultaneously, the unmanned surface vessel's position and attitude data generated reference signals for the UAV. The verification of the sea-air collaborative visual perception method used the YOLOv5 training dataset, which contains real target images captured by the UAV's onboard camera. The target detection model, based on the PyTorch deep learning framework and the YOLOv5 architecture, was trained on a computer equipped with an NVIDIA GeForce GTX 1660Ti GPU. The sea-air collaborative visual perception capability of the UAV-UAV collaborative system will be verified by the accuracy of dynamically locating fixed targets. The stereo camera target localization effect is as follows: Figure 7 As shown; Experimental results are as follows Figures 8 to 10 As shown; specifically, Figure 8 The experiment demonstrates the motion trajectory of the unmanned surface vessel (USV) along a fixed heading, as well as the trajectory of a drone tracking the USV in two-dimensional flight. The drone tracking error during the experiment is shown as follows: Figure 9 As shown; Figure 7 The middle border indicates the range within which the stereo camera can capture the target buoy; within this area, the localization performance of the proposed visual perception system is as follows: Figure 9 and Figure 10 As shown, the maximum positional error reached 2.23 meters; this error is partly due to camera resolution limitations, and drone motion jitter may also be one of the influencing factors.
[0028] The present invention has the following beneficial effects: This invention uses a binocular camera mounted on a drone to collect left and right images in real time, and combines stereo vision with a dynamic positioning framework that fuses platform pose. This can accurately obtain the three-dimensional position coordinates of the target, effectively overcoming the problems of low positioning accuracy and susceptibility to interference in complex marine environments caused by single vision or single platform. By employing multiple optimization and path generation mechanisms to process target position coordinates, a smooth, feasible reference tracking path that satisfies motion constraints can be generated for the UAV-UAV cooperative system, significantly improving the system's path adaptability and tracking efficiency in dynamic environments. To address the heterogeneous dynamic characteristics of UAVs and unmanned surface vessels (USVs), a hierarchical collaborative control strategy is designed to generate control laws for USVs and control laws and attitude reference signals for UAVs, respectively. This enables efficient collaborative operations between air and sea platforms and ensures stability and consistency during target tracking.
[0029] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.
Claims
1. A target tracking and control method for unmanned surface vessels based on sea-air cooperative visual perception, characterized in that, include: S1. Establish a nonlinear mathematical model of the UAV-unmanned vessel cooperative system; S2. The drone's binocular cameras acquire left images from the left camera and right images from the right camera in real time; a dynamic positioning framework for fusing stereo vision and platform pose is constructed, and target detection and coordinate processing are performed on the left and right images to obtain the position coordinates of the fused target. S3. Based on the location coordinates of the fused target, the reference tracking path of the UAV-unmanned vessel cooperative system is obtained through multiple optimization and path generation mechanisms. S4. Based on the reference tracking path of the UAV-UAV cooperative system and the nonlinear mathematical model of the UAV-UAV cooperative system, the control law of the UAV and the control law and attitude reference signal of the UAV are obtained through the hierarchical cooperative control strategy. S5. The UAV-UAV cooperative system is controlled by the control laws of the unmanned surface vessel, the control laws of the UAV, and the attitude reference signals to achieve the target tracking task.
2. The unmanned surface vessel target tracking and control method based on sea-air cooperative visual perception according to claim 1, characterized in that, The expression for the nonlinear mathematical model of the UAV-unmanned vessel cooperative system is as follows: （1） （2） （3） Among them, Equation (1) and Equation (2) are nonlinear mathematical models of the UAV-unmanned ship cooperative system, Equation (1) is the kinematic model, Equation (2) is the dynamic model, and Equation (3) is the expansion formula of some variables in Equation (2); In the formula, The positions of the unmanned vessel in the geodetic coordinate system are respectively Axis coordinates Axial coordinates and heading angle; Let be the velocity vector of the unmanned surface vessel, and ,in, The forward speed of the unmanned vessel. The drift speed of the unmanned boat. The bow roll rate of the unmanned vessel; The unmanned ship in three-dimensional coordinate space Axis coordinates Axis coordinates Axis coordinates, roll angle, pitch angle, and yaw angle; Let be the velocity vector of the drone, and ,in, For drones along The displacement velocity of the shaft, For drones along The displacement velocity of the shaft, For drones along The displacement velocity of the shaft, The pitch speed represents the attitude of the drone. The roll speed is the attitude of the drone. The yaw speed represents the attitude of the drone. For the unknown nonlinear damping term of the unmanned vessel; For the corresponding known actuator performance coefficients; The engine rotation speed of the unmanned vessel; The rudder angle for the unmanned vessel; These represent the environmental disturbance forces experienced by the unmanned vessel during forward motion, the environmental disturbance forces experienced by the unmanned vessel during lateral drift motion, and the environmental disturbance forces experienced by the unmanned vessel during bow turning motion. For the nonlinear damping term of the UAV; For the quality of the drone; In three-dimensional space axis, axis, Rotation matrix corresponding to the axial direction; It is the acceleration due to gravity; The total lift generated by the drone's rotor; These are the environmental interference forces experienced by the UAV during forward motion, during drift motion, during yaw motion, during pitch motion, during roll motion, and during yaw motion, respectively. Let be the diagonal moment of the UAV rotor; The rotational torque generated by the difference in rotor speed of the UAV; respectively drones orbit The moment of inertia of the axis, the drone's orbit The moment of inertia of the axis, the drone's orbit Moment of inertia of the shaft; The added mass of the unmanned vessel during its forward motion; The added mass during the drifting motion of the unmanned vessel; The added mass during the bow-turning motion of the unmanned vessel; This represents the moment of inertia of the drone's propeller blades about their own axis of rotation. This is the rotational resistance coefficient; Let be the rotor angular velocity, and ; This represents the first-order hydrodynamic parameter in the forward velocity degree of freedom of the unmanned vessel. The second-order hydrodynamic parameters represent the forward velocity degree of freedom of the unmanned vessel. The third-order hydrodynamic parameters represent the forward velocity degree of freedom of the unmanned vessel. The first-order hydrodynamic parameter represents the degree of freedom of the unmanned vessel's drift velocity. The second-order hydrodynamic parameters represent the degree of freedom of the unmanned vessel's drift velocity. The third-order hydrodynamic parameters represent the degree of freedom of the unmanned vessel's drift velocity. The first-order hydrodynamic parameter represents the degree of freedom of the unmanned vessel's bow turning speed; The second-order hydrodynamic parameters represent the degree of freedom of the unmanned vessel's bow turning speed; The third-order hydrodynamic parameters represent the degree of freedom of the unmanned vessel's bow turning speed.
3. The unmanned surface vessel target tracking and control method based on sea-air cooperative visual perception as described in claim 1, characterized in that, The execution steps of the dynamic positioning framework that fuses stereo vision with platform pose include: S21. Based on the left image obtained from the left camera and the right image obtained from the right camera, a target recognition algorithm is used to identify the target and obtain the target pixel coordinates in the left image and the target pixel coordinates in the right image. S22. Based on the target pixel coordinates in the left image and the target pixel coordinates in the right image, the precise disparity of the target center point in the left and right images is calculated using the SGBM algorithm. S23. Using the left image as the main image, based on the target pixel coordinates in the left image and the precise parallax of the target center points in the left and right images, obtain the camera coordinate system coordinates of the target through coordinate transformation. S24. Based on the target's camera coordinate system coordinates, combined with the UAV's own attitude and position information, the target's geodetic coordinate system coordinates are obtained through coordinate transformation equations, i.e., the fused target's position coordinates.
4. The unmanned surface vessel target tracking and control method based on sea-air cooperative visual perception as described in claim 3, characterized in that, The execution steps of the multiple optimization and path generation mechanism include: S31. Simplify the position coordinates of the target to be fused into two-dimensional planar coordinates; S32. Based on the two-dimensional plane coordinates, noise reduction and state estimation are performed using an extended Kalman filter to obtain the preliminary accurate positioning coordinates of the target; and the heading angle of the target is calculated based on the preliminary accurate positioning coordinates of the target; the preliminary accurate positioning coordinates of the target and the heading angle of the target are combined to obtain the preliminary optimized target position; S33. Based on the preliminary optimized target position, the smoothed target position is obtained through a moving average filter, expressed as: （4） In the formula, For the first The target position after optimization by the moving average filter at any given time, i.e., the smoothed target position; This represents the window width of the moving average filter; This is the current sampling time; The target location is for preliminary optimization; S34. Based on the smoothed target position, continuous reference tracking trajectory points are obtained through a cubic spline interpolation function, expressed as: （5） In the formula, For reference tracking trajectory points; These are parameters to be determined. System uptime; For the first Time to the The interval of system runtime at any given moment; For the first System runtime at any given moment; For the first System runtime at any given moment; S35. Combine continuous reference tracking trajectory points to obtain a reference tracking trajectory; map the reference tracking trajectory using a virtual mapping method to obtain the reference tracking path of the UAV-unmanned vessel cooperative system.
5. The unmanned surface vessel target tracking and control method based on sea-air cooperative visual perception as described in claim 2 or 4, characterized in that, The specific steps to obtain the control law of the unmanned ship are as follows: S411. Based on the UAV-UAV cooperative system's reference tracking path's UAV position information and the UAV's position in the geodetic coordinate system, define the UAV's kinematic error as follows: （6） In the formula, For unmanned ships Shaft error; For the location information of the unmanned vessel Axis coordinates; For unmanned ships Shaft error; For the location information of the unmanned vessel Axis coordinates; The Euclidean distance is the positional error of the unmanned vessel. This refers to the heading angle error of the unmanned vessel; The bow angle is the position information of the unmanned vessel. S412. Differentiate the Euclidean distance of the unmanned vessel's position error with the bow angle error of the unmanned vessel to obtain the following expression; （7） In the formula, for The derivative; As an auxiliary variable, and ; for The derivative; for The derivative; S413. Based on equations (6) and (7), a virtual control law for the forward motion and turning motion of the unmanned vessel is designed, and a first-order filter is used to process the virtual control law for the forward motion and turning motion of the unmanned vessel to obtain the output of the first-order filter; the expression of the virtual control law is: （8） In the formula, A virtual control law for forward motion; For the virtual control law of bow-rolling motion; and To design control parameters; The expression for processing the virtual control law of forward motion and turning motion using a first-order filter is as follows: （9） In the formula, This is the filtering time; This is the filtering error; The expected velocity reference of the unmanned vessel is the output of the first-order filter; for The derivative; It is an index variable, and , among which, when At time, the forward degree of freedom of the unmanned vessel is... At that time, the bow roll angle is the degree of freedom of the unmanned vessel; It is a virtual control law, and ; S414. Based on the output of the first-order filter and the velocities of the unmanned vessel in each direction, define the dynamic error of the unmanned vessel, and differentiate the dynamic error to obtain the following expression: （10） In the formula, For the dynamic error of the unmanned vessel, and ,in, The speed of the unmanned vessel in each direction; for The derivative; For the direct control law of the unmanned vessel, and when hour, ,when hour, ; for The derivative; The added mass of the unmanned vessel during its motion; The environmental disturbance forces experienced by the unmanned vessel during its movement; For the unknown nonlinear damping term of the unmanned vessel; S415. By approximating equation (10) using a fuzzy logic system, the following expression is obtained: （11） In the formula, The highest membership degree of the fuzzy logic system serving the unmanned vessel; The fuzzy basis function vector serving the fuzzy logic system of the unmanned vessel; To serve the fitting error of the fuzzy logic system for unmanned ships, and ,in, Unknown constants for the fuzzy logic system serving unmanned ships; Design parameters for the fuzzy logic system serving unmanned ships; The upper bound of the fitting error for the fuzzy logic system serving unmanned ships; S416. Introduce the following disturbance observer into the dynamics controller of the unmanned vessel: （12） In the formula, This is to provide the interference observer's observation of the environmental disturbance forces experienced by the unmanned vessel during its movement; for The derivative; As an auxiliary variable; for The derivative; Design variables; S417. Combining equations (10), (11), and (12), the final control law of the unmanned vessel is obtained through the backstepping framework, and its expression is: （13） In the formula, and For design parameters; These are the adaptive parameters for the unmanned surface vessel; The derivative of the adaptive parameters of the unmanned vessel is the adaptive law of the final control law of the unmanned vessel.
6. The unmanned surface vessel target tracking and control method based on sea-air cooperative visual perception as described in claim 2 or 4, characterized in that, The specific steps to obtain the control law and attitude reference signal of the UAV are as follows: S421. Based on the UAV's position information in the reference tracking path of the UAV-Unmanned Vessel Cooperative System and the UAV's position in the geodetic coordinate system, the kinematic error of the UAV is defined as follows: （14） In the formula, For the kinematic errors of the drone; This represents the position of the UAV in the geodetic coordinate system. The location information of the UAV in the reference tracking path of the UAV-unmanned vessel cooperative system; Let be the index variable of the UAV's spatial coordinate axes, and ,when At that time, for drones Degrees of freedom in the axial direction, when At that time, for drones Degrees of freedom in the axial direction, when At that time, for drones Degrees of freedom in the axial direction; S422. Differentiating equation (14), we obtain the following expression: （15） In the formula, for The derivative; The displacement velocity of the UAV along the spatial coordinate axes; for The derivative; S423. Based on equations (14) and (15), a virtual control law for the UAV's motion along the spatial coordinate axis is designed, and a first-order filter is used to process the virtual control law for the UAV's motion along the spatial coordinate axis to obtain the output of the first-order filter; the expression of the virtual control law is: （16） In the formula, The virtual control law for the motion of the UAV along the spatial coordinate axes; , , and For design parameters, and ; The expression for processing the virtual control law of the UAV's motion along the spatial coordinate axes using a first-order filter is as follows: （17） In the formula, This is the filtering time; This is the filtering error; The expected speed reference for the UAV is the output of the first-order filter; for The derivative; S424. Based on the output of the first-order filter and the displacement velocity of the UAV along the spatial coordinate axes, define the dynamic error of the UAV, and differentiate the dynamic error of the UAV to obtain the following expression: （18） In the formula, This represents the dynamic error of the drone, and ,in, The displacement velocity of the UAV along the spatial coordinate axes; for The derivative; For the unknown nonlinear damping term of the UAV; Rotation matrices corresponding to each axis direction in three-dimensional space; The environmental disturbance forces experienced by the unmanned vessel during its movement; S425. By approximating equation (18) using a fuzzy logic system, the following expression is obtained: （19） In the formula, The highest membership degree of the fuzzy logic system serving the drone; Fuzzy basis function vectors for fuzzy logic systems serving unmanned aerial vehicles; To address the fitting error of the fuzzy logic system serving the unmanned aerial vehicle (UAV), and ,in, Unknown constants for fuzzy logic systems serving unmanned aerial vehicles; Design parameters for fuzzy logic systems serving unmanned aerial vehicles; The upper bound of the fitting error for the fuzzy logic system serving unmanned aerial vehicles; S426. Introduce the following disturbance observer into the dynamics controller of the UAV: （20） In the formula, This refers to the observation value of the environmental interference force experienced by the UAV during its movement by the interference observer; for The derivative; As an auxiliary variable; for The derivative; , , Design variables; S427, combining equations (18), (19), and (20), the final control law of the UAV is obtained through the inverse stepping framework, and its expression is: （21） In the formula, This is the final control law for the drone; The rate of change of the expected speed of the drone after smoothing; For design parameters; These are the adaptive parameters for the drone; The derivative of the adaptive parameters of the UAV is the adaptive law of the final control law of the UAV. S428. Based on the final control law of the UAV, the attitude reference signal of the UAV is obtained through a nonlinear decoupling method, and its expression is: （22） In the formula, The desired pitch angle for the drone; The desired roll angle for the drone; The desired yaw angle for the drone; for The final control law in the axial direction; for The final control law in the axial direction; for The final control law in the axial direction.