Dynamic cooperative path planning and autonomous landing control method for unmanned aerial vehicle and unmanned ship

Abstract

The invention relates to a dynamic cooperative path planning and autonomous landing control method for an unmanned aerial vehicle and an unmanned ship, and the method comprises the following steps: collecting the state data of the unmanned ship and the unmanned aerial vehicle and the high sea condition environment data, and carrying out the data fusion processing; an improved fast extended random tree algorithm is adopted to plan a safe navigation path for the unmanned ship, the generated path of the unmanned ship is optimized based on an unmanned ship motion model considering environmental interference, and a model prediction control algorithm is adopted to plan a path for the unmanned ship to fly to a predicted landing point of the unmanned ship. Dynamic cooperation between the unmanned ship and the unmanned aerial vehicle is realized through information interaction between the unmanned ship and the unmanned aerial vehicle; the unmanned aerial vehicle recognizes an unmanned ship landing identifier through a visual servo guide system, calculates the position and attitude deviation relative to a landing point, adjusts flight control parameters according to interference conditions by using an adaptive control algorithm, and performs trajectory prediction compensation in combination with an unmanned ship motion model to realize autonomous landing. Compared with the prior art, safe landing of the unmanned aerial vehicle on the unmanned ship under the high sea condition is achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of cooperative control of unmanned aerial vehicles, and in particular to a method for dynamic cooperative path planning and autonomous landing control of a UAV and an unmanned boat. Background Art

[0002] With the continued development and utilization of the ocean, the application of unmanned boats (UDVs) and drones (UAVs) in marine operations is becoming increasingly widespread. UAVs, with their autonomous navigation, long endurance, and strong environmental adaptability, are capable of performing a variety of tasks in complex marine environments. UAVs, with their flexibility, diverse perspectives, and rapid response, enable efficient detection and data collection of target areas. Working together, these two technologies can significantly enhance the efficiency and scope of marine operations.

[0003] CN118484015A discloses a method for coordinated landing of a UAV and a boat in a dynamic environment, comprising the following steps: guiding the landing from an initial position to a first preset height range based on GPS positioning; performing a phased assessment of the hull fluctuation state using the attitude prediction of the unmanned boat, and guiding the UAV to descend to a second preset height range; in the pre-landing stage, the UAV and the unmanned boat determine a suitable dynamic landing area through collaborative path planning and water condition assessment; after the dynamic landing area is determined, the UAV dynamically lands with the boat-borne self-stabilizing platform as the target, and the unmanned boat locates it according to the received UAV posture signal and performs position following; when the unmanned boat stably maintains the position following state, the boat-borne self-stabilizing platform continuously adjusts the platform attitude to keep the attitude stable until the UAV lands at the center of the self-stabilizing platform, thereby achieving relatively accurate UAV-boat coordinated landing in a dynamic environment.

[0004] However, in high sea conditions (Sea State Level 5 and above), the ocean environment is extremely complex and dynamic. The combined effects of high waves, high winds, and complex ocean currents cause unmanned vehicles to experience severe disturbances during navigation, severely impacting their attitude stability and making them prone to path deviation. Furthermore, drones taking off from other unmanned vehicles or from shore in these conditions face numerous challenges when landing on unmanned vehicles.

[0005] From the perspective of the drone, the violent swaying caused by high sea conditions makes the drone's deck an unstable landing platform. The drone's six degrees of freedom (including roll, pitch, bow pitch, heave, sway, and surge) vary rapidly over time. Traditional landing guidance and positioning technologies struggle to adapt to these dynamic changes and cannot accurately provide a stable landing reference point for the drone. For example, in high wave conditions, the drone's heave amplitude can reach several meters, and the roll and pitch angles can increase significantly, posing a significant threat to the safety and accuracy of the drone's landing.

[0006] For drones, strong winds in high sea conditions can severely impact their flight attitude and trajectory control. Unstable wind speeds and varying wind directions make it difficult for drones to maintain a stable flight while approaching unmanned boats, making precise landings more challenging. Furthermore, complex weather conditions in high sea conditions, such as low visibility and heavy rainfall, can affect the drone's vision and sensor performance, making it impossible to accurately identify the unmanned boat's landing area and related guidance signs.

[0007] Existing collaborative path planning methods for drones and unmanned boats are mostly designed for ideal or relatively stable environments. However, they are unable to achieve real-time, efficient collaboration between drones and unmanned boats in the extremely dynamic environment of high seas. These methods struggle to quickly and accurately plan a safe and optimal landing path for the drone based on the unmanned boat's real-time motion and surrounding environmental changes. Furthermore, existing technologies for autonomous landing control lack comprehensive consideration of the dynamic characteristics of the unmanned boat and drone in high seas, making it impossible to reliably land the drone on unstable decks.

[0008] In summary, realizing dynamic collaborative path planning and autonomous landing control between drones and unmanned boats under high sea conditions is a key technical problem that needs to be urgently solved in the field of marine unmanned systems. Summary of the Invention

[0009] The purpose of this invention is to provide a method for dynamic collaborative path planning and autonomous landing control between a drone and an unmanned boat, addressing the difficulty of safely and accurately landing a drone on an unmanned boat in high sea conditions. This method utilizes real-time status information from the unmanned boat and drone, combined with high sea condition data, to perform dynamic collaborative path planning. Furthermore, it employs an innovative autonomous landing control strategy to achieve reliable landing of a drone on an unstable unmanned boat deck.

[0010] The purpose of the present invention can be achieved by the following technical solutions:

[0011] A method for dynamic collaborative path planning and autonomous landing control of a UAV and an unmanned boat comprises the following steps:

[0012] Data collection and processing steps: The unmanned boat and drone collect their own status data through their respective multi-sensor systems. At the same time, the environmental monitoring system obtains high sea state environmental data and integrates the collected data;

[0013] Dynamic collaborative path planning steps: Based on the fused data, an improved rapidly expanding random tree algorithm is used to plan a safe navigation path for the unmanned boat. The generated unmanned boat path is optimized based on the unmanned boat motion model that considers environmental interference. A model predictive control algorithm is used to plan a path for the drone to fly to the predicted landing point of the unmanned boat. Dynamic collaboration between the unmanned boat and the drone is achieved through information exchange between the two.

[0014] Autonomous landing control steps: The UAV uses the visual servo guidance system to identify the unmanned boat landing mark, calculates the position and attitude deviation relative to the landing point, uses the adaptive control algorithm to adjust the flight control parameters according to the interference situation, and combines the unmanned boat motion model to perform trajectory prediction compensation to achieve autonomous landing.

[0015] The multi-sensor system carried by the unmanned boat includes an inertial measurement unit, a global positioning system, a lidar and a visual sensor. The collected data include the unmanned boat's attitude information in three-dimensional space, position coordinates, velocity vector, surrounding point cloud data and images of the unmanned boat's deck and surrounding environment;

[0016] The multi-sensor system carried by the UAV includes an inertial measurement unit, a global positioning system and a visual sensor, and the collected data includes the UAV's own attitude angle, angular velocity, acceleration, position coordinates, velocity vector and environmental image;

[0017] The environmental monitoring system includes a meteorological satellite and a buoy monitoring system, and the collected data include wave spectrum, wind speed, wind direction and ocean current speed and direction in the local sea area.

[0018] The fusion processing of the collected data includes:

[0019] Self-state data fusion: The extended Kalman filter algorithm is used to fuse the inertial measurement unit and global positioning system data of the unmanned boat and drone. After pre-processing the lidar point cloud data, a feature matching-based algorithm is used to fuse it with the image data collected by the visual sensor to extract environmental feature information.

[0020] Environmental data fusion: The collected environmental data are standardized and fused using a weighted fusion algorithm. The weights of different data in the weighted fusion algorithm are determined based on the degree of influence of each environmental data on the movement of unmanned boats and drones.

[0021] The improved rapid expansion random tree algorithm adds the unmanned boat posture change constraint condition when planning the path for the unmanned boat, and combines the obstacle information obtained by the laser radar to construct the path, wherein,

[0022] Taking the unmanned boat as the research object, the state space is defined as: Where (x, y, z) is the three-dimensional position of the UAV in the geographic coordinate system, φ, θ, and ψ are the roll angle, pitch angle, and bow angle, respectively. At the same time, the control space is defined as: Among them, u1 controls the forward speed of the unmanned boat, and u2 controls the steering angle;

[0023] Improved rapid expansion random tree algorithm based on the current position of the unmanned boat is the starting point and the target location of the mission Perform random tree expansion for the endpoint, and in each expansion, from the state space Randomly sample a state x rand , calculate the tree node x by the following formula near to x rand Direction vector

[0024]

[0025] Among them, ||·|| represents the modulus of the vector;

[0026] Along direction vector Generate a new node x with a fixed step size Δs new :

[0027] After generating a new node, determine whether the new node meets the attitude constraint of the unmanned boat and whether it is in the obstacle area based on the point cloud data obtained by the laser radar. If the attitude constraint is met and there is no collision with an obstacle, the new node is added to the random tree; otherwise, the node is resampled and generated, where the attitude constraint includes the roll angle constraint |φ|≤φ max , pitch angle constraint |θ|≤θ max ,φ max and θ max is the maximum allowable angle set according to the structure of the unmanned boat and the sea state level; repeat the above process until a feasible path from the starting point to the end point is found;

[0028] Considering the interference of waves, wind speed and ocean currents on the movement of the unmanned boat, the additional forces and moments of the unmanned boat under the corresponding interference are calculated through the acquired environmental data. The unmanned boat motion model is established considering the additional forces and moments, and the generated feasible path is optimized.

[0029] When planning a path for a UAV, the model predictive control algorithm uses the real-time position and predicted landing point of the UAV as targets, combines the UAV's own state data and high sea state environmental data, and uses the UAV motion model that takes environmental interference into account to predict the motion state within a preset time in the future. It then constructs an objective function and solves the optimal control sequence to generate a flight path. The objective function J aims to minimize the time and energy consumption of the UAV to reach the predicted landing point:

[0030]

[0031] Among them, λ1 and λ2 are weight coefficients used to balance position error and energy consumption, N is the number of predicted future time steps, is the drone state vector, To predict the landing point, is the control input vector of the UAV, and k is the current moment.

[0032] The dynamic coordination between the unmanned boat and the drone is achieved through information interaction between the two as follows:

[0033] The unmanned boat sends its real-time status data, predicted trajectory, and environmental data to the drone, and the drone feeds back its status data and current path planning results to the unmanned boat.

[0034] When the motion state of the unmanned boat changes due to high sea conditions and exceeds a preset threshold, the unmanned boat immediately sends the new state data and predicted trajectory to the drone; after receiving the information, the drone re-runs the model predictive control algorithm, takes the new predicted landing point of the unmanned boat as the target, and optimizes and adjusts the flight path based on the updated environmental data and its own state; at the same time, the drone feeds the adjusted path back to the unmanned boat, and the unmanned boat fine-tunes its own navigation path based on the feedback information.

[0035] The UAV identifies the landing mark of the unmanned boat through the visual servo guidance system:

[0036] A specific landing mark is set on the deck of the unmanned boat. The drone collects images through visual sensors, processes the collected images using target detection algorithms, identifies the landing mark on the deck of the unmanned boat, and extracts the corner features on the mark using corner detection algorithms.

[0037] The method for calculating the position and attitude deviation relative to the landing point is specifically as follows:

[0038] According to the pinhole camera model, the transformation relationship between the image coordinate system and the UAV body coordinate system and the UAV coordinate system is established:

[0039] Assume that the intrinsic parameter matrix of the drone camera is where f x ,f y is the focal length, c x ,c y is the main point coordinate;

[0040] The three-dimensional coordinates P of the landing mark in the unmanned boat coordinate system are known i =(X i ,Y i ,Zi ) T , calculate its projection coordinates in the image coordinate system through the perspective projection transformation formula, the perspective projection transformation formula is:

[0041]

[0042] Among them, s is the scale factor, R is the rotation matrix, t is the translation vector, and u i 、v i They are corner feature p i The horizontal and vertical coordinates in the image coordinate system;

[0043] The above equations are solved by the least squares method to obtain the estimated values ​​of the rotation matrix and translation vector. The estimated rotation matrix is ​​decomposed to obtain the attitude deviation of the UAV relative to the UAV. The position deviation of the UAV relative to the UAV is determined based on the components of the estimated translation vector in the directions of the three coordinate axes.

[0044] The adaptive control algorithm comprises the following steps:

[0045] The environmental disturbance factors are equivalently modeled as disturbance forces and disturbance torques acting on the UAV, which serve as disturbance inputs in the adaptive control algorithm.

[0046] Adaptive backstepping control method is used to design the UAV control law:

[0047] The dynamic equation of the attitude error of the UAV is defined as: Among them, e θ is the attitude error vector, θ is the actual attitude angle vector of the UAV, ω d is the desired attitude angular velocity vector, R(θ) is the attitude rotation matrix;

[0048] Step-by-step construction of the Lyapunov function V by backstepping design i , and designed an adaptive control parameter update law based on the Lyapunov stability theory. Among them, the adaptive adjustment term in the adaptive control parameter update law is dynamically adjusted through real-time estimated interference input to ensure that the UAV attitude quickly tracks the changes of the unmanned boat deck.

[0049] In the prediction and compensation step, based on the unmanned boat motion model taking into account environmental interference, the extended Kalman filter algorithm or the particle filter algorithm is used to predict the motion state of the unmanned boat in the future. According to the prediction results, the trajectory compensation vector and attitude compensation vector of the UAV are calculated and incorporated into the UAV control instructions to adjust the UAV flight trajectory.

[0050] Compared with the prior art, the present invention has the following beneficial effects:

[0051] (1) The accuracy and reliability of data collection and processing are greatly enhanced

[0052] In the data acquisition and processing phase, this invention abandons the traditional single sensor or simple data processing mode and adopts an organic combination of high-precision, multi-type sensors. At the same time, it builds an innovative multi-source heterogeneous data fusion architecture, uses algorithms such as the extended Kalman filter, fuses data with different sampling frequencies and different physical properties, establishes a unified data spatiotemporal benchmark, and uses an adaptive weighted fusion algorithm to flexibly adjust the weights of each data source according to data reliability and environmental changes. This technical solution effectively overcomes the shortcomings of traditional methods in that data is easily interfered with and lacks accuracy in high sea conditions. It can accurately capture the status information of unmanned boats and drones, deeply integrate environmental data such as waves, wind speed, and ocean currents, and provide a solid data foundation for subsequent path planning and landing control, significantly improving the accuracy and reliability of the data.

[0053] (2) Dynamic collaborative path planning is more efficient and safer

[0054] In the dynamic collaborative path planning algorithm, the present invention makes targeted improvements and optimizations to the rapidly expanding random tree (RRT) algorithm and the model predictive control (MPC) algorithm. In the unmanned boat path planning process, the characteristics of the drastic changes in the attitude of the unmanned boat under high sea conditions are fully considered, strict attitude change constraints are added, and an unmanned boat motion model that considers environmental interference is introduced to make the planned path more in line with actual navigation needs; when planning the path of the unmanned aerial vehicle, the real-time position of the unmanned boat, the predicted landing point and the complex and changeable environmental data are closely combined to dynamically adjust the flight path. At the same time, through a real-time and efficient information interaction mechanism, close coordination between the unmanned aerial vehicle and the unmanned boat is achieved. Compared with the traditional path planning method, this method effectively solves the problem that path planning under high sea conditions is prone to deviation and cannot adapt to environmental changes. It can quickly plan a safe and reasonable path, greatly reducing the risk of collision caused by improper path planning, and significantly improving the success rate of task execution.

[0055] (3) The accuracy and stability of autonomous landing of drones are significantly improved

[0056] The autonomous landing control strategy of the present invention successfully overcomes the key technical difficulties of UAV landing under high sea conditions through the coordinated cooperation of a landing guidance system based on visual servoing, an adaptive control algorithm, and a predictive compensation mechanism. The dual-modal visual sensor and advanced image recognition algorithm give the UAV the ability to accurately identify the landing point of the unmanned boat under adverse weather conditions such as low visibility and heavy precipitation; the adaptive control algorithm can quickly and accurately adjust the UAV control parameters according to the real-time interference level to ensure that the UAV closely tracks the dynamic changes of the unmanned boat deck; the predictive compensation mechanism predicts the movement trend of the unmanned boat in advance and makes forward-looking adjustments to the UAV flight trajectory. Compared with traditional landing control methods, the present invention improves the landing performance of UAVs under high sea conditions, effectively guarantees the accuracy and stability of UAV landing, and greatly reduces the probability of landing failure.

[0057] (4) System environmental adaptability and versatility are significantly enhanced

[0058] The present invention takes into account the complex characteristics of high sea conditions, and fully considers the influence of interference factors such as waves, strong winds, and ocean currents throughout the entire process from data collection, path planning to landing control. Whether facing adverse weather conditions such as low visibility and heavy rainfall, or in extreme marine environments with sea conditions of level 5 or above, the present invention can operate continuously and stably and has strong environmental adaptability. In addition, the present invention has good compatibility with various types of drones and unmanned boats, breaking away from the dependence of traditional technologies on specific equipment, significantly expanding the application scenarios of unmanned systems in marine operations, and can be widely used in marine environmental monitoring, maritime emergency rescue, material delivery and other fields, greatly improving the versatility and application value of the technology. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] Figure 1 Flow chart of the method of the present invention. DETAILED DESCRIPTION

[0060] The present invention is described in detail below with reference to the accompanying drawings and specific embodiments. This embodiment is implemented based on the technical solution of the present invention, and provides a detailed implementation method and specific operation process, but the protection scope of the present invention is not limited to the following embodiments.

[0061] This embodiment provides a method for dynamic collaborative path planning and autonomous landing control of a UAV and an unmanned boat, which is particularly suitable for scenarios requiring collaborative operations between UAVs and unmanned boats, such as marine environment monitoring, maritime emergency rescue, and material delivery. Figure 1 As shown, the method includes the following steps:

[0062] Step 1) Data collection and processing: The unmanned boat and the drone collect their own status data through their respective multi-sensor systems, and use the environmental monitoring system to obtain high sea state environmental data, and then fuse the collected data.

[0063] Step 11) Data Collection

[0064] (1) Unmanned boat sensor configuration

[0065] The multi-sensor system carried by the unmanned boat includes an inertial measurement unit, a global positioning system, a lidar and a visual sensor. The collected data include the unmanned boat's attitude information, position coordinates, velocity vector, surrounding environment point cloud data and images of the unmanned boat's deck and surrounding environment in three-dimensional space.

[0066] A high-precision fiber optic gyro inertial measurement unit (IMU) is used, with a zero-bias stability of 0.01° / h. It can collect the attitude information of the unmanned boat in three-dimensional space in real time at a frequency of 100Hz under severe shaking in high sea conditions, including roll angle φ, pitch angle θ, bow angle ψ and their corresponding angular velocity. and acceleration a x 、a y 、a z In this embodiment, a high-precision fiber optic gyro inertial measurement unit (IMU) is installed on the deck and key positions of the unmanned boat and fixed with a shock-absorbing device to reduce the impact of hull vibration on measurement accuracy.

[0067] Using dual-frequency multi-constellation global positioning system (GPS), the positioning accuracy can reach centimeter level, and the geographic coordinates (x, y, z) and velocity vector of the unmanned boat are obtained at a frequency of 10Hz. The dual-frequency multi-constellation GPS antenna is installed in an open position on the top of the unmanned boat to ensure unobstructed signal reception.

[0068] A mechanical rotating laser radar is deployed at the front of the unmanned boat, with a scanning range of 360° and an angular resolution of 0.1°. It scans the surrounding environment at a frequency of 20Hz to obtain point cloud data. where p i Represents the three-dimensional coordinates of the i-th point, N is the number of point clouds, and is used to construct the obstacle map.

[0069] A high dynamic range (HDR) visual sensor with a frame rate of 30Hz is installed above the center of the unmanned boat deck. It captures images of the unmanned boat deck and surrounding environment through a wide-angle lens for identifying drones and landing area signs.

[0070] (2) UAV sensor configuration

[0071] The multi-sensor system carried by the drone includes an inertial measurement unit, a global positioning system and a visual sensor. The collected data include the drone's own attitude angle, angular velocity, acceleration, position coordinates, velocity vector and environmental image.

[0072] The micro-electromechanical system (MEMS) inertial measurement unit is used to collect its own attitude angle φ at a frequency of 200Hz while meeting the lightweight requirements of the UAV. d ,θ d , ψ d , angular velocity and acceleration a dx 、a dy 、a dz .

[0073] Equipped with a miniaturized high-precision GPS module, installed on the top of the drone, optimized antenna design to enhance signal reception capability, positioning accuracy of sub-meter level, and obtain the position of the drone at a frequency of 10Hz (x d ,y d ,z d ) and speed

[0074] A visual sensor is installed under the nose of the drone, integrating infrared and visible light dual modes to adapt to complex meteorological conditions such as low visibility in high sea conditions. It collects image data at a frequency of 30Hz for visual identification of unmanned boats.

[0075] (3) Environmental monitoring system

[0076] The environmental monitoring system includes meteorological satellites and buoy monitoring systems, and the data collected include wave spectrum, wind speed, wind direction, and ocean current speed and direction in local sea areas.

[0077] High-resolution meteorological data covering the operating area is obtained through meteorological satellites, including the wave spectrum S(ω), where ω is the angular frequency and the wave spectrum describes the distribution of wave energy with frequency; wind speed v w 、wind direction θ w . Use the buoy monitoring system to collect the ocean current speed in the local sea area in real time. and direction θ c The buoy is equipped with an acoustic Doppler current profiler (ADCP) with a measurement accuracy of 1% of the measured value.

[0078] Collected data is transmitted via wireless communication modules. Both the UAV and the drone are equipped with high-bandwidth, low-latency wireless communication equipment, using dedicated frequency bands for data exchange, ensuring stable and real-time data transmission. The UAV sends its own and environmental data to the data processing center, and the drone transmits its own status data to the UAV and the data processing center.

[0079] Step 12) Data Fusion

[0080] (1) Fusion of the state data of the unmanned boat and the drone: The extended Kalman filter (EKF) algorithm is used to fuse the inertial measurement unit and global positioning system data of the unmanned boat and the drone respectively. After preprocessing the lidar point cloud data, the feature matching-based algorithm is used to fuse it with the image data collected by the visual sensor to extract environmental feature information.

[0081] Taking the unmanned boat as an example, the state vector The system state transition equation is: Where f(·) is the nonlinear state transfer function, is process noise, which obeys Gaussian distribution Q k is the process noise covariance matrix. The measurement equation is: where h(·) is a nonlinear measurement function, To measure noise, it follows a Gaussian distribution R k is the measurement noise covariance matrix. Through the prediction and update steps of the EKF algorithm, the state of the unmanned vehicle is accurately estimated.

[0082] The same principle is used to fuse drone status data. After pre-processing such as denoising and segmentation, the lidar point cloud data is fused with the visual sensor image data using a feature matching algorithm to extract environmental feature information.

[0083] (2) Environmental data fusion: The collected environmental data are standardized and fused using a weighted fusion algorithm. The weights of different data in the weighted fusion algorithm are determined based on the degree of influence of each environmental data on the movement of the unmanned boat and the drone.

[0084] The wave spectrum S(ω) obtained by meteorological satellites and the ocean current speed monitored by buoys and direction θ c , and wind speed v w 、wind direction θ w Environmental data, such as the pod and the drone, undergoes data standardization and is fused using a weighted fusion algorithm. Weights are determined based on the degree to which each environmental data item affects the movement of the unmanned boat and drone. For example, in high sea conditions, wave spectrum data has a greater impact on the movement of the unmanned boat and is therefore given a higher weight. This fused data is further correlated with the status data of the unmanned boat and drone, providing comprehensive data support for subsequent path planning and landing control.

[0085] The environmental data is then correlated with the status data of the unmanned boat and drone.

[0086] The wave spectrum S(ω) is processed and the heave displacement is calculated by linear wave theory Among them A i is the amplitude of the i-th frequency component, determined by the wave spectrum S(ω); i is the angular frequency; ε i is the random phase; n is the number of frequency components. s It will be used to evaluate the vertical interference effect of waves on unmanned boats and drones in the subsequent process. s The impact on the stability of the unmanned boat's navigation attitude is to adjust the path of the unmanned boat to avoid sailing in the area where the waves are too large; in the autonomous landing control stage of the UAV, combined with Z s Predicting the vertical movement of the unmanned boat deck under the action of waves provides a basis for compensating the landing trajectory of the drone, ensuring that the drone is adapted to the vertical height of the unmanned boat deck when landing.

[0087] For ocean current speed monitored by buoys and direction θ c , converting it into a force on unmanned boats and drones According to the principles of fluid mechanics, Where ρ is the seawater density, S is the platform's windward or upstream area, and C d is the drag coefficient.

[0088] Wind speed v w 、wind direction θ w Also converted into wind power Among them, ρ a is the air density, S w is the frontal area, C dw is the drag coefficient.

[0089] By establishing a unified coordinate system, various environmental interference forces and moments are synthesized in this coordinate system, and the environmental data is associated and integrated with the status data of unmanned boats and drones, and the impact of the environment on the movement of unmanned boats and drones is comprehensively evaluated.

[0090] Step 2) Dynamic collaborative path planning: Based on the fused data, an improved rapidly expanding random tree algorithm is used to plan a safe navigation path for the unmanned boat. The generated unmanned boat path is optimized based on the unmanned boat motion model that considers environmental interference. The model predictive control algorithm is used to plan the path for the drone to fly to the predicted landing point of the unmanned boat. Dynamic collaboration between the two is achieved through information exchange between the unmanned boat and the drone.

[0091] In a complex and dynamic environment with high sea conditions, the dynamic collaborative path planning algorithm needs to accurately respond to the motion characteristics of the unmanned boat and drone and environmental interference. Specifically, it includes the following steps:

[0092] Step 21) Path planning for the unmanned boat

[0093] In high sea conditions, unmanned vehicles (UAVs) are subject to complex and variable motions influenced by waves, currents, and strong winds, making traditional path planning algorithms inadequate. This paper employs an improved Rapidly Expanding Random Tree (RRT) algorithm to fully consider the UAV's attitude changes and environmental interference when constructing a path. This algorithm also adds constraints on the UAV's attitude changes and incorporates obstacle information acquired by LiDAR to ensure path feasibility and safety.

[0094] Taking the unmanned boat as the research object, the state space is defined as: Where (x, y, z) is the three-dimensional position of the UAV in the geographic coordinate system, φ, θ, and ψ are the roll angle, pitch angle, and bow angle, respectively. At the same time, the control space is defined as: Among them, u1 controls the forward speed of the unmanned boat, and u2 controls the steering angle.

[0095] Improved rapid expansion random tree algorithm based on the current position of the unmanned boat is the starting point and the target location of the mission Perform random tree expansion for the endpoint, and in each expansion, from the state space Randomly sample a state x rand , calculate the tree node x by the following formula near to x rand Direction vector

[0096]

[0097] Among them, ||·|| represents the modulus of the vector;

[0098] Along direction vector Generate a new node x with a fixed step size Δs new :

[0099] After generating a new node, determine whether the new node meets the attitude constraint of the unmanned boat and whether it is in the obstacle area based on the point cloud data obtained by the lidar. If the attitude constraint is met and there is no collision with an obstacle, the new node is added to the random tree; otherwise, the node is resampled and generated, where the attitude constraint includes the roll angle constraint |φ|≤φ max , pitch angle constraint |θ|≤θ max ,φ max and θ max is the maximum allowable angle set according to the structure of the unmanned boat and the sea condition level; the above process is repeated until a feasible path from the starting point to the end point is found.

[0100] Taking into account the interference of waves, wind speed and ocean currents on the movement of the unmanned boat, the additional force and torque of the unmanned boat under the corresponding interference are calculated through the acquired environmental data. The unmanned boat motion model is established considering the additional force and torque, and the generated feasible path is optimized.

[0101] Taking the unmanned boat as an example, a motion model that takes environmental interference into account is established:

[0102] Where M is the mass matrix of the unmanned boat, is the acceleration vector, are the Coriolis force and centripetal force matrices, is the damping matrix, is the restoring force vector generated by gravity and buoyancy, τ is the control force vector generated by the thruster, is the environmental interference force vector.

[0103] Including wave disturbance force Ocean current interference force wind power Etc. Wave disturbance force It is determined based on the wave spectrum S(ω) and the interaction between the unmanned boat and the waves, for example, by calculating the Morrison equation. wind power At the same time, the torque generated by environmental interference is considered, such as the rolling moment M caused by waves φs , pitch moment M θs et al., modeled by ship motion theory.

[0104] In one embodiment, taking the sea wave interference as an example, the additional force of the unmanned boat under the action of the sea wave is calculated according to the sea wave spectrum S(ω) and additional torque

[0105]

[0106] Among them F si and M si is the force and moment amplitude of the ith frequency component, determined by the wave spectrum; and The additional forces and moments are incorporated into the UAV dynamics equations to optimize the generated path, allowing the UAV to navigate more smoothly along the planned path in high sea conditions.

[0107] The moment generated by the waves on the unmanned boat includes the rolling moment M φs , pitch moment M θs and yaw moment M ψs Based on linear wave theory and ship motion equations, the rolling moment M φs It can be expressed as Where ρ is the density of seawater, g is the acceleration of gravity, V is the displacement volume of the unmanned boat, is the initial stability, φ is the roll angle, A i 、ω i , ε i are the amplitude, circular frequency and phase of the wave respectively. θs The model is similarly performed using relevant ship motion parameters and wave parameters.

[0108] For ocean current disturbance, the ocean current speed monitored by buoys and direction θ c , calculate the force of ocean currents on unmanned boats and drones In the path planning of the unmanned boat, when the ocean current speed exceeds a certain threshold and the angle between the direction and the planned path of the unmanned boat is large, the The size and direction of the current are taken into account, and an avoidance strategy for areas affected by ocean currents is added during the path search process. For example, when the Rapidly Expanding Random Tree (RRT) algorithm expands a node, if the new node is in an area affected by strong ocean currents, the node's expansion priority is lowered or the expansion direction is reselected. At the same time, the additional power required to overcome the current force is calculated based on the power performance of the unmanned vehicle, and the energy consumption and safety of navigation in this area are evaluated to optimize the path.

[0109] For wind disturbance, the wind speed v w 、wind direction θ w Calculating wind force When planning the path of the unmanned boat, if it encounters strong winds, the wind resistance and navigation attitude restrictions of the unmanned boat are combined. According to the direction of action of the wind, the path is locally adjusted, such as increasing the curvature of the windward sailing path, reducing the impact of the crosswind on the rolling force of the unmanned boat, and ensuring the navigation safety of the unmanned boat under wind interference.

[0110] Step 22) Drone path planning

[0111] In high sea conditions, drones must precisely avoid environmental interference while quickly and safely approaching the unmanned boat's landing point. This paper employs a model predictive control (MPC) algorithm, using the unmanned boat's real-time position and predicted landing point as targets. This algorithm combines the drone's own state information with high sea condition data. Based on a drone motion model that considers environmental interference, the algorithm predicts its motion state within a preset timeframe. This algorithm then constructs an objective function and solves for the optimal control sequence to generate a flight path.

[0112] Establish a six-degree-of-freedom motion model of the UAV without considering environmental interference, and its state vector Where (x d ,y d ,z d ) is the position of the drone, (vdx ,v dy ,v dz ) is the speed, (φ d ,θ d ,ψ d ) is the attitude angle, is the attitude angular velocity. The dynamic equation can be expressed as:

[0113] where f d (·) is the nonlinear dynamic function, The input vector for control includes motor speed and rudder angle; is process noise, which obeys Gaussian distribution Q d is the process noise covariance matrix.

[0114] Referring to the content of unmanned boat path planning, environmental interference is included in the six-degree-of-freedom motion model of the UAV, such as is the environmental interference force on the UAV, M ed is the environmental interference torque, and its expression is determined according to the aerodynamic characteristics of the UAV and environmental factors.

[0115] In one embodiment, the ocean current interference force It is converted into an interference term on the UAV's flight trajectory. By predicting the motion deviation of the UAV under the interference of ocean currents, the control input is adjusted to make the UAV path avoid the area of ​​strong ocean current interference or maintain flight stability when crossing.

[0116] Wind interference is incorporated into the UAV dynamics model. During the MPC algorithm prediction process, the impact of wind on the UAV speed and attitude is considered. By optimizing the control sequence, the flight speed and attitude of the UAV are adjusted in real time to offset wind interference and ensure that the UAV can accurately fly to the predicted landing point of the unmanned boat.

[0117] Based on the UAV motion model, predict the state of the next N time steps Where k is the current moment. At the same time, according to the unmanned boat motion model, the predicted landing point position in the next N time steps is obtained

[0118] Then, define the objective function J to minimize the time and energy consumption of the UAV to reach the predicted landing point:

[0119]

[0120] Among them, λ1 and λ2 are weight coefficients used to balance position error and energy consumption, N is the number of predicted future time steps, is the drone state vector, To predict the landing point, is the control input vector of the UAV, and k is the current moment.

[0121] Under the conditions of satisfying the UAV dynamic constraints (such as maximum speed, maximum acceleration, maximum attitude angle change rate) and safety constraints (such as obstacle avoidance distance), the optimal control sequence is obtained by solving the minimum value of the objective function J. Only the first control Applied to a drone, the above prediction and optimization process is repeated over time, the control variables are continuously updated, and the flight path of the drone is generated.

[0122] Step 23) Dynamic coordination between the UAV and the drone is achieved through information exchange between the two.

[0123] The unmanned boat sends its real-time status information (position (x, y, z), attitude (φ, θ, ψ), speed ), predict the trajectory and environmental information (waves, wind speed, ocean current data) to the drone, and the drone sends its own status information (position (x d ,y d ,z d ), posture (φ d ,θ d ,ψ d ),speed ) and the current path planning results are fed back to the unmanned boat;

[0124] When the motion state of the unmanned boat changes due to high sea conditions and exceeds the preset threshold, such as the roll angle exceeds the set threshold φ threshold Or the position deviation is greater than Δx max The unmanned boat immediately sends the new status information and predicted trajectory to the drone; after receiving the information, the drone suspends the current path execution and re-runs the model predictive control algorithm, taking the new predicted landing point of the unmanned boat as the target, and optimizes and adjusts the flight path in combination with the updated environmental data and its own status; at the same time, the drone feeds back the adjusted path information to the unmanned boat, and the unmanned boat fine-tunes its own navigation path based on the feedback information, ensuring that the two always maintain dynamic coordination in high sea conditions, creating conditions for the drone to land safely.

[0125] Step 3) Autonomous landing control step: The UAV uses the visual servo guidance system to identify the landing mark of the unmanned boat, calculates the position and attitude deviation relative to the landing point, uses the adaptive control algorithm to adjust the flight control parameters according to the interference situation, and combines the unmanned boat motion model to perform trajectory prediction compensation to achieve autonomous landing.

[0126] In high sea conditions, the deck of an unmanned vessel is subject to dramatic dynamic changes, posing a significant challenge to autonomous drone landing. This invention leverages these characteristics to achieve precise drone landing on unstable decks through visual servo guidance, adaptive control, and predictive compensation mechanisms.

[0127] Step 31) Landing Guidance System Based on Visual Servoing

[0128] In order to achieve accurate identification and positioning of the landing point of the unmanned boat by the drone, a landing guidance system based on visual servoing is constructed. The system fully considers factors such as low visibility and deck shaking in high sea conditions to ensure the reliability and accuracy of visual information.

[0129] Step 311) Visual sensor selection and image acquisition: The UAV is equipped with a dual-mode visual sensor (visible light and infrared). The visible light sensor has high resolution (e.g., 4K resolution) and wide dynamic range (120dB), which is suitable for environments with drastic changes in light in high sea conditions. The infrared sensor uses an uncooled focal plane array, which can operate in low visibility conditions such as nighttime or dense fog, effectively capturing the features of the UAV deck. The sensor acquires images at a frame rate of 30Hz, obtaining image sequence I. t , where t represents time.

[0130] Step 312) Landing mark recognition and feature extraction: Set a specific landing mark on the unmanned boat deck, such as a pattern consisting of concentric circles and characteristic corner points. Use a deep learning-based target detection algorithm (such as the improved YOLOv5 algorithm) to detect the collected image I. t Processing is performed to identify the landing mark. The corner features p on the mark are extracted using the Harris corner detection algorithm. i =(u i ,v i ) T , where (u i ,v i ) are the horizontal and vertical coordinates of the corner points in the image coordinate system, i = 1, 2, …, n, and n is the number of corner points.

[0131] Step 313) Calculate the position and attitude deviation relative to the landing point

[0132] According to the pinhole camera model, the transformation relationship between the image coordinate system and the UAV body coordinate system and the UAV coordinate system is established:

[0133] Assume that the intrinsic parameter matrix of the drone camera is where f x ,f y is the focal length, c x ,c y The main point coordinates.

[0134] The three-dimensional coordinates P of the landing mark in the unmanned boat coordinate system are known i =(X i ,Y i ,Z i ) T , calculate its projection coordinates in the image coordinate system through the perspective projection transformation formula, the perspective projection transformation formula is:

[0135]

[0136] Among them, s is the scale factor, R is the rotation matrix, t is the translation vector, and u i 、v i They are corner feature p i The horizontal and vertical coordinates in the image coordinate system.

[0137] The three-dimensional coordinates P of the landing mark in the unmanned boat coordinate system are known i =(X i ,Y i ,Z i ) T It is pre-set and known, and the camera internal parameter matrix K can be obtained through camera calibration. In practical applications, the image containing the landing mark is collected by the visual sensor, and the corner feature p on the mark is extracted using the corner detection algorithm. i , get its coordinates in the image coordinate system (u i ,v i ).

[0138] After substituting the known quantities into the formula, we can obtain a set of equations about the rotation matrix R and the translation vector t. Solving the above equations by the least squares method, we can obtain the estimated values ​​of the rotation matrix and translation vector. The rotation matrix R contains the attitude information of the UAV relative to the unmanned boat. By decomposing the rotation matrix, such as using the Euler angle decomposition method, we can obtain the attitude deviation of the UAV relative to the unmanned boat Δθ=(α err ,β err ,γ err ) T , where α err , β err , γ err are the attitude angle deviations around different coordinate axes. The translation vector t reflects the position information of the UAV relative to the unmanned boat, and its components in the three coordinate axis directions correspond to the position deviation Δx=(x err ,y err ,z err ) T , that is, the position deviation and attitude deviation of the UAV relative to the unmanned boat are obtained.

[0139] Step 32) Adaptive Control Algorithm

[0140] Step 321) Environmental interference modeling

[0141] The environmental disturbance factors are equivalently modeled as disturbance forces and disturbance torques acting on the UAV, which are used as disturbance inputs in the adaptive control algorithm to achieve effective compensation for environmental disturbances.

[0142] Taking the interference of ocean waves as an example, based on the linear wave theory, the interference force F of ocean waves on drones is d It can be expressed as: F d =ρg∫ S n·ξdS, where ρ is the seawater density, g is the acceleration of gravity, S is the projected area of ​​the UAV on the sea surface, n is the sea surface normal vector, and ξ is the wave displacement vector.

[0143] For drones, the airflow changes caused by waves will produce interference torque on them. Through aerodynamic theory and combined with the model of the impact of waves on the surrounding airflow field, the relationship between the wave interference torque and the drone's posture is established.

[0144] In addition to the disturbance of waves, as mentioned above, the disturbance force of ocean currents A model has been established, locating its point of action near the center of mass of the unmanned boat or drone, and calculating the ocean current disturbance torque based on the moment arm relationship. For wind disturbances, a wind disturbance torque model is established based on the positional relationship between the wind action point, the drone's center of mass, and various coordinate axes. This model comprehensively considers various disturbance forces and torques, providing accurate disturbance input for the subsequent adaptive control algorithm.

[0145] Step 322) The adaptive control law involves

[0146] Adaptive backstepping control method is used to design the UAV control law:

[0147] The dynamic equation of the attitude error of the UAV is defined as: Among them, e θ is the attitude error vector, θ is the actual attitude angle vector of the UAV, ω d is the desired attitude angular velocity vector, R(θ) is the attitude rotation matrix;

[0148] Step-by-step construction of the Lyapunov function V by backstepping design i , and designed an adaptive control parameter update law based on the Lyapunov stability theory. Among them, the adaptive adjustment term in the adaptive control parameter update law is dynamically adjusted through real-time estimated interference input to ensure that the UAV attitude quickly tracks the changes of the unmanned boat deck.

[0149] For example, for the control of attitude angle θ, the adaptive control law is Among them, k θis the proportional coefficient, which is based on the attitude error e θ The size of the attitude error is calculated by generating a control amount in a certain proportion to quickly reduce the attitude deviation. θ =θ-θ d , θ is the current actual attitude angle of the UAV, θ d is the attitude angle expected to follow the changes of the unmanned boat deck. It is an adaptive adjustment item, which is dynamically adjusted by real-time estimation of the size of environmental interference. In high sea conditions, interference factors such as waves and strong winds are complex and changeable, and the interference force and torque obtained by interference modeling will change in real time. Adaptive algorithms, such as the adaptive control parameter update law based on Lyapunov stability theory, are used to adjust the interference force and torque in real time according to the estimated values ​​of the interference force and torque and the dynamic equation of the attitude error. For example, when the disturbance force increases and the posture deviation tends to increase, the adaptive algorithm adjusts It generates an additional control quantity, which is related to k θ e θ Work together to increase the total control amount u θ , thereby quickly adjusting the drone's attitude, ensuring that the drone's attitude can quickly track changes in the unmanned boat deck, effectively resist environmental interference, and maintain a stable flight attitude.

[0150] Step 33) Prediction Compensation

[0151] Since the movement of unmanned boats in high sea conditions is time-varying and uncertain, in order to improve the accuracy of drone landing, a prediction compensation mechanism is introduced to predict the movement trend of the unmanned boat in advance and adjust the flight trajectory of the drone.

[0152] As mentioned above, the motion model of the unmanned boat considering environmental interference uses the extended Kalman filter (EKF) algorithm or the particle filter (PF) algorithm to predict the motion state of the unmanned boat in the future T time, and obtains the predicted position x in the future T time. u,pred (t+T) and predicted pose θ u,pred (t+T).

[0153] According to the prediction results, the trajectory compensation vector and attitude compensation vector of the UAV are calculated. Let the current position of the UAV be x d , the posture is θ d , then the trajectory compensation vector Δx comp and attitude compensation vector Δθ comp for:

[0154] Δx comp =x u,pred (t+T)-x d ,

[0155] Δθ comp =θu,pred (t+T)-θ d .

[0156] The trajectory compensation and attitude compensation are incorporated into the control instructions of the UAV to adjust the flight trajectory of the UAV so that it can land accurately on the dynamically changing deck of the unmanned boat.

[0157] Through the collaborative work of the above-mentioned visual servo-based landing guidance system, adaptive control algorithm and predictive compensation mechanism, the autonomous landing control strategy of the present invention can effectively overcome the challenges brought by high sea conditions and realize reliable autonomous landing of the UAV on the deck of the unmanned boat.

[0158] The above describes in detail the preferred embodiments of the present invention. It should be understood that those skilled in the art can make numerous modifications and variations based on the concepts of the present invention without inventive effort. Therefore, any technical solutions that can be derived by those skilled in the art through logical analysis, reasoning, or limited experimentation based on the concepts of the present invention and the prior art should be within the scope of protection defined by the claims.

Claims

1. A method for dynamic collaborative path planning and autonomous landing control of a UAV and an unmanned boat, characterized in that: The following steps are involved: Data collection and processing steps: The unmanned boat and drone collect their own status data through their respective multi-sensor systems. At the same time, the environmental monitoring system obtains high sea state environmental data and integrates the collected data; Dynamic collaborative path planning steps: Based on the fused data, an improved rapidly expanding random tree algorithm is used to plan a safe navigation path for the unmanned boat. The generated unmanned boat path is optimized based on the unmanned boat motion model that considers environmental interference. A model predictive control algorithm is used to plan a path for the drone to fly to the predicted landing point of the unmanned boat. Dynamic collaboration between the unmanned boat and the drone is achieved through information exchange between the two. Autonomous landing control steps: The UAV uses the visual servo guidance system to identify the unmanned boat landing mark, calculates the position and attitude deviation relative to the landing point, uses the adaptive control algorithm to adjust the flight control parameters according to the interference situation, and combines the unmanned boat motion model to perform trajectory prediction compensation to achieve autonomous landing.

2. The method for dynamic collaborative path planning and autonomous landing control of a UAV and an unmanned boat according to claim 1, characterized in that: The multi-sensor system carried by the unmanned boat includes an inertial measurement unit, a global positioning system, a lidar and a visual sensor. The collected data include the unmanned boat's attitude information in three-dimensional space, position coordinates, velocity vector, surrounding point cloud data and images of the unmanned boat's deck and surrounding environment; The multi-sensor system carried by the UAV includes an inertial measurement unit, a global positioning system and a visual sensor, and the collected data includes the UAV's own attitude angle, angular velocity, acceleration, position coordinates, velocity vector and environmental image; The environmental monitoring system includes a meteorological satellite and a buoy monitoring system, and the collected data include wave spectrum, wind speed, wind direction and ocean current speed and direction in the local sea area.

3. The method for dynamic collaborative path planning and autonomous landing control of a UAV and an unmanned boat according to claim 2, characterized in that: The fusion processing of the collected data includes: Self-state data fusion: The extended Kalman filter algorithm is used to fuse the inertial measurement unit and global positioning system data of the unmanned boat and drone. After pre-processing the lidar point cloud data, a feature matching-based algorithm is used to fuse it with the image data collected by the visual sensor to extract environmental feature information. Environmental data fusion: The collected environmental data are standardized and fused using a weighted fusion algorithm. The weights of different data in the weighted fusion algorithm are determined based on the degree of influence of each environmental data on the movement of unmanned boats and drones.

4. The method for dynamic collaborative path planning and autonomous landing control of a UAV and an unmanned boat according to claim 1, characterized in that: The improved rapid expansion random tree algorithm adds the unmanned boat posture change constraint condition when planning the path for the unmanned boat, and combines the obstacle information obtained by the laser radar to construct the path, wherein, Taking the unmanned boat as the research object, the state space is defined as: Where (x, y, z) is the three-dimensional position of the UAV in the geographic coordinate system, φ, θ, and ψ are the roll angle, pitch angle, and bow angle, respectively. At the same time, the control space is defined as: Among them, u1 controls the forward speed of the unmanned boat, and u2 controls the steering angle; Improved rapid expansion random tree algorithm based on the current position of the unmanned boat is the starting point and the mission target location Perform random tree expansion for the endpoint, and in each expansion, from the state space Randomly sample a state x rand , calculate the tree node x by the following formula near to x rand Direction vector Among them, ||·|| represents the modulus of the vector; Along direction vector Generate a new node x with a fixed step size Δs new : After generating a new node, determine whether the new node meets the attitude constraint of the unmanned boat and whether it is in the obstacle area based on the point cloud data obtained by the laser radar. If the attitude constraint is met and there is no collision with an obstacle, the new node is added to the random tree; otherwise, the node is resampled and generated, where the attitude constraint includes the roll angle constraint |φ|≤φ max , pitch angle constraint |θ|≤θ max ,φ max and θ max is the maximum allowable angle set according to the structure of the unmanned boat and the sea state level; repeat the above process until a feasible path from the starting point to the end point is found; Considering the interference of waves, wind speed and ocean currents on the movement of the unmanned boat, the additional forces and moments of the unmanned boat under the corresponding interference are calculated through the acquired environmental data. The unmanned boat motion model is established considering the additional forces and moments, and the generated feasible path is optimized.

5. The method for dynamic collaborative path planning and autonomous landing control of a UAV and an unmanned boat according to claim 1, characterized in that: When planning a path for a UAV, the model predictive control algorithm uses the real-time position and predicted landing point of the UAV as targets, combines the UAV's own state data and high sea state environmental data, and uses the UAV motion model that takes environmental interference into account to predict the motion state within a preset time in the future. It then constructs an objective function and solves the optimal control sequence to generate a flight path. The objective function J aims to minimize the time and energy consumption of the UAV to reach the predicted landing point: Among them, λ1 and λ2 are weight coefficients used to balance position error and energy consumption, N is the number of predicted future time steps, is the drone state vector, To predict the landing point, is the control input vector of the UAV, and k is the current moment.

6. The method for dynamic collaborative path planning and autonomous landing control of a UAV and an unmanned boat according to claim 1, characterized in that: The dynamic coordination between the unmanned boat and the drone is achieved through information interaction between the two as follows: The unmanned boat sends its real-time status data, predicted trajectory, and environmental data to the drone, and the drone feeds back its status data and current path planning results to the unmanned boat. When the motion state of the unmanned boat changes due to high sea conditions and exceeds a preset threshold, the unmanned boat immediately sends the new state data and predicted trajectory to the drone; after receiving the information, the drone re-runs the model predictive control algorithm, takes the new predicted landing point of the unmanned boat as the target, and optimizes and adjusts the flight path based on the updated environmental data and its own state; at the same time, the drone feeds the adjusted path back to the unmanned boat, and the unmanned boat fine-tunes its own navigation path based on the feedback information.

7. The method for dynamic collaborative path planning and autonomous landing control of a UAV and an unmanned boat according to claim 1, characterized in that: The UAV identifies the landing mark of the unmanned boat through the visual servo guidance system: A specific landing mark is set on the deck of the unmanned boat. The drone collects images through visual sensors, processes the collected images using target detection algorithms, identifies the landing mark on the deck of the unmanned boat, and extracts the corner features on the mark using corner detection algorithms.

8. The method for dynamic collaborative path planning and autonomous landing control of a UAV and an unmanned boat according to claim 1, characterized in that: The method for calculating the position and attitude deviation relative to the landing point is specifically as follows: According to the pinhole camera model, the transformation relationship between the image coordinate system and the UAV body coordinate system and the UAV coordinate system is established: Assume that the intrinsic parameter matrix of the drone camera is where f x ,f y is the focal length, c x ,c y is the main point coordinate; The three-dimensional coordinates P of the landing mark in the unmanned boat coordinate system are known i =(X i ,Y i ,Z i ) T , calculate its projection coordinates in the image coordinate system through the perspective projection transformation formula, the perspective projection transformation formula is: Among them, s is the scale factor, R is the rotation matrix, t is the translation vector, and u i 、v i They are corner feature p i The horizontal and vertical coordinates in the image coordinate system; The above equations are solved by the least squares method to obtain the estimated values ​​of the rotation matrix and translation vector. The estimated rotation matrix is ​​decomposed to obtain the attitude deviation of the UAV relative to the UAV. The position deviation of the UAV relative to the UAV is determined based on the components of the estimated translation vector in the directions of the three coordinate axes.

9. The method for dynamic collaborative path planning and autonomous landing control of a UAV and an unmanned boat according to claim 1, characterized in that: The adaptive control algorithm comprises the following steps: The environmental disturbance factors are equivalently modeled as disturbance forces and disturbance torques acting on the UAV, which serve as disturbance inputs in the adaptive control algorithm. Adaptive backstepping control method is used to design the UAV control law: The dynamic equation of the attitude error of the UAV is defined as: Among them, e θ is the attitude error vector, θ is the actual attitude angle vector of the UAV, ω d is the desired attitude angular velocity vector, R(θ) is the attitude rotation matrix; Step-by-step construction of the Lyapunov function V by backstepping design i , and designed an adaptive control parameter update law based on the Lyapunov stability theory. Among them, the adaptive adjustment term in the adaptive control parameter update law is dynamically adjusted through the real-time estimated interference input to ensure that the UAV attitude quickly tracks the changes of the unmanned boat deck.

10. The method for dynamic collaborative path planning and autonomous landing control of a UAV and an unmanned boat according to claim 1, characterized in that: In the prediction and compensation step, based on the unmanned boat motion model taking into account environmental interference, the extended Kalman filter algorithm or the particle filter algorithm is used to predict the motion state of the unmanned boat in the future. According to the prediction results, the trajectory compensation vector and attitude compensation vector of the UAV are calculated and incorporated into the UAV control instructions to adjust the UAV flight trajectory.

Images