Unmanned aerial vehicle water operation control method and system, medium and program product

Abstract

The invention discloses an unmanned aerial vehicle overwater operation control method and system, a medium and a program product, and relates to the field of general control or regulation systems.The method comprises the steps that an unmanned aerial vehicle is controlled to fly to a preset approach route to a target landing area, and the horizontal speed of the unmanned aerial vehicle is adjusted to a preset horizontal approach speed; the predetermined approach route comprises a parabolic trajectory; the operation power of all rotor motors is controlled to be reduced to a preset shutdown threshold value, so that the unmanned aerial vehicle descends to the water surface along a parabolic path based on the horizontal inertial speed and the gravity effect; the vertical distance between the unmanned aerial vehicle and the water surface is monitored in real time, and an instantaneous height value is obtained through calculation; when the instantaneous height value is reduced to a preset triggering height, controlling a rotor motor to operate so as to adjust the unmanned aerial vehicle to contact with the water surface in a preset horizontal attitude; and when the unmanned aerial vehicle is in the water surface inversion state, the unmanned aerial vehicle is controlled to execute a turnover action. According to the invention, the cruising ability of the unmanned aerial vehicle in water operation can be improved.

Description

Technical Field

[0001] This application relates to the general field of control or regulation systems, and more particularly to a method, system, medium, and program product for controlling unmanned aerial vehicles (UAVs) on water. Background Technology

[0002] With the rapid development of drone technology, its application areas have expanded from traditional aerial photography and inspection to more complex and specialized scenarios. Especially in water-related tasks such as hydrological surveys, water quality monitoring, marine resource investigations, and maritime emergency search and rescue, drones are playing an increasingly important role due to their efficiency and flexibility. In these applications, drones not only need to perform tasks in the air, but also often need to have the ability to land, moor, and take off again on the surface of the target water area.

[0003] In related technologies, a water landing scheme based on vertical attitude control is provided to achieve safe landing of unmanned aerial vehicles (UAVs) on water. This scheme typically involves the UAV hovering after reaching the target water area, using onboard altitude sensors (such as millimeter-wave radar or lidar) to measure the distance between the UAV and the dynamic water surface in real time. Subsequently, based on the altitude information fed back from the sensors, the flight control system controls the UAV's four or more rotor motors to descend at a preset, low vertical speed. Upon reaching a lower altitude, kinetic energy is recovered, and the UAV is guided to land in the water.

[0004] However, the high-frequency dynamic adjustments made by the flight control system in related technologies to maintain attitude stability result in significant energy loss. At the same time, the slow descent process causes the drone to hover in a high-power state for too long, resulting in excessive power consumption per landing and affecting the drone's already limited endurance. Summary of the Invention

[0005] This application provides a method, system, medium, and program product for controlling unmanned aerial vehicles (UAVs) on water, which can improve the endurance of UAVs on water.

[0006] In a first aspect, this application provides a method for controlling unmanned aerial vehicle (UAV) operations on water, applied to a control system. The method includes: controlling the UAV to fly to a predetermined approach route to a target water landing area, and adjusting the UAV's horizontal speed to a preset horizontal approach speed; the predetermined approach route includes a parabolic trajectory; controlling the operating power of all rotor motors to decrease to a preset shutdown threshold, so that the UAV descends towards the water surface along a parabolic trajectory based on horizontal inertial velocity and gravity; monitoring the vertical distance between the UAV and the water surface in real time and calculating the instantaneous altitude value; when the instantaneous altitude value decreases to a preset trigger altitude, controlling the rotor motors to operate to adjust the UAV to contact the water surface with a preset horizontal attitude; and controlling the UAV to perform a flipping action when the UAV is in an inverted state on the water surface.

[0007] In the above embodiment, the control system controls the UAV to glide parabolically using inertia, shutting off the motors for most of the descent, significantly reducing the energy consumption of traditional hovering descent methods. The motors are only activated momentarily before landing on the water for attitude adjustment, ensuring a safe and stable landing, and the UAV finally returns to operational status through a rollover maneuver. By saving energy while ensuring safety, the system effectively extends the UAV's single-operation time and improves its overall endurance.

[0008] In conjunction with some embodiments of the first aspect, in some embodiments, the step of controlling the UAV to fly to a predetermined approach route to the target landing area and adjusting the UAV's horizontal speed to a preset horizontal approach speed specifically includes: determining the target landing point in the target landing area and calculating the real-time azimuth angle from the current position to the target landing point; calculating the direction adjustment amount and speed adjustment amount based on the real-time azimuth angle, the UAV's current speed, and the current position; planning and generating a transition route for the UAV to adjust its attitude and speed from the current position, and a predetermined approach route for landing at the target landing point after parameter adjustment, based on the direction adjustment amount and speed adjustment amount; and adjusting the UAV's flight parameters so that the UAV enters the predetermined approach route from the transition route at a horizontal speed of the preset horizontal approach speed.

[0009] In the above embodiments, the control system can accurately plan a path that smoothly transitions from the current position to the predetermined approach route based on the real-time status of the UAV and the target position. This ensures that the UAV can enter the energy-saving gliding phase in the optimal way, avoids additional maneuvers due to inaccurate route alignment, and improves the automation level and trajectory accuracy of the water landing process.

[0010] In conjunction with some embodiments of the first aspect, in some embodiments, after determining the target landing point in the target landing area and calculating the real-time azimuth angle from the current position to the target landing point, the method further includes: acquiring the distribution information of water surface obstacles and dynamic parameters of water flow in the target landing area, and constructing a three-dimensional obstacle map; calculating a safe flight corridor based on the three-dimensional obstacle map and the UAV's body size parameters; and optimizing the predetermined approach route based on the safe flight corridor to generate an optimal approach path so that the UAV does not collide with water surface obstacles.

[0011] In the above embodiments, the control system actively senses and analyzes the water environment before planning the flight path. By constructing a three-dimensional obstacle map and calculating a safe flight corridor, the system intelligently optimizes the predetermined approach route, enabling the UAV to actively avoid static and dynamic obstacles on the water surface. This improves the safety of landing in complex water environments, effectively prevents equipment damage or mission failure due to collisions, and ensures the smooth progress of the entire water operation.

[0012] In some embodiments of the first aspect, prior to the step of calculating the direction adjustment and speed adjustment based on the real-time azimuth angle, the current speed of the UAV, and the current position, the method further includes: determining the free deceleration parameters of the UAV on the predetermined approach route based on environmental data and the flight test data of the UAV; and determining the preset horizontal approach speed by back-calculation based on the free deceleration parameters and the boundary condition that the horizontal speed of the UAV at the target landing point is zero.

[0013] In the above embodiments, the control system can accurately predict the deceleration characteristics of the UAV during the unpowered gliding phase based on actual flight data and environmental influences. By calculating the optimal initial approach velocity, it ensures that the UAV's horizontal velocity is exactly zero when it reaches the target point. This maximizes the accuracy of the water landing, avoids landing point deviations caused by inaccurate velocity estimation, and further improves energy efficiency.

[0014] In conjunction with some embodiments of the first aspect, in some embodiments, after determining the preset horizontal approach speed by back-calculation based on the free deceleration parameters and the boundary condition that the horizontal speed of the UAV at the target landing point is zero, the method further includes: constructing a parabolic trajectory of a predetermined approach route based on the preset horizontal approach speed and the free deceleration parameters; determining the initial flight altitude of the UAV on the parabolic trajectory, so that the UAV reaches the target landing point when the horizontal speed is zero.

[0015] In the above embodiments, after determining the optimal approach speed, the control system further constructs a complete and executable parabolic trajectory and determines the precise initial altitude for entering this trajectory. This makes the entire energy-saving descent process fully parameterized and predictable, ensuring that the UAV can fly strictly according to the preset energy-optimal path and finally arrive at the target point exactly at the moment when the horizontal speed is exhausted. This achieves perfect coordination in time and space, ensuring the reliability and accuracy of the water landing scheme.

[0016] In conjunction with some embodiments of the first aspect, in some embodiments, after the step of real-time monitoring of the vertical distance between the UAV and the water surface and calculating the instantaneous altitude value, the method further includes: calculating the real-time vertical descent speed of the UAV; and when the vertical descent speed exceeds a preset safe descent speed threshold, controlling the rotor motor to operate to reduce the real-time vertical descent speed of the UAV.

[0017] In the above embodiments, the control system adds a crucial safety safeguard to the unpowered gliding descent process. By monitoring and controlling the vertical descent speed in real time, it can effectively cope with abnormal environmental factors such as sudden downdrafts, preventing the drone from violently crashing into the water surface due to excessive descent speed. This proactive intervention mechanism improves the safety of the water landing process without significantly increasing energy consumption, ensuring the structural safety of the drone under various weather conditions.

[0018] In conjunction with some embodiments of the first aspect, in some embodiments, after controlling the drone to perform a flipping action when the drone is in an inverted state on the water surface, the method further includes: acquiring real-time wind direction data and wind speed intensity information of the target water-landing area, calculating the influence coefficient of wind force on the drone's water surface attitude; determining wind force compensation parameters based on the influence coefficient and the drone's current water surface position; adjusting the drone's water surface operation parameters based on the wind force compensation parameters, generating a water surface navigation trajectory, so that the drone remains stable under the action of wind force.

[0019] In the above embodiments, the control system extends the UAV's operational capabilities from the air to the water surface. After the UAV lands on the water, it can actively adjust its attitude or navigation parameters based on real-time wind information to counteract the effects of wind. This ensures that the UAV can maintain a stable position or navigate along a predetermined trajectory when performing tasks such as monitoring, sampling, or acting as a relay station on the water surface, expanding the application scenarios of the UAV and improving its stability and reliability in water surface operations.

[0020] In a second aspect, embodiments of this application provide a control system comprising: one or more processors and a memory; the memory being coupled to the one or more processors, the memory being used to store computer program code, the computer program code including computer instructions, the one or more processors invoking the computer instructions to cause the control system to perform the method described in the first aspect and any possible implementation thereof.

[0021] Thirdly, embodiments of this application provide a computer program product containing instructions that, when the computer program product is run on a control system, cause the control system to perform the method described in the first aspect and any possible implementation thereof.

[0022] Fourthly, embodiments of this application provide a computer-readable storage medium including instructions that, when executed on a control system, cause the control system to perform the method described in the first aspect and any possible implementation thereof.

[0023] Understandably, the control system provided in the second aspect, the computer program product provided in the third aspect, and the computer storage medium provided in the fourth aspect are all used to execute the methods provided in the embodiments of this application. Therefore, the beneficial effects they can achieve can be referred to the beneficial effects in the corresponding methods, and will not be repeated here.

[0024] One or more technical solutions provided in the embodiments of this application have at least the following technical effects or advantages: 1. By employing a technical solution that reduces the operating power of all rotor motors to a preset shutdown threshold after the UAV enters the predetermined approach route, allowing it to descend towards the water surface along a parabolic trajectory based on horizontal inertial velocity and gravity, and only re-controlling the rotor motors to adjust its attitude and contact the water surface when the instantaneous altitude value drops to a preset trigger altitude, the UAV can glide in a nearly unpowered manner for most of the water landing process, avoiding the situation in traditional vertical descent schemes where a large amount of electrical energy is continuously consumed to maintain hovering and slow descent.

[0025] 2. Because the system adopts a technical solution that first determines the target landing point and calculates the real-time azimuth angle when planning the flight path, and then calculates the direction and speed adjustment based on the current speed, position and other dynamic information of the UAV, and generates a transition flight path for attitude and speed adjustment to guide the UAV to smoothly enter the predetermined approach flight path, the control system can ensure that the UAV completes the flight path alignment and parameter adjustment in an efficient, accurate and automated manner before performing complex parabolic landing.

[0026] 3. Because the system adopts a technical solution that uses actual environmental data and the UAV's own flight test data to determine its free deceleration parameters on the predetermined approach route, and then uses this as a basis to back-calculate the preset horizontal approach speed by combining the boundary condition that the horizontal speed of the target landing point is zero, the control system can accurately model and predict the UAV's trajectory during the unpowered gliding phase, and fully consider the influence of real factors such as air resistance on the deceleration process. Attached Figure Description

[0027] Figure 1 This is a flowchart illustrating a method for controlling unmanned aerial vehicle (UAV) operations on water in an embodiment of this application. Figure 2 This is another flowchart illustrating the UAV waterborne operation control method in the embodiments of this application; Figure 3 This is a schematic diagram of the physical device structure of a control system in an embodiment of this application. Detailed Implementation

[0028] The terminology used in the following embodiments of this application is for the purpose of describing particular embodiments only and is not intended to be limiting of this application. As used in the specification of this application, the singular expressions “a,” “an,” “the,” “the,” and “this” are intended to include the plural expressions as well, unless the context clearly indicates otherwise. It should also be understood that the term “and / or” as used in this application refers to any or all possible combinations including one or more of the listed items.

[0029] Hereinafter, the terms "first" and "second" are used for descriptive purposes only and should not be construed as implying or suggesting relative importance or implicitly indicating the number of indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature, and in the description of the embodiments of this application, unless otherwise stated, "multiple" means two or more.

[0030] In the specific implementation scenario of this application, the various technical terms and processing logic together constitute a closed loop for efficient, energy-saving, and safe UAV water operations. For example, in a reservoir water quality sampling mission, the operator first designates a sampling point on a map, which is the center of the target water-landing area. The control system then plans a predetermined approach route to this area. This route is not a simple straight line, but includes a calculated parabolic trajectory, which is key to achieving energy saving. When the UAV reaches the starting point of this trajectory, the control system reduces the motor power to a preset shutdown threshold, such as maintaining only the minimum power supply of the system, allowing the UAV to begin gliding without power. During this process, the onboard millimeter-wave radar continuously measures the vertical distance between the UAV and the undulating water surface, obtaining a series of instantaneous altitude values. When the altitude drops to, for example, a preset trigger altitude of 3 meters, the system determines that a final attitude adjustment is needed, so it wakes up the motors, allowing the UAV to gently touch the water in a preset horizontal attitude. In addition, the UAV is equipped with a flipping function and waterproof devices. After entering the water, it can perform a flipping action, generating torque through differential motor speed to turn the fuselage face up, facilitating sampling work.

[0031] The following describes the process of the method provided in this implementation. Please refer to [link / reference]. Figure 1 This is a flowchart illustrating a method for controlling unmanned aerial vehicle (UAV) operations on water in an embodiment of this application.

[0032] S101. Control the UAV to fly to the predetermined approach route to the target water landing area, and adjust the UAV's horizontal speed to the preset horizontal approach speed.

[0033] Among them, the target water landing area refers to the water area where the UAV is planned to land; the planned approach route refers to the flight path pre-planned to achieve parabolic descent; and the preset horizontal approach speed refers to the specific horizontal speed that the UAV needs to reach when entering the starting point of the parabolic trajectory. This speed is a key initial condition for subsequent unpowered gliding to the target point.

[0034] Specifically, this step is executed as the initial stage of the water landing process after the UAV receives the water landing command. The control system first determines the target water landing area based on the mission command or GPS coordinates, and plans a predetermined approach route to that area. Next, the control system controls the UAV to fly from its current position to the starting point of the route, and during the flight, it precisely adjusts its horizontal flight speed to the preset horizontal approach speed by adjusting the motor output power, while simultaneously adjusting the flight altitude to the predetermined value, preparing for the next step of unpowered gliding.

[0035] In some embodiments, this step can be controlled in several ways: Optionally, the control system employs a closed-loop control algorithm based on PID (Proportional-Integral-Derivative), using the waypoints and preset speed of the predetermined approach route as target values. By continuously comparing the actual position and speed fed back by GPS and IMU, the flight control commands are adjusted in real time to guide the UAV to accurately track the route and speed. Optionally, the control system employs a model predictive control (MPC) method to predict the UAV's trajectory over a future period and optimize the control input, enabling it to reach the predetermined approach state more smoothly and energy-efficiently. It is understood that other path planning and tracking algorithms can also be used to achieve the convergence of the UAV to the predetermined route and speed; this is not limited here.

[0036] A key issue that may arise during this process is interference from the external environment, particularly the influence of crosswinds, which can cause the drone to deviate from its intended flight path. To address this, the control system integrates data from GPS, IMU (Inertial Measurement Unit), and pitot tube in real time to accurately estimate wind speed and direction. Once lateral drift is detected, the flight control algorithm automatically calculates a compensation amount. This compensation adjusts the drone's roll angle to generate a lateral force to counteract the wind's influence, while simultaneously adjusting the yaw angle to keep the nose aligned with the heading. This ensures that the drone can fly precisely along its intended approach route even in windy conditions.

[0037] S102. Control the operating power of all rotor motors to a preset shutdown threshold, so that the UAV descends to the water surface along a parabolic trajectory based on horizontal inertial velocity and gravity.

[0038] Among them, the rotor motor refers to the power unit that drives the rotor of the UAV to rotate; the preset shutdown threshold is an extremely low power value, not a complete power cut-off, which is intended to stop the motor from outputting lift but can quickly respond to subsequent start commands; the horizontal inertial velocity is the preset horizontal approach velocity adjusted in step S101; the parabolic trajectory refers to the motion trajectory formed by an object with an initial horizontal velocity under the action of gravity and air resistance alone.

[0039] Specifically, this step is executed when the control system confirms that the UAV has accurately reached the starting point of the predetermined approach route, i.e., the position, altitude, and horizontal speed all meet the preset conditions. The control system sends commands to the electronic speed controllers (ESCs) of all rotor motors to rapidly reduce their operating power to the preset shutdown threshold. At this point, the UAV loses active lift and, under the combined action of its own gravity and initial horizontal speed, begins to glide and descend towards the water surface along an approximately parabolic trajectory, thus entering the core energy-saving stage.

[0040] In some embodiments, motor control for this step can be implemented in several ways: Optionally, the control system directly broadcasts a PWM / DShot signal to all ESCs, reducing the throttle channel value to a minimum (or a preset threshold) to achieve rapid power shutdown; alternatively, the control system executes a preset power ramp-down program to smoothly reduce the power from cruise mode to a shutdown threshold within a few hundred milliseconds, thereby reducing the impact on the engine structure. It is understood that other motor control protocols or methods can also be used to achieve rapid power reduction, and this is not limited to these methods.

[0041] A key issue that may arise during this process is inconsistent motor response, where individual motors fail to descend to the stopping threshold as instructed, causing the drone to rotate or roll unexpectedly during the initial gliding phase. To address this, the control system immediately monitors the current or speed feedback of each motor after issuing the command. If the feedback value of one or more motors fails to drop below the threshold within a specified time, the system determines it as a motor malfunction, immediately aborts the parabolic descent procedure, and executes either an emergency hover or a standard powered descent procedure. The system also reports the malfunction to the ground station to ensure flight safety.

[0042] S103. Real-time monitoring of the vertical distance between the drone and the water surface, and calculation of the instantaneous altitude value.

[0043] The instantaneous altitude value represents the real-time vertical distance between the bottom of the drone's fuselage and the water surface below at a given moment. This value is typically obtained using a specialized altimeter sensor.

[0044] Specifically, from step S102 when the UAV begins its unpowered descent, the control system continuously activates and reads data from the altimeter. Because the water surface may fluctuate due to waves, the raw data returned by the sensor may contain noise and fluctuations. Therefore, the control system needs to process this raw data, for example, through filtering algorithms, to calculate a more stable and reliable instantaneous altitude value. This value is the key basis for determining when to trigger the next attitude adjustment.

[0045] In some embodiments, height monitoring in this step can be achieved in several ways: optionally, millimeter-wave radar can be used as the height sensor, which has the advantages of good penetration, is not easily affected by water mist or changes in lighting, and can stably measure the distance to the water surface; optionally, lidar can be used for height measurement, which has the advantages of high measurement accuracy, small spot size, and can more accurately reflect instantaneous wave height. The control system can fuse data from two or more sensors to improve the robustness and accuracy of the measurement. It is understood that other ranging technologies such as ultrasonic waves can also be used, and this is not limited here.

[0046] A key issue that may arise during this step is the dramatic fluctuations in altitude measurement data caused by water waves. Using the raw data directly could lead to errors in subsequent altitude determination, either too early or too late. To address this, the control system typically employs a Kalman filter to process the altitude data. This filter combines the UAV's kinematic model (predicted value) with sensor measurements (observed value), effectively filtering out high-frequency noise and outputting a smooth estimate that more closely approximates the true altitude. In this way, even on turbulent water, the system can obtain a stable instantaneous altitude value, ensuring accurate triggering of subsequent actions.

[0047] It's important to note that the model parameters in the Kalman filter here primarily refer to the process noise covariance matrix Q and the measurement noise covariance matrix R. These parameters are not obtained through conventional machine learning "training," but rather optimized through "tuning." The specific method is as follows: First, under stationary or known motion conditions, a large number of readings from altimeters (such as millimeter-wave radar) are collected, and their variances are statistically analyzed. A typical value of this variance can be used as the initial value for the measurement noise R. For example, if the radar data fluctuates around the mean at a true altitude of 10 meters with a standard deviation of 0.1 meters, then R can be set to 0.1^2 = 0.01. Second, the process noise Q represents the uncertainty in the model's predictions, mainly originating from unmodeled dynamics (such as sudden airflow) and IMU noise. During actual flight, the value of Q can be continuously adjusted by comparing the filter output with higher-precision ground truth values ​​(such as differential GPS) until the filter's tracking performance (both quickly responding to real changes and effectively suppressing noise) reaches its optimal level. The standard for training (tuning) is to minimize the covariance of the state estimation error so that the filter output is as close as possible to the true state overall.

[0048] This Kalman filter is a linear dynamic system model. Its core is the state vector X, which in this scenario contains at least two states: the UAV's vertical height h relative to the water surface and its vertical velocity v_z, i.e., X = [h, v_z]^T. The model's inputs include: 1) the state estimate X_{k-1} from the previous time step; 2) the vertical acceleration a_z provided by the IMU (Inertial Measurement Unit); and 3) the current altitude measurement Z_k provided by the altimeter. The model is defined by two equations: State transition equation (prediction model): X_k = F*X_{k-1} + G*u_{k-1}. Where F is the state transition matrix, describing how the state evolves over time, in the form [[1, dt], [0, 1]]; G is the control input matrix, in the form [[0.5*dt^2], [dt]]; and u is the control input, i.e., the vertical acceleration a_z measured by the IMU. This equation predicts the current altitude and velocity.

[0049] Observation equation (measurement model): Z_k = H*X_k + v_k. Where H is the observation matrix, which describes how the state is mapped to the measurement value, in the form of [1, 0], indicating that only the height h can be directly measured; v_k is the measurement noise.

[0050] The model output is the optimal estimate of the current state after filtering, X_k_hat=[h_hat,v_z_hat]^T, which is the optimal instantaneous height value and vertical descent speed.

[0051] During each control cycle of the UAV's gliding descent (e.g., every 20 milliseconds), the control system executes a Kalman filter loop. First, using the state transition equation and the optimal estimate from the previous time step, the prior state estimate for the current time step is predicted. Then, the latest altitude reading Z_k from an altimeter such as radar is acquired. Next, the residual between the prior estimate and the actual measurement is calculated. Finally, based on the Kalman gain (calculated from Q, R, and the state covariance matrix), the prior estimate is corrected using the residual to obtain the posterior (optimal) state estimate for the current time step. This output altitude h_hat is used for trigger altitude determination in S104, and the speed v_z_hat is used for safe speed monitoring in S208, thereby achieving effective suppression of noise interference and accurate control of critical states.

[0052] S104. When the instantaneous altitude value drops to the preset trigger altitude, control the rotor motor to operate, so as to adjust the UAV to contact the water surface in a preset horizontal attitude.

[0053] The preset trigger height is a height threshold set in advance based on the drone's performance and safety requirements, such as 2-5 meters; the preset horizontal attitude refers to the state where the drone's pitch and roll angles are close to zero degrees to ensure a smooth water landing.

[0054] Specifically, during the unpowered descent of the drone, the control system continuously compares the instantaneous altitude value calculated in S103 with the preset trigger altitude. Once it detects that the instantaneous altitude value is less than or equal to the preset trigger altitude, the control system immediately executes this step. It quickly sends start and speed adjustment commands to all rotor motors, using the thrust generated by the motors in a short time to rapidly adjust the drone's attitude, eliminating any tilt that may occur during gliding, and keeping it as level as possible at the moment of contact with the water surface, thereby reducing impact and preventing rollover.

[0055] In some embodiments, attitude adjustment in this step can be achieved in several ways: Optionally, the system invokes a high-gain attitude stabilization PID controller. Once triggered, this controller will forcefully and rapidly adjust the speed difference between the motors based on the attitude error fed back by the IMU, quickly leveling the aircraft. Optionally, the system executes a pre-programmed open-loop control sequence, that is, based on the attitude at the time of triggering, instantaneously provides a pre-calibrated pulse power to a specific motor to generate a reverse overturning torque to quickly correct the attitude. It is understood that other advanced attitude control algorithms, such as sliding mode control, can also be used, and are not limited here.

[0056] A key issue that may arise during this process is the delay in the motor's rotation from a standstill to generating effective thrust. If the trigger altitude is set too low, the drone may hit the water before it can complete attitude adjustment. To address this, the preset trigger altitude setting needs to be thoroughly tested and calculated. This value must not only consider the time required for attitude adjustment but also factors such as motor start-up delay and aerodynamic response delay, while also allowing for a certain safety margin. For example, if testing shows that it takes 0.5 seconds for the drone to stabilize from the start command, during which time the drone will descend 2 meters, then the trigger altitude must be at least 2 meters, typically set to 3 meters or higher, to ensure sufficient time and altitude for adjustment.

[0057] S105. When the drone is in an upside-down state on the water surface, control the drone to perform a flipping action.

[0058] Among them, the inverted state on the water surface refers to the stable floating state in which the UAV naturally flips over after entering the water due to its structural design (such as the top float), with its belly facing up and its top facing down; the flipping action refers to an active and controlled change in the body attitude of the UAV, which restores the UAV from the inverted state to the normal operating state with its belly facing down.

[0059] Specifically, this step is executed once the control system detects through IMU data (especially accelerometer and gyroscope data) that the drone has successfully landed on the water and is stable in an inverted position. The control system activates a special flight control mode, precisely controlling some or all of the rotor motors to generate unbalanced thrust, thereby creating a strong flipping torque on the water surface, causing the drone to flip 180 degrees around its longitudinal or transverse axis. After completing the flip, the drone can then perform water surface operations or prepare for takeoff again.

[0060] In some embodiments, the flipping action can be achieved in several ways: Optionally, the control system controls two diagonally opposite motors to rotate at high speed, while the other two motors remain stationary or at low speed, using the generated strong torque to flip the drone; alternatively, if the drone is equipped with waterproof motors and propellers, one side of the motor can be controlled to rotate forward while the other side rotates in reverse (if the ESC supports this), completing the flipping on the water surface in a manner similar to differential steering. It is understood that mechanical methods such as internal counterweight movement can also be used to assist in achieving the flipping action, but this is not limited here.

[0061] A key issue that may arise during this step is the failure of the rollover maneuver. This could be due to insufficient power to overcome water resistance, or excessive force causing the drone to become unstable again or even submerge completely. To address this, the control logic for the rollover maneuver is typically closed-loop. The control system continuously monitors the IMU's angular velocity and angle data during the rollover. It dynamically adjusts the motor output based on a preset rollover angular velocity curve. If the rollover speed is too slow, power is increased; if the speed is too fast or the angle is about to overshoot, power is reduced or a reverse motor is activated for braking, thus achieving a fast yet controlled and smooth rollover process.

[0062] It should be noted that this step first defines the target's roll state, for example, the target roll angle is 180 degrees. The control system uses a PID controller to generate differential thrust commands. The input error e(t) of this controller is the difference between the target angle and the current angle fed back by the IMU, i.e., e(t) = 180° - roll_current(t). The controller's output Output(t) consists of three terms: proportional, integral, and derivative: Output(t) = Kp*e(t) + Ki*∫e(t)dt + Kd*de(t) / dt. The proportional term Kp*e(t) provides the main flipping torque; the larger the error, the larger the torque. The differential term Kd*de(t) / dt acts as damping, used to suppress oscillations during the flipping process and prevent angular overshoot; de(t) / dt is the roll angular velocity measured by the IMU. The integral term Ki*∫e(t)dt is used to eliminate steady-state errors. For example, when the UAV has difficulty flipping to exactly 180 degrees due to water flow or slight asymmetry, the persistent small error will accumulate through the integral term, eventually generating sufficient torque to overcome the resistance. The Output(t) value output by the controller is ultimately mapped to the thrust difference ΔT between the diagonal motors (such as motors 1 and 3) and the other two diagonal motors (motors 2 and 4), thereby generating a precise flipping torque. For example, if Output(t) is positive, the speed of motors 1 and 3 is increased, while the speed of motors 2 and 4 is decreased, achieving controlled flipping.

[0063] The following provides a more detailed description of the process of the method provided in this implementation. Please refer to [link / reference]. Figure 2 This is another flowchart illustrating the UAV waterborne operation control method in this application embodiment.

[0064] S201. Determine the target landing point within the target landing area and calculate the real-time azimuth angle from the current position to the target landing point.

[0065] Among them, the target landing point is a precise geographical coordinate (latitude and longitude) in the target landing area, which is the final target of the UAV landing; the current position refers to the real-time geographical coordinates obtained by the UAV through GPS or other positioning systems; the real-time azimuth angle is the angle between the direction from the current position of the UAV to the target landing point and due north.

[0066] Specifically, the control system executes this step first upon mission commencement or receipt of a landing command. The system obtains the precise coordinates of the target landing point from the mission planning module or ground station instructions. Simultaneously, it continuously reads the UAV's current position coordinates from the onboard GPS module. Then, using these two coordinates, the system calculates the azimuth angle of the target landing point relative to the UAV's current position in real-time and at high frequency, employing spherical trigonometric functions (such as the Haversine formula) or a small-scale planar geometric approximation. This azimuth angle is the core basis for subsequent heading adjustments.

[0067] In some embodiments, this calculation can be performed in several ways: Optionally, the flight management unit (FMU) within the control system can directly integrate a function library for calculating azimuth and distance, allowing for quick output of results upon input of two GPS coordinates; alternatively, the calculation can be performed at the ground control station, and the calculated target azimuth can be sent to the UAV as part of the navigation instructions. It is understood that any effective mathematical calculation method can be used as long as the direction can be determined from the coordinates of two points, and no limitation is made here.

[0068] A key issue that may arise during this step is the fluctuation or temporary loss of GPS signal accuracy, leading to inaccurate calculations of the current position and real-time azimuth, thus affecting course stability. To address this problem, the control system typically employs multi-sensor fusion positioning technology. This technology fuses the absolute position information from GPS with high-frequency attitude and acceleration information provided by the IMU (e.g., through an extended Kalman filter, EKF). When GPS signals are temporarily weak, the system primarily relies on dead reckoning by the IMU to maintain the estimated position and course, correcting the data once the GPS signal is restored. This results in a smoother and more reliable output of the current position and azimuth, improving navigation accuracy.

[0069] S202. Calculate the direction adjustment amount and speed adjustment amount based on the real-time azimuth angle, the current speed of the UAV and its current position.

[0070] Among them, the direction adjustment amount refers to the angle value that the current heading needs to be changed in order to make the UAV face the target landing point; the current speed includes the horizontal speed and vertical speed of the UAV, which is calculated by the fusion of sensors such as GPS and IMU; the speed adjustment amount refers to the amount by which the current speed needs to be increased or decreased in order to achieve the preset approach speed.

[0071] Specifically, after S201 calculates the real-time azimuth angle, the control system immediately executes this step. It compares the real-time azimuth angle with the current heading angle fed back by the UAV's IMU; the difference is the direction adjustment amount. Simultaneously, the system compares the UAV's current horizontal speed with the target approach speed (a preset value or a value dynamically calculated based on distance); the difference is the speed adjustment amount. These two adjustment amounts are quantified control targets that directly guide the next steps in flight path planning and flight parameter adjustments.

[0072] In some embodiments, the adjustment calculation in this step can be implemented in several ways: Optionally, the direction adjustment can be directly taken as the shortest angle difference between the real-time azimuth and the current heading angle; the speed adjustment is calculated through a proportional controller, i.e., the adjustment is proportional to the speed error; Optionally, more complex navigation algorithms, such as L1 navigation or nonlinear guidance law (NLGL), can be used. These algorithms not only calculate the instantaneous adjustment but also generate a smooth transition trajectory, allowing the UAV to naturally enter the target heading. It is understood that any algorithm capable of generating control error signals based on the current state and the target state can be applied, and no limitation is made here.

[0073] A key challenge during this process is balancing the priority and magnitude of directional and speed adjustments. Improper adjustments can lead to drastic changes in the UAV's flight attitude or energy waste. To address this, the control system employs a cooperative control strategy. For example, when there is a significant directional deviation, the system prioritizes steering maneuvers and appropriately reduces speed to minimize the turning radius and energy consumption. Once the direction is roughly aligned, speed adjustments are then the primary focus. Furthermore, the calculation of adjustments is constrained by dynamic constraints such as the maximum permissible tilt angle and maximum acceleration, ensuring that all adjustments are performed within the UAV's safe flight envelope, achieving smooth and efficient flight state transitions.

[0074] S203. Based on the directional and speed adjustment amounts, plan and generate a transitional flight path for the UAV to adjust its attitude and speed from its current position, and a predetermined approach flight path for the UAV to land at the target water landing point after parameter adjustments.

[0075] The transition route is a dynamically generated temporary path connecting the UAV's current location with the starting point of the planned approach route; the planned approach route is the main route that includes a key parabolic trajectory and is used for final landing. The core of this step is to seamlessly connect the two segments.

[0076] Specifically, after calculating the direction and speed adjustments, the control system performs path synthesis. Using the adjustments obtained in step S202, combined with the UAV's performance model (such as maximum turning angular velocity, maximum acceleration / deceleration, etc.), it plans a practically flyable transition route. The goal of this route is that at its endpoint, the UAV's heading, speed, and position perfectly match the starting parameters of the predetermined approach route. Simultaneously, the system also finalizes or loads the complete predetermined approach route. This forms a complete and continuous flight path from the current position to the target landing point.

[0077] In some embodiments, the path generation in this step can be achieved in several ways: Optionally, the transition path consists of a series of dense intermediate waypoints, which the UAV flies over sequentially to complete attitude and speed adjustments; alternatively, the transition path is represented by a parametric curve (such as a Bézier curve or a Dubins curve), which generates a smoother path, beneficial for flight stability and energy saving. It is understood that any path planning algorithm can be used as long as a flyable path connecting the current state and the target state can be generated, and no limitation is made here.

[0078] A key issue that may arise during this step is that the planned transition routes might be overly complex or abrupt, exceeding the UAV's maneuverability, or requiring frequent replanning in dynamically changing environments, resulting in a heavy computational burden. To address this, the control system employs a layered strategy during planning. The upper layer performs coarse global path planning to determine the general direction; the lower layer performs fine-grained local trajectory generation, responsible for smoothing and satisfying dynamic constraints. Simultaneously, the planning algorithm sets a replanning trigger threshold. A complete replanning is only initiated when the UAV deviates significantly from the current planned trajectory. Within small deviations, the flight control's closed-loop tracking capability automatically corrects the deviation, thus balancing path optimality and computational efficiency.

[0079] S204. Adjust the flight parameters of the UAV so that the UAV enters the predetermined approach route from the transition route at a preset horizontal approach speed.

[0080] The flight parameters mainly refer to the UAV's attitude angles (pitch, roll, yaw) and the rotational speeds of each rotor motor. This step is the execution phase planned in step S203.

[0081] Specifically, after the complete flight path is generated in step S203, the control system immediately begins executing this step. It decomposes the transition flight path into a series of continuous control commands, precisely adjusting the UAV's flight parameters through the underlying attitude and position controllers. For example, to achieve turning, the roll angle is controlled; to achieve acceleration and deceleration, the pitch angle and total throttle are controlled. The entire process is a closed-loop feedback control. The system continuously monitors the UAV's status, ensuring that it flies strictly along the transition flight path, and ultimately, at the moment of entering the predetermined approach flight path, its horizontal speed is exactly equal to the preset horizontal approach speed.

[0082] In some embodiments, parameter adjustment in this step can be achieved in several ways: Optionally, a cascaded PID control structure can be used, with the outer loop being a position controller that outputs the desired speed based on the position error; the middle loop being a speed controller that outputs the desired attitude angle based on the speed error; and the inner loop being an attitude controller that outputs control inputs to each motor based on the attitude error. Optionally, more advanced control methods, such as adaptive control or robust control, can be used. These methods can better cope with model uncertainties and external disturbances, improving tracking accuracy. It is understood that any flight control algorithm capable of accurately tracking a predetermined trajectory and velocity profile can be applied, and no limitation is made here.

[0083] A key issue that may arise during this step is energy management. If the adjustment process is too aggressive, with frequent large accelerations, decelerations, and turns, it will consume a significant amount of unnecessary energy, contradicting the initial goal of energy-efficient water landing. To address this, the control system follows an energy-optimal principle when adjusting parameters. In step S203, when planning the flight path, energy consumption is considered as an optimization indicator, generating a transition path that is as smooth as possible, avoiding abrupt maneuvers. During execution, the flight control algorithm also limits the amplitude and rate of change of control commands, prioritizing more energy-efficient control methods (e.g., slow acceleration through small pitch angles rather than rapid throttle), thereby minimizing energy consumption before entering the main flight path while ensuring tracking accuracy.

[0084] S205: Control the operating power of all rotor motors to a preset shutdown threshold, so that the UAV descends to the water surface along a parabolic trajectory based on horizontal inertial velocity and gravity.

[0085] Refer to step S102, which will not be repeated here.

[0086] S206. Real-time monitoring of the vertical distance between the drone and the water surface to calculate the instantaneous altitude value.

[0087] Refer to step S103, which will not be repeated here.

[0088] S207. Calculate the real-time vertical descent speed of the drone.

[0089] The real-time vertical descent speed refers to the speed of the drone in the vertical direction, which is usually obtained by differentiating the altitude value over time.

[0090] Specifically, after the drone begins its descent, the control system performs this step while continuously acquiring instantaneous altitude values ​​(step S206). The system records altitude values ​​at a series of consecutive time points and estimates the current vertical velocity through differential calculations (e.g., (current altitude - previous altitude) / time interval). To obtain a smoother and more accurate velocity value, a sliding window is typically used to perform linear regression fitting on multiple consecutive altitude values; the slope of this regression represents a more reliable vertical descent velocity. This velocity value is a key parameter for determining whether the descent process is safe.

[0091] In some embodiments, the velocity calculation in this step can be implemented in several ways: Optionally, the velocity estimate can be obtained by directly performing a first-order difference on the altitude estimate after Kalman filtering; alternatively, the vertical velocity can be included as one of the state variables and directly incorporated into the state equation of the Kalman filter mentioned in S206, and the filter will output the optimal altitude estimate and velocity estimate simultaneously. It is understood that the vertical velocity can also be directly measured using an airborne Doppler radar or laser velocimeter, which is not limited here.

[0092] A key issue that may arise during this step is that when calculating velocity based on altitude difference, noise from altitude measurements can be amplified, leading to drastic fluctuations in the calculated velocity value, making it unsuitable for direct judgment. To address this, in addition to the aforementioned filtering methods, the control system also incorporates vertical acceleration data from the IMU (Inertial Measurement Unit). By integrating the vertical acceleration measured by the IMU (subtracting the gravitational acceleration g), a change in vertical velocity can also be obtained. Fusing this velocity calculated based on inertia with the velocity calculated based on altitude difference (e.g., using a complementary filter) yields an accurate estimate of vertical velocity that exhibits both good dynamic response and no long-term drift, providing a reliable basis for safety monitoring.

[0093] S208. When the vertical descent speed exceeds the preset safe descent speed threshold, control the rotor motor to operate in order to reduce the real-time vertical descent speed of the UAV.

[0094] The preset safe descent speed threshold is a maximum permissible descent speed set based on the structural strength of the UAV and its ability to withstand water impact, for example, 5 m / s.

[0095] Specifically, this step is a safety monitoring and intervention loop executed in parallel with S207. The control system continuously compares the calculated real-time vertical descent speed with a preset safety threshold. During normal gliding descent, the vertical speed should be within a safe range. However, if abnormal conditions such as strong downdrafts cause the vertical speed to suddenly increase and exceed the threshold, the control system will immediately trigger the intervention mechanism. It will briefly activate the rotor motors to provide an upward thrust (equivalent to "braking") until the vertical descent speed returns to within a safe range, and then reduce the motor power back to the shutdown threshold, allowing the drone to continue gliding.

[0096] In some embodiments, deceleration control in this step can be implemented in several ways: Optionally, a simple threshold controller can be used to apply a fixed, preset "braking" power to the motor once the speed is exceeded until the target speed is reached; alternatively, a PID controller can be used, whose input is the speed error (current speed - safety threshold) and output is the required additional lift, which can more smoothly and accurately control the speed around the threshold. It is understood that the control method can be selected according to different requirements for smoothness and response speed, and is not limited here.

[0097] A key challenge during this step is coordinating the intervention with the main descent process. Excessive or abrupt braking can significantly increase energy consumption, contradicting the goal of energy conservation. To address this, the control system employs control logic with dead zones and hysteresis. For example, intervention is only initiated when the speed exceeds the safety threshold (e.g., 5 m / s) by a certain margin (e.g., exceeding 5.5 m / s), and intervention ceases only when the speed drops below the safety threshold to a lower value (e.g., 4.5 m / s). This hysteresis prevents the system from repeatedly starting and stopping the motor due to minor fluctuations near the critical point, ensuring the overall continuity and energy efficiency of the descent process, and intervening only when a genuine danger exists.

[0098] S209. When the instantaneous altitude value drops to the preset trigger altitude, control the rotor motor to operate, so as to adjust the UAV to contact the water surface in a preset horizontal attitude.

[0099] Refer to step S104, which will not be repeated here.

[0100] S210. When the drone is in an upside-down state on the water surface, control the drone to perform a flipping action.

[0101] Refer to step S105, which will not be repeated here.

[0102] In some embodiments, the control system further enhances the safety of the water landing process by acquiring information on the distribution of obstacles on the water surface and dynamic parameters of the water flow in the target water landing area, and constructing a three-dimensional obstacle map; calculating a safe flight corridor based on the three-dimensional obstacle map and the UAV's body size parameters; and optimizing the predetermined approach route based on the safe flight corridor to generate the optimal approach path, so that the UAV does not collide with obstacles on the water surface.

[0103] Among them, the distribution information of obstacles on the water surface can be obtained in real time by airborne lidar, millimeter-wave radar or visual sensors, or by pre-installed electronic nautical charts; the three-dimensional obstacle map is a data structure that includes the location, size and shape of obstacles; the safe flight corridor refers to the passage in three-dimensional space through which drones can fly safely after all obstacles and their safety buffer zones have been eliminated.

[0104] Specifically, before or during the planning of the predetermined approach route, the control system executes this obstacle avoidance process. It first integrates multi-source information to construct a 3D environment model centered on the target landing point in memory. Then, it "inflates" obstacles on the map according to the UAV's size (e.g., rotor diameter), creating no-fly zones. Next, within a safe flight corridor comprised of all spaces outside the no-fly zones, the system uses path planning algorithms (such as A or RRT) to find the optimal path from the approach point to the target point. This path becomes the new, obstacle-avoidance-optimized predetermined approach route, and subsequent flights will follow this path.

[0105] In some embodiments, path optimization in this step can be implemented in several ways: optionally, the space is rasterized, and Algorithm A is run on a 3D raster map to search for an optimal path; optionally, a sampling-based RRT (Rapid Expanding Random Tree Optimization) algorithm is used, which can efficiently find asymptotically optimal collision-free paths in complex high-dimensional spaces. It is understood that any algorithm capable of planning feasible paths in an environment that considers obstacles can be applied, and no limitation is made here.

[0106] A key challenge during this step is handling dynamic obstacles, such as moving boats, and the impact of water flow on the drone's final drift position. To address this, the obstacle map built by the control system is dynamically updated. It uses sensor data to predict the future trajectories of dynamic obstacles and reserves space for them to occupy over a future period when planning a safe flight corridor. For water flow, the system obtains hydrological forecast data or estimates the speed and direction of the current by observing the drone's drift after landing, and when selecting the final landing point, it allows for a certain lead time upstream to compensate for the drift distance from landing to stabilization.

[0107] It's important to note that during obstacle avoidance optimization, the acquired obstacle information is mapped onto a 3D grid map, with each grid cell marked as either "free" or "occupied." Then, an "dilation" process is performed, mathematically involving a morphological "dilation" operation on the occupied areas. Specifically, the radius of the UAV's enveloping sphere R (e.g., half the rotor diameter) and the safety margin d, i.e., R_safe = R + d, is used as the structuring element radius for the dilation operation. All grid cells marked as "occupied" are traversed, and all "free" grid cells within a distance of R_safe are also marked as "occupied." After this processing, all remaining "free" grid cells on the map constitute a "safe flight corridor" where the UAV's center of mass can fly. Finally, based on this dilated map, the A algorithm is used to plan the path. Algorithm A finds the optimal path by evaluating the cost function f(n) = g(n) + h(n) for each grid node n, where g(n) is the actual path cost (e.g., flight distance) from the starting point to node n, and h(n) is the estimated cost from node n to the destination (usually using Euclidean distance as a heuristic function). Starting from the starting point, the algorithm continuously expands by selecting the node with the smallest f(n) until the target point is reached, thus generating an optimal entry path that avoids all expansion obstacles while being as short as possible.

[0108] In some embodiments, the control system achieves precise point-to-point water landing. That is, the control system determines the free deceleration parameters of the UAV on the predetermined approach route based on environmental data and UAV flight test data. Based on the free deceleration parameters, and combined with the boundary condition that the horizontal velocity of the UAV at the target water landing point is zero, the system performs back-calculation to determine the preset horizontal approach speed.

[0109] Among them, environmental data includes air density, wind speed, etc.; flight test data refers to the speed decay data recorded from multiple unpowered gliding tests at different speeds; free deceleration parameters are usually one or more coefficients used to describe the relationship between air resistance and speed. For example, in the simplified model F_drag=k*v^2, k is the free deceleration parameter.

[0110] Specifically, this step is typically performed online before flight or during the mission. The control system analyzes a large amount of historical flight data, or performs a brief calibration maneuver over the mission area (such as gliding briefly after a sudden power outage from high-speed level flight), to fit the free deceleration model and parameters of the UAV under the current environment. Once these parameters are obtained, the system can establish a precise mathematical relationship from initial velocity and glide distance to terminal velocity. Based on this relationship, and setting a target (terminal horizontal velocity of zero, glide distance equal to the horizontal distance from the approach point to the target point), the system can precisely calculate the initial velocity required to achieve this target, i.e., the preset horizontal approach velocity, by solving the mathematical equation in reverse.

[0111] It's important to note that the training of the free deceleration model in this step is essentially a system parameter identification process. First, under various representative weather conditions (different wind speeds and air densities), the drone is controlled to perform multiple "powered gliding" tests: after accelerating to different speed levels, the motor power drops to a stopping threshold, and GPS ground speed, airspeed, position, and IMU attitude data are recorded for a period of time afterward. Once sufficient datasets are collected, parameter fitting is performed. For example, for the F_drag=c*v_air^2 model, its equation of motion is m*a=-c*v_air^2. Acceleration is obtained by numerically differentiating the collected velocity time series, or by directly using IMU data, a series of (v_air, a) data points can be obtained. The training criterion is to use the least squares method to find the optimal drag coefficient c, minimizing the mean squared error (MSE) between the model's predicted acceleration -c / m*v_air^2 and the actual measured acceleration a for all data points. The final coefficients c (or a set of coefficients for more complex models) are the trained model parameters.

[0112] The free deceleration model is a dynamic model describing the effect of air resistance on a drone during unpowered gliding. A commonly used and effective model is the quadratic drag model, where air resistance F_drag is proportional to the square of the airspeed v_air: F_drag = c * v_air^2, in the opposite direction to the airspeed. Here, c is a combined parameter (c = 0.5 * ρ * S * Cd) that integrates air density ρ, the drone's frontal area S, and the drag coefficient Cd. The model inputs are the drone's real-time airspeed v_air (measured by an airspeed tube or obtained by vector synthesis of ground speed and wind speed) and the drone's mass m. The model output is the deceleration due to air resistance a_drag = -(c / m) * v_air^2. Considering wind, the drone's equation of motion relative to the ground is dv_ground / dt = a_drag(v_air), where v_air = v_ground - v_wind. This model can predict the velocity change of a drone gliding from an arbitrary initial ground speed in a given wind field.

[0113] The core purpose of this model is to perform "backward calculations" to determine the preset horizontal approach velocity v0. In practice, the control system first determines the horizontal distance d_target between the target landing point and the starting point of the predetermined approach route. Then, it sets boundary conditions: at the end of gliding, the horizontal ground velocity v_f at the target point is 0. The control system establishes a mathematical relationship between the initial ground velocity v0, the gliding distance d, and the wind speed v_wind by integrating the above motion equation dv_ground / dt = -(c / m)*(v_ground - v_wind)^2. By solving this relationship (usually numerically), when d = d_target and v_f = 0, the required initial horizontal ground velocity v0 is obtained. This v0 is the "preset horizontal approach velocity" that the UAV must reach at the approach point to ensure that the energy is exhausted precisely upon reaching the target point, achieving a precise, targeted landing.

[0114] Taking the simplified model m*dv / dt=-k*v^2 under windless conditions as an example, to relate it to distance, we use the chain rule dv / dt=(dv / dx)*(dx / dt)=v*(dv / dx), the equation becomes m*v*(dv / dx)=-k*v^2, which simplifies to m*(dv / dx)=-k*v. This is a separable differential equation. After separating the variables, we get (m / k)*(1 / v)dv=-dx. Taking definite integrals over both sides, the velocity is integrated from the initial approach velocity v0 to the final velocity 0, and the distance is integrated from 0 to the target gliding distance d_target: ∫(v0,0)(m / k)*(1 / v)dv=-∫(0,d_target)dx. The integral on the left is (m / k)*[ln(v)]|(v0, 0), which is a divergent integral, indicating that the velocity cannot decay precisely to zero over a finite distance. Therefore, in practical applications, the final target velocity vf will be set to a very small positive value (e.g., 0.1 m / s), or the model will include a linear drag term. If the model is m*dv / dt=-k1*v^2-k2*v, then the back-calculation is usually performed in reverse using numerical integration (e.g., the Runge-Kutta method): starting from (x=d_target, v=vf), the inverse integration is performed in small negative time steps -dt, updating the velocity and position at each step, until the position x=0, at which point the velocity is the required initial velocity v0. For example, given m=5kg, k1=0.1, k2=0.2, d_target=100m, and vf=0.1m / s, the inverse numerical integration program will start from v=0.1m / s, gradually increase the speed, and accumulate the calculated small displacements until the total displacement reaches 100m. The speed at this point is the calculated preset horizontal approach speed.

[0115] A key issue that may arise during this step is the influence of wind, especially headwinds and tailwinds, which can alter the drone's gliding distance and thus affect the accuracy of deceleration parameters. To address this, wind speed must be taken into account when determining the free deceleration parameters. The control system first estimates the wind speed vector using onboard sensors. When calculating deceleration parameters, the drone's airspeed (speed relative to the air) is used, not its ground speed (speed relative to the ground). When calculating the preset horizontal approach velocity, the final target speed is the ground speed. For example, in a headwind of 5 m / s, to achieve a zero ground speed landing, the drone needs to maintain an airspeed of 5 m / s just before contact with the water. The entire calculation process is based on an "airspeed model," and finally, the required ground speed is converted based on the wind speed to accurately compensate for the wind's influence.

[0116] In some embodiments, after determining the approach speed, the control system will accurately plan an energy-saving path. That is, the control system will construct a parabolic trajectory of the predetermined approach route based on the preset horizontal approach speed and free deceleration parameters; determine the initial flight altitude of the UAV on the parabolic trajectory, so that the UAV reaches the target landing point when the horizontal speed is zero.

[0117] Constructing a parabolic trajectory refers to calculating the complete spatiotemporal path from the initial state to the target state based on a physical model, including the altitude and horizontal position at each moment; the initial flight altitude refers to the starting altitude at which the UAV begins to perform unpowered gliding.

[0118] Specifically, after determining the preset horizontal approach velocity in step S202, the control system executes this step to finally determine the complete descent trajectory. Using the known initial horizontal velocity, free deceleration parameters, and gravitational acceleration g, the system can accurately integrate the total time and total horizontal glide distance required for the UAV to glide from the start of gliding until its horizontal velocity drops to zero. Simultaneously, by integrating the vertical motion, the total descent height of the UAV during this time can be calculated. This total descent height, plus the height of the landing point (usually 0) and a certain safety height margin, determines the ideal initial flight altitude of the UAV when entering the gliding trajectory.

[0119] In some embodiments, trajectory construction in this step can be achieved in several ways: Optionally, the relationship between altitude, distance, and time can be directly obtained by analytically solving the equations of motion, and the parameters can be substituted; alternatively, when the model is too complex to be solved analytically, a numerical integration method (such as the Runge-Kutta method) can be used to simulate the flight process step by step from the initial state until the horizontal velocity is zero, thereby obtaining the complete trajectory and the required initial altitude. It is understood that the method of constructing the trajectory depends on the complexity of the physical model used, and no limitation is made here.

[0120] A key issue that may arise during this step is the discrepancy between the actual flight environment and the model. For example, air density varies with altitude, which can cause deviations between the pre-built fixed trajectory and the actual flight trajectory. To address this, the control system can employ online trajectory correction technology. During the UAV's gliding descent, the system continuously compares the actual position with the pre-built ideal trajectory. If a deviation occurs, the system can fine-tune the attitude (e.g., slightly tilting up or down) to change the UAV's lift-to-drag ratio, thereby fine-tuning the gliding trajectory and bringing it back towards the predetermined target point. This closed-loop trajectory tracking capability can compensate for model inaccuracies, further improving the final water landing accuracy.

[0121] In some embodiments, to ensure the stability of the UAV when operating on the water surface, the control system acquires real-time wind direction and wind speed intensity information of the target water landing area, calculates the influence coefficient of wind on the UAV's water surface attitude, determines wind compensation parameters based on the influence coefficient and the UAV's current water surface position, and adjusts the UAV's water surface operation parameters based on the wind compensation parameters to generate a water surface navigation trajectory, so that the UAV remains stable under the action of wind.

[0122] Among them, the influence coefficient is a model parameter that quantifies the thrust and torque generated by wind on the drone; the wind compensation parameter is the additional thrust or torque that needs to be applied to counteract the influence of wind; the water surface operation parameters include the motor speed and attitude in the water; and the water surface navigation trajectory refers to the planned path when water surface movement is required.

[0123] Specifically, this process is executed by the control system when the UAV completes its rollover maneuver (refer to step S210) and prepares to perform a task on the water surface (such as fixed-point sampling or mobile observation). The system first acquires real-time wind information via an onboard wind speed sensor or received meteorological data. Based on pre-calibrated hydrodynamic and aerodynamic models, it calculates the lateral thrust and rotational torque that the current wind will generate on the floating UAV. To maintain positional stability, the system calculates a compensating thrust of equal magnitude but opposite direction and converts it into the required rotational speed of each underwater propeller (i.e., rotor). If the UAV is to navigate along a certain trajectory, wind-induced drift is taken into account during path planning, generating a corrected route that ultimately reaches the target.

[0124] It should be noted that the pre-calibrated hydrodynamic and aerodynamic models used here have their parameters obtained through experimental identification. First, in a controlled pool environment, wind fields with varying speeds and directions are created using a wind tunnel or large fan. The drone is placed on the water surface, and its drift position, velocity, and attitude changes under wind influence are recorded using a motion capture system or high-precision GPS. Simultaneously, the underwater rotor is actively controlled to generate known thrust and torque, and the drone's response is recorded. These input-output data allow for the identification of key model parameters. For example, for the wind model F_wind=C_w*v_wind^2, the wind coefficient C_w can be fitted by measuring the drift force at different wind speeds. For the water drag model F_drag_water=C_d1*v+C_d2*v^2, the water drag coefficients C_d1 and C_d2 can be fitted by applying a known thrust and measuring the stabilized velocity. The training (identification) standard is to minimize the difference between the forces and torques predicted by the model and the forces and torques required to maintain equilibrium in the experiment.

[0125] This model is a six-degree-of-freedom model that integrates aerodynamics and hydrodynamics. Its core function is to describe the force balance of the UAV on the water surface. The model's inputs include: 1) Real-time wind speed and direction v_wind; 2) The drone's current velocity v_water and angular velocity ω in relation to the water; 3) The thrust T_i applied to each rotor by the control system.

[0126] The model consists of a set of force and moment equations: ΣF=F_wind+F_drag_water+F_thrust=m*a; ΣM=M_wind+M_drag_water+M_thrust=I*α.

[0127] Here, F_wind represents the force generated by the wind acting on the above-water portion of the UAV, F_drag_water represents the drag generated by the water acting on the underwater portion of the UAV, and F_thrust represents the total thrust generated by the rotor in the water. The M series represents the corresponding torques. Each force / torque is a function of its corresponding variable (such as wind speed and water speed) and the corresponding influence coefficient. The model outputs the total thrust F_thrust and total torque M_thrust required to achieve a specific motion state (such as maintaining position a=0, α=0) under the current environment; these two values ​​are called the "wind compensation parameters".

[0128] When operating on the water, this model is used for both feedforward and feedback control.

[0129] 1. Feedforward Compensation: The control system inputs the real-time measured wind speed v_wind into the model, which directly calculates the compensation thrust F_thrust_ff and torque M_thrust_ff required to counteract the wind force. These compensation values ​​are used as feedforward terms and directly added to the output of feedback controllers such as PID controllers. This proactively counteracts wind force before GPS detects position deviation, significantly improving the stability and accuracy of position holding.

[0130] 2. Feedback controller parameter tuning: The hydrodynamic parameters (such as damping coefficient) provided by the model can be used to design the parameters (Kp, Ki, Kd) of the PID controller for the water surface position more accurately, so that the controller response is more in line with the physical reality and avoids overshoot and oscillation.

[0131] For example, to maintain a fixed position, the system calculates the feedforward compensation thrust, and then the GPS position feedback PID controller calculates a corrected thrust. The two are superimposed, and the power distribution algorithm is used to calculate the specific rotational speed required for each rotor, thereby achieving stable resistance to wind and waves.

[0132] In some embodiments, the compensation control for this step can be implemented in several ways: Optionally, for a stationary position task, a surface position PID controller is used to convert the GPS position deviation into a compensation thrust command; alternatively, for a navigation task, a feedforward compensation term is added to the path tracking controller to directly cancel the drift force calculated based on the wind speed. It is understood that any algorithm capable of suppressing wind interference and achieving surface position or trajectory control can be applied, and no limitation is made here.

[0133] A key challenge during this implementation step is the complexity and time-varying dynamics of the UAV on the water surface. Influenced by waves and its own attitude, accurate modeling is difficult, leading to poor compensation results. To address this, the control system can employ adaptive or intelligent control methods. For example, the system can identify hydrodynamic parameters online in real time and continuously adjust the compensation model based on actual control performance. Alternatively, a fuzzy logic controller can be used, incorporating the rules experienced operators use to combat wind and waves, enabling the UAV to perform compensation as flexibly as a human. In this way, robust and efficient water surface stabilization control can be achieved even with an inaccurate model.

[0134] In this embodiment, an innovative water-landing method is adopted, which involves stopping the motor, gliding down along a parabolic trajectory using inertia, and only briefly turning on the drone to adjust its attitude before landing. This reduces the energy consumption of the drone during the water-landing process and effectively solves the problems of high energy consumption and limited endurance caused by the hovering vertical descent method in the prior art. As a result, it achieves a comprehensive effect of significantly improving the endurance and efficiency of drone water operations while ensuring the safety and accuracy of water landing.

[0135] The control system in the embodiments of this invention is described below from the perspective of hardware processing. Please refer to [link / reference needed]. Figure 3 This is a schematic diagram of the physical device structure of the control system in an embodiment of this application.

[0136] It should be noted that, Figure 3 The structure of the control system shown is merely an example and should not impose any limitation on the functionality and scope of use of the embodiments of the present invention.

[0137] like Figure 3 As shown, the control system includes a CPU 301, which can perform various appropriate actions and processes according to a program stored in ROM 302 or a program loaded into RAM 303 from storage section 308, such as executing the methods described in the above embodiments. RAM 303 also stores various programs and data required for system operation. The CPU 301, ROM 302, and RAM 303 are interconnected via bus 304. I / O interface 305 is also connected to bus 304.

[0138] The following components are connected to I / O interface 305: input section 306 including audio input devices, push-button switches, etc.; output section 307 including liquid crystal display (LCD) and audio output devices, indicator lights, etc.; storage section 308 including hard disks, etc.; and communication section 309 including network interface cards such as LAN (Local Area Network) cards, modems, etc. Communication section 309 performs communication processing via a network such as the Internet. Drive 310 is also connected to I / O interface 305 as needed. Removable media 311, such as disks, optical disks, magneto-optical disks, semiconductor memories, etc., are installed on drive 310 as needed so that computer programs read from them can be installed into storage section 308 as needed.

[0139] In particular, according to embodiments of the present invention, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of the present invention include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing computer programs for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via communication section 309, and / or installed from removable medium 311. When the computer program is executed by CPU 301, it performs the various functions defined in the present invention.

[0140] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of the present invention. Each block in a flowchart or block diagram may represent a module, program segment, or portion of code, which contains one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those shown in the drawings.

[0141] Specifically, the control system of this embodiment includes a processor and a memory. The memory stores a computer program, and when the computer program is executed by the processor, it implements the UAV waterborne operation control method provided in the above embodiment.

[0142] In another aspect, the present invention also provides a computer-readable storage medium, which may be included in the control system described in the above embodiments; or it may exist independently and not incorporated into the control system. The storage medium carries one or more computer programs that, when executed by a processor of the control system, cause the control system to implement the UAV waterborne operation control method provided in the above embodiments.

[0143] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit it. Although this application 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 of the technical features. Such 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 this application.

[0144] As used in the above embodiments, depending on the context, the term "when..." can be interpreted as meaning "if...", "after...", "in response to determining...", or "in response to detecting...". Similarly, depending on the context, the phrase "when determining..." or "if (the stated condition or event) is interpreted as meaning "if determining...", "in response to determining...", "when (the stated condition or event) is detected", or "in response to detecting (the stated condition or event)".

Claims

1. A method for controlling unmanned aerial vehicle (UAV) operations on water, characterized in that, Applied to a control system, the method includes: Control the drone to fly to a predetermined approach route to the target water landing area, and adjust the drone's horizontal speed to a preset horizontal approach speed; the predetermined approach route includes a parabolic trajectory; The operating power of all rotor motors is reduced to a preset shutdown threshold, causing the UAV to descend towards the water surface along the parabolic trajectory based on horizontal inertial velocity and gravity. The vertical distance between the drone and the water surface is monitored in real time, and the instantaneous altitude value is calculated. When the instantaneous altitude value drops to the preset trigger altitude, the rotor motor is controlled to operate in order to adjust the UAV to contact the water surface in a preset horizontal attitude; When the drone is in an upside-down state on the water surface, control the drone to perform a flipping action.

2. The method according to claim 1, characterized in that, The steps of controlling the UAV to fly to the predetermined approach route to the target water landing area and adjusting the UAV's horizontal speed to a preset horizontal approach speed specifically include: Determine the target water-landing point within the target water-landing area, and calculate the real-time azimuth angle from the current position to the target water-landing point; Calculate the direction adjustment amount and speed adjustment amount based on the real-time azimuth angle, the current speed of the UAV, and the current position; Based on the direction adjustment amount and the speed adjustment amount, a transitional route for the UAV to adjust its attitude and speed starting from the current position is planned and generated, and a predetermined approach route for the UAV to land at the target water landing point after parameter adjustment is generated. The flight parameters of the UAV are adjusted so that the UAV enters the predetermined approach route from the transition route at a preset horizontal approach speed.

3. The method according to claim 2, characterized in that, After determining the target landing point within the target landing area and calculating the real-time azimuth angle from the current position to the target landing point, the method further includes: Obtain information on the distribution of obstacles on the water surface and dynamic parameters of the water flow in the target water-contacting area, and construct a three-dimensional obstacle map; Calculate the safe flight corridor based on the three-dimensional obstacle map and the UAV's body size parameters; Based on the safe flight corridor, the predetermined approach route is optimized for obstacle avoidance to generate the optimal approach path, so that the UAV does not collide with the obstacles on the water surface.

4. The method according to claim 2, characterized in that, Before the step of calculating the direction adjustment and speed adjustment based on the real-time azimuth angle, the current speed of the UAV, and the current position, the method further includes: Based on environmental data and the flight test data of the UAV, determine the free deceleration parameters of the UAV on the predetermined approach route; Based on the free deceleration parameters, and combined with the boundary condition that the horizontal velocity of the UAV at the target landing point is zero, a back-calculation is performed to determine the preset horizontal approach velocity.

5. The method according to claim 4, characterized in that, After the step of determining the preset horizontal approach velocity by back-calculating based on the free deceleration parameters and the boundary condition that the horizontal velocity of the UAV at the target landing point is zero, the method further includes: Based on the preset horizontal approach speed and the free deceleration parameters, construct the parabolic trajectory of the predetermined approach route; Determine the initial flight altitude of the UAV on the parabolic trajectory, so that the UAV reaches the target landing point when its horizontal velocity is zero.

6. The method according to claim 1, characterized in that, After the step of real-time monitoring of the vertical distance between the drone and the water surface and calculating the instantaneous altitude value, the method further includes: Calculate the real-time vertical descent speed of the drone; When the vertical descent speed exceeds a preset safe descent speed threshold, the rotor motor is controlled to operate in order to reduce the real-time vertical descent speed of the UAV.

7. The method according to claim 1, characterized in that, After the step of controlling the drone to perform a flipping motion when the drone is in an inverted state on the water surface, the method further includes: Acquire real-time wind direction and wind speed intensity information of the target water landing area, and calculate the influence coefficient of wind force on the water surface attitude of the UAV; Based on the influence coefficient and the current water surface position of the UAV, the wind compensation parameters are determined; The water surface operation parameters of the UAV are adjusted based on the wind compensation parameters to generate a water surface navigation trajectory, so that the UAV remains stable under the action of wind.

8. A control system, characterized in that, The control system includes: one or more processors and a memory; the memory is coupled to the one or more processors, the memory is used to store computer program code, the computer program code including computer instructions, and the one or more processors call the computer instructions to cause the control system to perform the method as described in any one of claims 1-7.

9. A computer-readable storage medium comprising instructions, characterized in that, When the instruction is executed on the control system, it causes the control system to perform the method as described in any one of claims 1-7.

10. A computer program product, characterized in that, When the computer program product is run on the control system, the control system performs the method as described in any one of claims 1-7.

Images