Water-air multi-rotor sampling aircraft control method and system

Abstract

The invention provides a water-air multi-rotor sampling aircraft control method and system, and relates to the technical field of water-air amphibious aircraft control, and the method comprises the steps: obtaining sensor data of an aircraft in a water-air environment, constructing a state space model, and outputting a flight state; a sampling task path is generated through the task planning layer, a flight environment is judged and a control mode is switched through the medium adaptive layer, and an execution signal is generated through the execution control layer; the water pressure difference of the upper and lower surfaces is monitored, the water filling and draining state of the buoyancy compartment is adjusted through an underwater pressure difference balance control method, and the inclination angle and rotating speed of a propeller are adjusted through a propeller cooperative control algorithm to generate reverse flow thrust. And when the aircraft arrives at a sampling point, braking the propeller, adjusting the attitude of the mechanical arm until the sampling head is aligned with the target, executing the sampling action and recording the sampling data, so that high-precision positioning and attitude control can be performed on the aircraft, and seamless switching and cooperation of control logic during water-air cross-medium operation are ensured.

Description

Technical Field

[0001] The present invention relates to the technical field of water-air amphibious aircraft control, and in particular to a water-air multi-rotor sampling aircraft control method and system. Background Art

[0002] With the increasing integration of drone technology and environmental monitoring, existing amphibious aerial vehicles, while capable of surface takeoff and landing, have limited underwater capabilities, making it difficult to ensure stability and accuracy during underwater sampling. Current amphibious aerial vehicles primarily rely on buoyancy devices and underwater thrusters to achieve surface and underwater mobility. However, these vehicles are significantly affected by currents and suffer from instability during underwater missions. Due to the lack of precise underwater positioning and attitude control mechanisms, existing equipment suffers from low accuracy during underwater sampling and is prone to loss of control in complex underwater environments, increasing the difficulty and risk of sampling.

[0003] Therefore, it is necessary to provide a water-air multi-rotor sampling aircraft control method and system to solve the above technical problems. Summary of the Invention

[0004] In order to solve the above technical problems, the present invention provides a control method and system for a water-to-air multi-rotor sampling aircraft, which is used to solve the problems of low positioning accuracy and unstable attitude control of existing water-to-air amphibious aircraft during underwater sampling, and the inability to achieve seamless switching and coordination of water-to-air cross-media operations.

[0005] The present invention provides a method for controlling a water-air multi-rotor sampling aircraft, the method comprising: Acquire multi-source sensor data of a sampling aircraft in a water-air environment, perform spatiotemporal calibration on the multi-source sensor data using a Kalman filter fusion algorithm and construct an aircraft state space model, and output flight state data of the sampling aircraft, including real-time position, attitude angle, linear velocity vector, and angular velocity vector; Based on the aircraft state space model, the mission planning layer generates the optimal sampling mission path, the medium adaptation layer determines the current flight environment and automatically switches the flight control mode, and the execution control layer generates a flight execution signal; In an underwater environment, the water pressure difference between the upper and lower surfaces of the sampling aircraft is monitored, and the filling and drainage status of the buoyancy chamber is dynamically adjusted using an underwater pressure difference balance control method. At the same time, based on the water flow velocity vector and inertial measurement data, a thruster vector cooperative control algorithm is used to coordinately adjust the inclination angle and speed of the four vector thrusters to generate anti-water flow thrust; When the sampling aircraft reaches the preset sampling point, the vector thruster is braked, and the posture of the sampling robotic arm is adjusted through the visual servo control algorithm until the sampling head is aligned with the sampling target, triggering the solenoid valve to perform the sampling action and synchronously record the sampling data.

[0006] Preferably, the underwater depth value, the upper and lower surface water pressure values, the water flow velocity vector, and the inertial measurement data are collected respectively by an underwater depth sensor, a water pressure sensor, a Doppler velocimeter, and an inertial measurement unit, and the aerial position data, the aerial height value, and the visual image data are collected respectively by an aerial GPS, a barometric altimeter, and a visual positioning sensor, and the underwater depth value, the upper and lower surface water pressure values, the water flow velocity vector, the inertial measurement data, and the aerial position data, the aerial height value, and the visual image data are summarized to generate the multi-source sensor data.

[0007] Preferably, the calculation process of the Kalman filter fusion algorithm is as follows: Compute a priori state estimates and the prior state estimate covariance as follows: Then calculate the Kalman gain , posterior state estimation and the posterior state estimate covariance as follows: Where, Represents the prior state estimate at time k, including the real-time position, attitude angle, linear velocity vector and angular velocity vector of the sampled aircraft; Represents the state transfer matrix, which is used to reflect the transfer relationship of the aircraft state vector from time k-1 to time k, and has a mapping relationship with the water velocity vector and inertial measurement data; represents the transpose of the state transition matrix; represents the posterior state estimate at time k-1; Represents the control input matrix, which is used to characterize the influence of the control variable on the aircraft state vector; represents the control vector at time k; represents the prior error covariance at time k, which is used to reflect the uncertainty of the prior state estimation and is related to the sensor noise; represents the posterior error covariance at time k-1; represents the process noise covariance, which is used to characterize the uncertainty of the aircraft state space model; represents the Kalman gain matrix, which is used to balance the weight of the prior state estimate and the sensor observation vector; Represents the observation matrix, which is used to map the aircraft state vector to the sensor observation space; represents the transpose of the observation matrix; represents the observation noise covariance, i.e., the sensor measurement error; represents the sensor observation vector at time k, i.e., multi-source sensor data; represents the posterior state estimate at time k; I represents the identity matrix; represents the posterior error covariance at time k.

[0008] Preferably, when the task planning layer uses a fast exploration random tree algorithm to generate the optimal sampling task path, the corresponding optimization objective function is ,in, They represent the starting time and ending time of path planning respectively; v(k) represents the linear velocity vector of the sampling aircraft at time k; represents the Euclidean norm of the linear velocity vector v(k); a(k) represents the acceleration vector of the sampling aircraft at time k, that is, the differential calculation result of the linear velocity vector v(k) at time k; represents the Euclidean norm of the acceleration vector a(k); represents the angular velocity vector of the sampling aircraft at time k; represents the weight of control speed smoothness; Represents the angular velocity vector The Euclidean norm of ; represents the weight of controlling acceleration smoothness; represents the control attitude stability weight.

[0009] Preferably, judging the current flight environment through the medium adaptation layer and automatically switching the flight control mode specifically includes: When the underwater depth value l<0.5m and the aerial altitude value h>1m, the current flight environment is determined to be an aerial flight environment and the flight control mode is automatically switched to the aerial flight control mode; When the underwater depth value l satisfies , determining that the current flight environment is a surface flight environment and automatically switching to a surface flight control mode; When the underwater depth value l>2m, the current flight environment is determined to be an underwater flight environment and the underwater flight control mode is automatically switched; The aerial flight control mode is based on the GPS and the visual positioning sensor, and controls the position and attitude of the sampling aircraft by collaboratively adjusting the six-rotor speed; The surface flight control mode is to close part of the rotors, start the buoyancy tank to keep the aircraft floating, and control the sampling aircraft to move on the surface of the water through the two vector thrusters; The underwater flight control mode is to completely close the rotor, start the four vector thrusters, and adjust the sampling aircraft to a pressure difference balance control state.

[0010] Preferably, during the switching process of the flight control mode, a smooth transition algorithm is used to generate a mixed control signal and output it to the sampling aircraft. The calculation formula of the mixed control signal is as follows: Where U(k) represents the mixed control signal at time k; Indicates the control signal before mode switching; Indicates the control signal after mode switching; represents the switching weight factor, ; c represents the control switching rate; Indicates the switching start time.

[0011] Preferably, the underwater pressure difference balance control method uses a PID controller to adjust the opening of the filling and drainage pump of the buoyancy tank in real time and dynamically adjust the filling and drainage state of the buoyancy tank until the buoyancy change generated by the filling and drainage of the buoyancy tank is ,in, represents the water pressure difference deviation of the sampling aircraft at time k, that is, the difference between the water pressure difference between the upper and lower surfaces and the pressure difference in the equilibrium state; Indicates PID control parameters.

[0012] Preferably, the thruster vector cooperative control algorithm solves the inclination angle of the vector thruster by inverse kinematics and speed ,in, represents the water velocity vector; represents the inertial measurement data; m and n represent the inverse kinematics mapping function in the thruster vector cooperative control algorithm; And the inclination The adjustment range is , the speed The adjustment range is .

[0013] Preferably, the adjustment process of the visual servo control algorithm is as follows: Identify the sampled target through HOG feature extraction and SVM classifier; Based on the position of the sampling target, the Lucas-Kanade optical flow algorithm is used to determine the desired position of the sampling head; Calculate the joint control amount of the sampling manipulator based on the image Jacobian matrix ,in, represents the Jacobian pseudoinverse; represents the expected position coordinates of the sampling head; o represents the current position coordinates of the sampling head; Represents the control gain coefficient.

[0014] A water-air multi-rotor sampling aircraft control system, the control system comprising: a model building module for acquiring multi-source sensor data of a sampling aircraft in a water-air environment, performing spatiotemporal calibration on the multi-source sensor data using a Kalman filter fusion algorithm and constructing an aircraft state space model, and outputting flight state data of the sampling aircraft, including real-time position, attitude angle, linear velocity vector, and angular velocity vector; A multi-layer control module is configured to generate an optimal sampling mission path based on the aircraft state space model through a mission planning layer, determine the current flight environment and automatically switch the flight control mode through a medium adaptation layer, and generate a flight execution signal through an execution control layer; An underwater adjustment module is used to dynamically adjust the filling and drainage status of the buoyancy chamber by monitoring the water pressure difference between the upper and lower surfaces of the sampling aircraft in an underwater environment using an underwater pressure difference balance control method, and at the same time, based on the water velocity vector and inertial measurement data, using a thruster vector cooperative control algorithm to coordinately adjust the inclination angle and speed of the four vector thrusters to generate anti-water flow thrust; The sampling adjustment module is used to brake the vector thruster when the sampling aircraft reaches the preset sampling point, and adjust the posture of the sampling robotic arm through the visual servo control algorithm until the sampling head is aligned with the sampling target, triggering the solenoid valve to perform the sampling action and synchronously record the sampling data.

[0015] Compared with related technologies, the control method and system of a water-air multi-rotor sampling aircraft provided by the present invention has the following beneficial effects: The present invention acquires multi-source sensor data from a sampling aircraft in a water-air environment, uses a Kalman filter fusion algorithm to perform spatiotemporal calibration on the multi-source sensor data, and constructs an aircraft state-space model. The flight state data of the sampling aircraft, including real-time position, attitude angle, linear velocity vector, and angular velocity vector, is output. Based on the aircraft state-space model, the mission planning layer generates an optimal sampling mission path. The medium adaptation layer determines the current flight environment and automatically switches the flight control mode. The execution control layer generates a flight execution signal. In an underwater environment, the underwater pressure differential balance control method is used to dynamically adjust the filling and drainage status of the buoyancy chamber by monitoring the water pressure difference between the upper and lower surfaces of the sampling aircraft. Simultaneously, a thruster vector cooperative control algorithm is used to coordinately adjust the inclination angle and speed of the four vector thrusters based on the water velocity vector and inertial measurement data to generate anti-water flow thrust. When the sampling aircraft reaches a preset sampling point, the vector thrusters are braked, and the attitude of the sampling manipulator arm is adjusted using a visual servo control algorithm until the sampling head is aligned with the sampling target. The solenoid valve is triggered to execute the sampling action and the sampling data is simultaneously recorded. This allows for high-precision positioning and attitude control of the aircraft, ensuring seamless switching and coordination of its water-air cross-medium operations.

[0016] The present invention can improve underwater positioning accuracy and reduce attitude angle deviation through pressure difference balance control and thruster vector collaborative algorithm, and then can perform high-precision positioning and attitude control on the aircraft, ensuring that the aircraft maintains a stable attitude in complex underwater environments. The method of the present invention adopts an adaptive control algorithm, which can effectively offset the influence of water flow, and can still keep the aircraft stable in a strong water flow environment, reducing the risk of sampling operations. The present invention can significantly reduce the switching time of the water-air mode, improve the cross-media control efficiency, do not need to rely on complex machine learning algorithms, and reduce the demand for hardware computing resources. The method of the present invention can reduce the sampling position deviation, improve the sampling efficiency, ensure the accuracy of the synchronous collection of water quality parameters, meet the high-precision requirements of environmental monitoring, and improve the pertinence and accuracy of environmental monitoring through depth-attitude dual closed-loop control and sampling collaborative control. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 A flow chart of a method for controlling a water-air multi-rotor sampling aircraft provided by an embodiment of the present invention; Figure 2 A system block diagram of a water-air multi-rotor sampling aircraft control system provided by an embodiment of the present invention; Figure 3 A schematic diagram of the hardware structure of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0018] To make the objectives, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts shall fall within the scope of protection of the present invention.

[0019] like Figure 1 FIG. 1 is a flow chart of a method for controlling a water-air multi-rotor sampling aircraft according to an embodiment of the present invention. Figure 1The execution subject of the method shown may be a software and / or hardware device. The execution subject of the present application may include but is not limited to at least one of the following: user equipment, network equipment, etc. Among them, user equipment may include but is not limited to computers, smart phones, personal digital assistants (PDAs) and the electronic devices mentioned above. Network equipment may include but is not limited to a single network server, a server group consisting of multiple network servers, or a cloud based on cloud computing consisting of a large number of computers or network servers, wherein cloud computing is a type of distributed computing, a super virtual computer composed of a group of loosely coupled computers. This embodiment does not limit this. It includes steps S1 to S4, as follows: S1, acquiring multi-source sensor data of a sampling aircraft in a water-air environment, performing spatiotemporal calibration on the multi-source sensor data using a Kalman filter fusion algorithm and constructing an aircraft state space model, and outputting flight state data of the sampling aircraft, including real-time position, attitude angle, linear velocity vector, and angular velocity vector; Among them, the sampling aircraft in the present invention is an unmanned aerial vehicle with the ability to operate in water and air multi-media. It provides flight power through multiple sets of propellers and integrates sampling actuators. It can complete the sample collection task of the target point in the water and air environment, and has the composite function of cross-media movement and operation. Multi-source sensor data is heterogeneous data collected by various sensor devices deployed on the sampling aircraft, which is used to characterize the real-time state of the aircraft in a multi-media environment. The Kalman filter fusion algorithm is an optimal data processing method based on Bayesian estimation theory. It recursively calibrates the time series and spatial coordinates of multi-source sensor data, thereby suppressing noise interference, improving data consistency, and constructing an accurate characterization model of the aircraft's motion state. The aircraft state space model refers to a mathematical model constructed using the state variable method. It describes the motion state and dynamic characteristics of the aircraft in three-dimensional space through state quantities such as position vectors, attitude angle parameters, linear velocity vectors and angular velocity vectors, providing a theoretical basis for control decisions.

[0020] Flight status data is a collection of physical quantities that characterize the motion characteristics of an aircraft in space. It reflects its real-time motion and environmental interaction during flight. Real-time position refers to the aircraft's three-dimensional spatial coordinates, which are dynamically updated over time and accurately describe the aircraft's geometric position in space. Attitude angles refer to the aircraft's angular parameters, including: pitch angle, the angle between the aircraft's longitudinal axis and the horizontal plane, representing head-up or head-down attitude; roll angle, the tilt angle of the aircraft's lateral axis relative to the horizontal plane, representing left or right roll attitude; and yaw angle, the angle between the projection of the aircraft's longitudinal axis on the horizontal plane and a reference direction, representing turning attitude. The linear velocity vector represents the speed of the aircraft's center of mass in translational motion in space. It is a vector property consisting of magnitude and direction. The magnitude represents the displacement of the aircraft per unit time, while the direction is tangential to the trajectory. It is typically decomposed into components in a three-dimensional coordinate system and is used to describe the aircraft's translational motion characteristics. The angular velocity vector represents the physical quantity of the speed of the aircraft's rotational motion around its own center of mass. It is a vector attribute, and its magnitude indicates the speed of rotation, that is, the angular change per unit time. Its direction follows the right-hand rule and is along the direction of the rotation axis. It is usually decomposed into the rotation rate around the three axes of the aircraft body, namely the pitch axis, roll axis, and yaw axis, which are used to describe the rate of change of the aircraft's attitude.

[0021] In a water-air cross-medium environment, a multidimensional perception network can be constructed by deploying an inertial measurement unit (IMU), depth sensor, water pressure sensor, and visual positioning device on the sampler to acquire heterogeneous data representing the aircraft's motion state. A Kalman filter fusion algorithm based on Bayesian estimation theory can then be used to synchronize the time series and unify the spatial coordinates of sensor data with different sampling frequencies and spatial coordinate systems. This involves iterative computation of the state prediction equation and the observation update equation to suppress measurement noise and eliminate systematic bias, achieving spatiotemporal calibration of the data. Furthermore, a state-space model of the aircraft can be constructed based on the Newton-Euler dynamics equations. Motion parameters such as real-time position, attitude angle, linear velocity vector, and angular velocity vector can be incorporated into a unified mathematical framework to form a state vector describing the aircraft's dynamic characteristics, providing a precise representation of the state for subsequent control decisions.

[0022] S2, based on the aircraft state space model, generates an optimal sampling mission path through the mission planning layer, determines the current flight environment and automatically switches the flight control mode through the medium adaptation layer, and generates a flight execution signal through the execution control layer; It can be understood that the mission planning layer is the top-level module in the hierarchical control architecture. Based on the environmental map and sampling mission requirements, it uses a path optimization algorithm to generate the optimal flight trajectory that meets operational accuracy and efficiency. The medium adaptation layer is the environmental perception and mode decision-making layer in the hierarchical control architecture. It uses multi-source data fusion analysis to identify the current medium environment and automatically switch to the corresponding control strategy and dynamic model to ensure stable cross-medium motion. The flight control mode refers to a set of control strategies defined based on the aircraft's environment and mission requirements. It includes dynamic models, sensor configurations, and control algorithms optimized for different scenarios. For example, the airborne mode uses closed-loop position control based on the navigation satellite system, while the underwater mode switches to vector propulsion control based on water pressure and Doppler velocity measurement. The medium adaptation layer enables automatic mode switching, ensuring stable cross-medium motion and control accuracy. The execution control layer is the bottom-level execution layer in the hierarchical control architecture. It converts upper-level decisions into specific hardware control signals, including rotor speed adjustment, thruster pitch control, and robotic arm movement commands, to achieve precise aircraft movement and operation. Flight execution signals are hardware-driven instructions generated by the executive control layer, translating upper-layer decisions into concrete physical actions. These signals are the ultimate carrier of instructions for aircraft motion control and task execution. Their accuracy directly impacts the flight trajectory and sampling accuracy. Specific signal forms include pulse-width modulation (PWM), voltage control, and current control.

[0023] The control method of the present invention utilizes a three-level hierarchical control system to decompose and execute sampling tasks. Specifically, the task planning layer utilizes a heuristic search algorithm to generate an optimal sampling path based on an environmental map and sampling requirements. It then optimizes the initial trajectory using a path smoothing algorithm by comprehensively evaluating the distribution of environmental obstacles, energy consumption thresholds, and sampling point priorities, ensuring that the trajectory meets the vehicle's dynamic constraints. The medium adaptation layer constructs a multi-sensor fusion environmental recognition model. By analyzing air pressure gradients, water pressure change rates, and visual scene texture features, it uses a pattern recognition algorithm to determine the current airborne, surface, or underwater environment. It then automatically switches to the corresponding control strategy based on the environmental attributes. For example, in the air mode, it uses position closed-loop control based on the global navigation satellite system, while in the underwater mode, it switches to vector propulsion control based on water pressure feedback. A gain scheduling mechanism is then used to achieve smooth transitions in control parameters. The execution control layer converts high-level control instructions into hardware-executable drive signals, adjusts the speed of the multi-rotor motor using pulse width modulation technology, and controls the pitch angle of the vector propulsion using a servo system. Combined with feedforward compensation and feedback correction algorithms, it suppresses system nonlinearities and external disturbances, ensuring the precise execution of control instructions.

[0024] S3, in an underwater environment, by monitoring the water pressure difference between the upper and lower surfaces of the sampling aircraft, dynamically adjusting the filling and draining states of the buoyancy chamber using an underwater pressure difference balance control method, and simultaneously using a thruster vector cooperative control algorithm to coordinately adjust the inclination angle and speed of the four vector thrusters based on the water velocity vector and inertial measurement data to generate anti-water flow thrust; It should be noted that the underwater pressure differential balance control method dynamically balances the vertical forces acting on the water by monitoring the water pressure differential between the upper and lower surfaces of the vehicle in real time, adjusting the filling and draining status of the buoyancy chamber to offset water flow disturbances and maintain the vehicle's depth stability. The thruster vector coordinated control algorithm is a control method based on fluid dynamics theory. Based on the acquired water velocity vector and inertial measurement data, it coordinates the inclination angle and speed of multiple sets of vector thrusters to generate a thrust vector opposite the direction of the water flow, compensating for environmental interference forces and achieving underwater hovering and trajectory tracking. The buoyancy chamber is a sealed cavity structure installed on the sampling vehicle. By filling or draining water or gas, the overall buoyancy of the vehicle is adjusted to achieve underwater depth control and vertical balance. Its operating principle is: when the vehicle needs to dive, it is drained to reduce weight, and when it needs to surface, it is filled with water to increase weight. This closed-loop control is combined with feedback from water pressure sensors to offset vertical water flow disturbances.

[0025] The water velocity vector is a physical quantity that characterizes the flow characteristics of water. It is a vector property that includes both velocity magnitude and direction. This vector is typically measured by a Doppler velocimeter (DVL) using the acoustic Doppler effect or calculated based on fluid dynamics models. It is used to describe the dynamic effects of water flow on an aircraft. In coordinated thruster vector control, this vector serves as the basis for disturbance compensation, generating equal and opposite thrust vectors to counteract the effects of water flow and maintain the aircraft's positional stability. Inertial measurement data, collected by an inertial measurement unit (IMU), represents the aircraft's motion parameters. These data include triaxial acceleration, which characterizes linear acceleration, and triaxial angular velocity, which characterizes rotational velocity. This data is measured in real time by sensors such as gyroscopes and accelerometers. After integration using algorithms such as the Kalman filter, it is used to calculate the aircraft's attitude angle, trajectory, and dynamic characteristics. It serves as a key input to the aircraft's state-space model and supports coordinated control and attitude correction of underwater thrusters. A vector thruster is a propulsion device with directional control capabilities. By varying the direction and magnitude of thrust, it provides three-dimensional vector power to an aircraft. Typically consisting of a propeller, servo actuators, and a drive mechanism, it can independently adjust the thrust components in pitch, yaw, and roll, enabling underwater six-degree-of-freedom motion control. In the control method presented in this paper, four vector thrusters collaboratively adjust inclination angle and speed to generate countercurrent thrust and attitude control torque, serving as the core actuator for high-precision underwater positioning.

[0026] In complex underwater flow environments, a dual closed-loop control mechanism can be used to achieve stable hovering and trajectory tracking of the aircraft. Specifically, the pressure differential balance control mechanism monitors the water pressure gradient in real time through pressure sensors deployed on the upper and lower surfaces of the aircraft. Based on the principles of fluid statics, it calculates the vertical force deviation. Then, it uses a fuzzy PID control algorithm to dynamically adjust the filling and drainage rate of the buoyancy chamber. By changing the overall buoyancy of the aircraft, it offsets the vertical disturbance caused by the water flow. Finally, it introduces the anti-disturbance control technology to estimate and compensate for unknown water flow disturbances in real time, maintain the neutral buoyancy state of the aircraft, and optimize the depth control accuracy. The vector propulsion cooperative control mechanism can establish an underwater propulsion dynamics model based on the momentum theorem, obtain the water flow velocity vector through the Doppler velocimeter, and combine the angular velocity data calculated by the inertial measurement unit to adopt the thruster vector cooperative control algorithm to cooperatively adjust multiple groups of vector thrusters. Then, the inclination angle and speed of each thruster can be finely controlled to generate the thrust vector and attitude control torque opposite to the direction of the water flow, compensate for environmental disturbances and realize six-degree-of-freedom motion control. Finally, the sliding mode observer is introduced to estimate the water flow disturbance and model uncertainty to enhance the robustness of the control system in complex flow fields.

[0027] S4, when the sampling aircraft reaches the preset sampling point, the vector thruster is braked, and the posture of the sampling manipulator is adjusted through the visual servo control algorithm until the sampling head is aligned with the sampling target, the solenoid valve is triggered to perform the sampling action and the sampling data is recorded synchronously.

[0028] Preset sampling points refer to pre-set sampling locations. The visual servo control algorithm is a closed-loop control strategy based on machine vision. It uses image sensors to capture the visual features of the sampling target, calculates the positional deviation of the sampling manipulator, and drives the manipulator's joints to adjust the sampling head's posture to achieve precise alignment with the target point, meeting the spatial accuracy requirements of the sampling operation. The sampling manipulator is a multi-degree-of-freedom mechanical actuator integrated into the aircraft. Joint actuation enables spatial position and posture adjustment of the end-cap sampling head. Its waterproof and sealed structure enables it to grasp and collect target samples underwater. The sampling head is a specialized tool integrated into the end of the manipulator, capable of target recognition, sample collection, and packaging. The sampling target refers to the physical entity or environmental parameter to be collected, including but not limited to: water quality parameters such as dissolved oxygen, pH, turbidity, and microbial concentration; sediment samples such as bottom particles, pollutant content, and sediment layer structure; and biological samples such as plankton, benthic organisms, and microbial communities. A solenoid valve is an electrically controlled fluid control element. The magnetic force generated by energizing an electromagnetic coil drives the valve core to open and close the sampling channel. It is used to automate operations such as water sample collection and storage. The sampling process includes the following: During the positioning phase, a visual servo control algorithm drives the robotic arm to align the sampling head with the target point; during the collection phase, the solenoid valve triggers the sampling channel, activating a water pump or drilling device to retrieve the sample; during the packaging phase, the sample is transferred to a sealed chamber and simultaneously injected with a protective agent or cryopreserved; during the reset phase, the robotic arm returns to its initial position and initiates a self-cleaning procedure to prevent cross-contamination. Each stage is controlled by sequential logic and force feedback to ensure operational accuracy and sample integrity. Sampling data refers to the multidimensional information recorded during the sampling process, including: spatial data (i.e., latitude, longitude, depth, and altitude of the sampling point); environmental parameters (i.e., temperature, salinity, pressure, and water velocity); sample attributes (i.e., volume, weight, color, and spectral characteristics); and a timestamp (i.e., the sampling instant, accurate to the millisecond level, for temporal and spatial data correlation).

[0029] After the sampling aircraft reaches the preset sampling point, closed-loop control of the sampling operation can be achieved through visual guidance. Specifically, a high-definition camera installed at the end of the sampling robotic arm can be used to obtain environmental images, and a feature point extraction algorithm combined with a deep learning semantic segmentation model can be used to achieve robust recognition of the sampling target. The three-dimensional posture is solved through the perspective projection transformation algorithm to determine the spatial relative position relationship between the sampling head and the target. Then, based on the robot kinematic model, the posture deviation of the visual feedback can be converted into the control quantity of the robotic arm joint. Finally, an adaptive impedance control algorithm can be used to sense the contact force in real time during the process of adjusting the robotic arm posture to avoid hard collisions between the sampling head and the target. Through Cartesian space trajectory planning, the end effector of the robotic arm can be controlled to move along the optimal path until the axis of the sampling head is accurately aligned with the target point.

[0030] Furthermore, a sequential logic controller can trigger the solenoid valve assembly to open the sampling channel and collect samples. Sampling modes are automatically switched based on the target attributes, such as water, sediment, or biological samples. For example, when sampling water, a micro-pump is activated to draw the sample into the storage chamber, while when sampling sediment, a drilling mechanism is driven for stratified sampling. While sampling, the spatial coordinates, environmental parameters, and timestamps of the sampling points are simultaneously recorded. Data encryption technology ensures the integrity and traceability of the sampled data.

[0031] The control method of the present invention completes algorithm verification through a hardware-in-the-loop simulation platform: a closed-loop test system including a multi-sensor data simulator, an aircraft dynamics model and an actuator is constructed, and simulation experiments are carried out under different water flow velocities and water transparency conditions. The results show that the underwater positioning accuracy can reach the centimeter level, the sampling success rate exceeds 90%, and it can effectively suppress water flow disturbances within a certain flow rate range. Actual water test data show that this control method realizes autonomous sampling operations in cross-media environments, providing a highly robust technical solution for fields such as water environment monitoring and ecological surveys. Its hierarchical control architecture and adaptive compensation strategy significantly improve the operational adaptability and mission execution efficiency of the aircraft in complex environments.

[0032] The underwater depth value, the upper and lower surface water pressure values, the water flow velocity vector, and the inertial measurement data are collected respectively through the underwater depth sensor, water pressure sensor, Doppler velocimeter, and inertial measurement unit, and the aerial position data, the aerial height value, and the visual image data are collected respectively through the aerial GPS, barometric altimeter, and visual positioning sensor. The underwater depth value, the upper and lower surface water pressure values, the water flow velocity vector, the inertial measurement data, the aerial position data, the aerial height value, and the visual image data are summarized to generate the multi-source sensor data.

[0033] In practical applications, a cross-media environmental perception network is constructed by deploying multiple sensor types on sampling aircraft. In underwater operations, depth sensors are used to obtain vertical water depth parameters, water pressure sensors monitor the pressure differential between the upper and lower surfaces of the aircraft, Doppler velocimeters are used to calculate water velocity vectors based on the acoustic Doppler effect, and an inertial measurement unit (IMU) collects motion state data such as triaxial acceleration and angular velocity. During the aerial phase, navigation satellite system receivers acquire three-dimensional positioning data, a barometric altimeter measures relative altitude, and a visual positioning sensor captures environmental texture feature images. These heterogeneous sensory information, including underwater depth parameters, water pressure differential data, water velocity vectors, and inertial measurement data, as well as aerial position coordinates, altitude information, and visual image data, are synchronized and format-standardized to form a multi-source sensor dataset, providing multidimensional information support for subsequent cross-media motion state calculation and control decision-making. This multi-source data fusion mechanism, through spatiotemporal registration and noise suppression, ensures the consistency and reliability of sensor data from different media and types, laying the foundation for precise control and intelligent aircraft operations.

[0034] The calculation process of the Kalman filter fusion algorithm is as follows: Compute a priori state estimates and the prior state estimate covariance as follows: Then calculate the Kalman gain , posterior state estimation and the posterior state estimate covariance as follows: Where, Represents the prior state estimate at time k, including the real-time position, attitude angle, linear velocity vector and angular velocity vector of the sampled aircraft; Represents the state transfer matrix, which is used to reflect the transfer relationship of the aircraft state vector from time k-1 to time k, and has a mapping relationship with the water velocity vector and inertial measurement data; represents the transpose of the state transition matrix; represents the posterior state estimate at time k-1; Represents the control input matrix, which is used to characterize the influence of the control variable on the aircraft state vector; represents the control vector at time k; represents the prior error covariance at time k, which is used to reflect the uncertainty of the prior state estimation and is related to the sensor noise; represents the posterior error covariance at time k-1; represents the process noise covariance, which is used to characterize the uncertainty of the aircraft state space model; represents the Kalman gain matrix, which is used to balance the weight of the prior state estimate and the sensor observation vector; Represents the observation matrix, which is used to map the aircraft state vector to the sensor observation space; represents the transpose of the observation matrix; represents the observation noise covariance, i.e., the sensor measurement error; represents the sensor observation vector at time k, i.e., multi-source sensor data; represents the posterior state estimate at time k; I represents the identity matrix; represents the posterior error covariance at time k.

[0035] It can be understood that the Kalman filter fusion algorithm achieves optimal estimation of multi-source sensor data through recursive calculation. Specifically, based on the posterior state estimate at time k-1 and the control input at time k, a prior state estimate at time k is constructed using the state transition matrix. This estimate includes motion parameters such as the aircraft's real-time position, attitude angle, linear velocity vector, and angular velocity vector. The prior error covariance is also calculated simultaneously. This parameter represents the uncertainty of the prior state estimate and is related to the process noise covariance and the posterior error covariance at time k-1. The process noise covariance reflects the uncertainty of the aircraft's state space model. The Kalman gain matrix is ​​then used to balance the weights of the prior state estimate and sensor observation data. The calculation of this gain matrix involves the prior error covariance, the observation matrix and its transpose, and the observation noise covariance. The observation matrix maps the aircraft state vector to the sensor observation space, and the observation noise covariance represents the sensor measurement error. Based on this gain matrix, the deviation between the sensor observation vector and the prior state estimate is weighted and corrected to obtain the posterior state estimate at time k, achieving an optimized update of the aircraft's motion state. Finally, the posterior error covariance is updated by multiplying the identity matrix with the gain matrix and the observation matrix, providing initial parameters for the recursive calculation at the next moment. This algorithm, through iterative state prediction and observation update operations, continuously suppresses sensor noise and eliminates system bias, ensuring the spatiotemporal consistency of multi-source data and the accuracy of state estimation.

[0036] When the task planning layer uses the fast exploration random tree algorithm to generate the optimal sampling task path, the corresponding optimization objective function is: ,in, They represent the starting time and ending time of path planning respectively; v(k) represents the linear velocity vector of the sampling aircraft at time k; represents the Euclidean norm of the linear velocity vector v(k); a(k) represents the acceleration vector of the sampling aircraft at time k, that is, the differential calculation result of the linear velocity vector v(k) at time k; represents the Euclidean norm of the acceleration vector a(k); represents the angular velocity vector of the sampling aircraft at time k; represents the weight of control speed smoothness; Represents the angular velocity vector The Euclidean norm of ; represents the weight of controlling acceleration smoothness; represents the control attitude stability weight.

[0037] In practical applications, when the mission planning layer uses the rapid exploration random tree algorithm to generate the optimal sampling path, it constructs an optimization target system centered on motion smoothness and attitude stability. The rapid exploration random tree algorithm achieves path optimization by randomly sampling and expanding the search tree in the state space, combined with an objective function containing multi-dimensional motion constraints. Specifically, the optimization objective function models the dynamic characteristics of the sampled aircraft and performs weighted integral operations on the linear velocity vector modulus, acceleration vector modulus, and angular velocity vector modulus within the time interval. Among them, the linear velocity vector modulus reflects the rate of translational motion of the aircraft, the acceleration vector, as the rate of change of the linear velocity vector with respect to time, characterizes the degree of fluctuation of the motion speed, and the angular velocity vector modulus reflects the speed of change of the aircraft's attitude.

[0038] It should be noted that the weight coefficient is used to balance the optimization priorities of different motion indicators. The weight parameter for controlling speed smoothness is used to reduce the speed fluctuation of the aircraft's translational motion and improve the smoothness of the path; the weight parameter for controlling acceleration smoothness is used to suppress sudden changes in acceleration and deceleration during motion, reducing energy consumption and mechanical stress; the weight parameter for controlling attitude stability focuses on constraining the aircraft's rotational motion amplitude to ensure attitude accuracy during sampling operations. Through this optimization mechanism, the rapid exploration random tree algorithm can generate an optimal sampling path that meets dynamic constraints while avoiding environmental obstacles, achieving coordinated optimization of sampling efficiency and motion stability. This path planning method effectively solves the problem of path generation with multiple constraints in complex environments through the combination of random sampling and dynamic optimization, providing theoretical support for the autonomous operation of water-air multi-rotor sampling aircraft.

[0039] The method of judging the current flight environment and automatically switching the flight control mode through the medium adaptation layer specifically includes: When the underwater depth value l<0.5m and the aerial altitude value h>1m, the current flight environment is determined to be an aerial flight environment and the flight control mode is automatically switched to the aerial flight control mode; When the underwater depth value l satisfies , determining that the current flight environment is a surface flight environment and automatically switching to a surface flight control mode; When the underwater depth value l>2m, the current flight environment is determined to be an underwater flight environment and the underwater flight control mode is automatically switched; The aerial flight control mode is based on the GPS and the visual positioning sensor, and controls the position and attitude of the sampling aircraft by collaboratively adjusting the six-rotor speed; The surface flight control mode is to close part of the rotors, start the buoyancy tank to keep the aircraft floating, and control the sampling aircraft to move on the surface of the water through the two vector thrusters; The underwater flight control mode is to completely close the rotor, start the four vector thrusters, and adjust the sampling aircraft to a pressure difference balance control state.

[0040] The medium adaptation layer achieves accurate classification of cross-water and air environments by constructing an environmental recognition model that fuses multi-sensor data. This recognition process is based on the joint judgment of underwater depth parameters and aerial altitude parameters: when the underwater depth parameter is less than 0.5m and the aerial relative altitude parameter exceeds 1m, it is determined to be an aerial flight environment; if the underwater depth parameter is within the range of 0-2m, it is identified as a surface flight environment; when the underwater depth parameter exceeds 2m, it is determined to be an underwater flight environment. This hierarchical judgment mechanism can dynamically perceive the medium environment in which the aircraft is located through logical operations on multi-source data, providing a decision-making basis for adaptive switching of control modes.

[0041] After determining that the environment is aerial, the system automatically activates a fusion positioning mechanism based on the navigation satellite system and visual positioning sensors. Satellite positioning data is used to provide a global coordinate reference, combined with local feature matching from the visual positioning sensors to establish a closed-loop control feedback link. In this aerial flight control mode, the aircraft's position and attitude are controlled through coordinated speed regulation of the six-rotor power system. Based on a multi-rotor dynamics model, position deviations and attitude angle errors are converted into speed adjustments for each rotor. By varying the magnitude and distribution of rotor lift, the aircraft's translational and rotational motion in three-dimensional space is controlled. This fully utilizes the low drag characteristics of the aerial environment, enabling agile maneuvers through speed differential adjustment, while ensuring positioning accuracy that meets the sampling mission requirements.

[0042] When the aircraft is in a water environment, it actively shuts down some of its rotors to reduce water resistance, and activates the constant pressure regulation mechanism of the buoyancy tank to maintain its floating state. At this time, the buoyancy tank automatically adjusts the filling and discharge status of the internal fluid medium by monitoring the changes in water pressure in real time, ensuring that the aircraft maintains neutral buoyancy on the water surface. Surface movement control relies on the coordinated work of two vector thrusters. By adjusting the thrust direction and magnitude of the thrusters, forward and backward movement and steering operations in the horizontal plane are achieved. This control mode takes into account both the stability of floating on the water and the flexibility of movement, solves the problem of dynamic discontinuity when the aircraft transitions at the water-air interface, and provides a transitional control solution for cross-media operations.

[0043] After entering the underwater environment, the rotor propulsion system is completely shut down to avoid water resistance losses. The four sets of vector thrusters are activated to form a three-dimensional power system. Then, pressure sensors deployed on the upper and lower surfaces of the aircraft collect water pressure differences in real time. Based on the principles of fluid statics, the vertical force deviation is calculated, and the filling and drainage rates of the buoyancy chamber are adjusted to achieve a neutral buoyancy state underwater. At the same time, the four sets of vector thrusters coordinately adjust the inclination angle and speed to generate a thrust vector opposite to the direction of the water flow, compensating for environmental water flow disturbances. This resolves the contradiction between water flow interference and positioning accuracy in underwater environments and ensures the stability of sampling operations.

[0044] During the switching process of the flight control mode, a smooth transition algorithm is used to generate a mixed control signal and output it to the sampling aircraft. The calculation formula of the mixed control signal is as follows: Where U(k) represents the mixed control signal at time k; Indicates the control signal before mode switching; Indicates the control signal after mode switching; represents the switching weight factor, ; c represents the control switching rate; Indicates the switching start time.

[0045] During the flight control mode switching process, a smooth transition algorithm can be used to generate a mixed control signal and output it to the sampling aircraft to ensure the smoothness and continuity of the control mode switching. This algorithm constructs a switching weight factor to perform a weighted fusion of the control signals before and after the mode switch, achieving a smooth transition of the control parameters. Specifically, the mixed control signal is synthesized by dynamically weighting the control signal before and after the switch according to time. The switching weight factor takes the form of a logical function with a value range between 0 and 1, achieving a gradual transition from the old control signal to the new control signal. The switching rate parameter is used to adjust the speed of the transition process. A larger switching rate will make the transition process complete faster, while a smaller switching rate will make the transition process smoother. The switching start time defines the starting time point of the transition process.

[0046] In this way, while maintaining the operational stability of the aircraft, seamless switching between different control modes can be achieved, effectively avoiding flight attitude oscillation or instability caused by sudden changes in control parameters, and ensuring the safety and reliability of the sampling aircraft throughout the entire operation process.

[0047] The underwater pressure difference balance control method uses a PID controller to adjust the opening of the filling and drainage pump of the buoyancy tank in real time and dynamically adjust the filling and drainage state of the buoyancy tank until the buoyancy change caused by the filling and drainage of the buoyancy tank is ,in, represents the water pressure difference deviation of the sampling aircraft at time k, that is, the difference between the water pressure difference between the upper and lower surfaces and the pressure difference in the equilibrium state; Indicates PID control parameters.

[0048] It should be noted that the underwater pressure difference balance control method uses a PID (proportional integral differential) controller to calculate the required buoyancy change based on the deviation between the water pressure difference between the upper and lower surfaces of the aircraft and the pressure difference in the equilibrium state, thereby achieving precise control of the aircraft's underwater buoyancy. This allows the opening of the buoyancy tank's filling and drainage pumps to be adjusted in real time to dynamically adjust the filling and drainage status of the buoyancy tank.

[0049] Specifically, when the sampling aircraft is underwater, water pressure sensors deployed on its upper and lower surfaces monitor the water pressure differential in real time. This pressure differential is then compared with the preset equilibrium pressure differential to obtain a water pressure differential deviation. The PID controller uses this deviation as an input signal and, through a linear combination of proportional, integral, and differential operations, generates a control variable to adjust the opening of the filling and drainage pumps. The proportional term is proportional to the current deviation and can quickly respond to changes in the water pressure differential. The integral term accumulates the deviation to eliminate static errors. The differential term reflects the rate of change of the deviation, allowing for prediction of the deviation's changing trend and proactive adjustments.

[0050] In this way, the PID controller can dynamically adjust the filling and draining states of the buoyancy chamber based on real-time changes in the water pressure differential deviation, ensuring that the buoyancy changes generated by the filling and draining of the buoyancy chamber precisely match the adjustment requirements of the water pressure differential deviation. Ultimately, this achieves pressure differential balance for the aircraft underwater, ensuring that it maintains a stable suspension state in complex water flow environments and providing a reliable foundation for underwater sampling operations. This underwater pressure differential balance control method fully utilizes the simple structure and strong robustness of the PID control algorithm, can effectively cope with the uncertainty and interference factors of the underwater environment, and improves the stability and reliability of the aircraft's underwater operations.

[0051] The thruster vector cooperative control algorithm solves the inclination angle of the vector thruster through inverse kinematics and speed ,in, represents the water velocity vector; represents the inertial measurement data; m and n represent the inverse kinematics mapping function in the thruster vector cooperative control algorithm; And the inclination The adjustment range is , the speed The adjustment range is .

[0052] It is understandable that the thruster vector coordinated control algorithm uses inverse kinematics theory to solve the inclination angle and speed parameters of the vector thruster to achieve dynamic compensation for water flow disturbances. The algorithm uses the water flow velocity vector and inertial measurement data as input and establishes a nonlinear mapping relationship between the environmental dynamic parameters and the thruster control quantity through an inverse kinematic mapping function. The water flow velocity vector represents the dynamic effect of the ambient water flow, while the inertial measurement data reflects the motion state of the aircraft itself. Together, the two form the core basis for thruster control decisions.

[0053] During the specific solution process, the algorithm, based on the fluid dynamics equations and the aircraft kinematic model, substitutes the water velocity vector and inertial measurement data into the inverse kinematics mapping function to calculate the required inclination angle and speed of each vector thruster. The adjustment range of the inclination parameter is limited to the range of -45°C to 45°C. This range ensures that the thruster can generate sufficient vector thrust components while avoiding the loss of fluid dynamic efficiency caused by excessive inclination angles. The speed parameter adjustment range covers 0 to 10,000 rpm, meeting the full range of control requirements from static hovering to high-speed propulsion.

[0054] Through this inverse kinematics solution mechanism, the thruster vector cooperative control algorithm can dynamically generate the optimal combination of thruster control parameters based on real-time environmental disturbances and the aircraft's motion state. By collaboratively adjusting the inclination angle and speed, each vector thruster can synthesize a thrust vector in three-dimensional space that is opposite to the water flow disturbance, while generating the necessary attitude control torque to achieve six-degree-of-freedom stable control of the aircraft in an underwater environment. This algorithm fully utilizes the omnidirectional thrust adjustment capability of the vector thruster, combined with the precise solution of inverse kinematics, effectively improving the aircraft's anti-interference ability and motion control accuracy in complex water flow environments, providing reliable power support for underwater sampling operations.

[0055] The adjustment process of the visual servo control algorithm is as follows: Identify the sampled target through HOG feature extraction and SVM classifier; Based on the position of the sampling target, the Lucas-Kanade optical flow algorithm is used to determine the desired position of the sampling head; Calculate the joint control amount of the sampling manipulator based on the image Jacobian matrix ,in, represents the Jacobian pseudoinverse; represents the expected position coordinates of the sampling head; o represents the current position coordinates of the sampling head; Represents the control gain coefficient.

[0056] In practical applications, the visual servo control algorithm achieves precise alignment of the sampling manipulator with the target through a multi-stage image processing and motion control strategy. First, the algorithm uses oriented gradient histogram feature extraction technology to convert the visual features of the sampled target into a high-dimensional feature vector. This is then used by a support vector machine classifier for pattern recognition, enabling accurate detection and classification of the target. This process extracts the gradient directional distribution features of local image regions and combines them with a machine learning classifier to construct a robust description of the target's appearance, ensuring recognition accuracy under varying lighting conditions and viewing angles.

[0057] Based on the identified sampling target position, the algorithm then uses optical flow estimation to calculate the target's motion trajectory within the image sequence. By analyzing pixel brightness changes between adjacent frames and incorporating spatiotemporal gradient constraints, the algorithm builds a motion model of the target point in the image plane, thereby predicting the desired position of the sampling head. This motion estimation method, based on feature point tracking, can capture the target's dynamic changes in real time, providing precise visual feedback for the robotic arm's motion planning.

[0058] During the joint control quantity calculation stage, the algorithm constructs a mapping relationship between visual error and joint motion based on the image Jacobian matrix. By calculating the coordinate deviation between the current position and the desired position, combined with the pseudo-inverse operation of the Jacobian matrix, the error in the image space is converted into the control quantity in the joint space of the manipulator. The control gain coefficient is used to adjust the error convergence speed to ensure that the control system responds quickly within a stable range. This control strategy based on inverse kinematics realizes closed-loop control from visual perception to motion execution by establishing a nonlinear mapping between image space and joint space, effectively improving the accuracy and robustness of the manipulator's operation. The entire algorithm ensures that the sampling head can accurately align with the target in complex environments through the coordinated work of feature extraction, motion estimation and error compensation, providing a reliable operational basis for subsequent sampling operations.

[0059] like Figure 2 FIG. 1 is a system block diagram of a control system for a water-air multi-rotor sampling aircraft provided by an embodiment of the present invention. The control system includes: a model building module for acquiring multi-source sensor data of a sampling aircraft in a water-air environment, performing spatiotemporal calibration on the multi-source sensor data using a Kalman filter fusion algorithm and constructing an aircraft state space model, and outputting flight state data of the sampling aircraft, including real-time position, attitude angle, linear velocity vector, and angular velocity vector; A multi-layer control module is configured to generate an optimal sampling mission path based on the aircraft state space model through a mission planning layer, determine the current flight environment and automatically switch the flight control mode through a medium adaptation layer, and generate a flight execution signal through an execution control layer; An underwater adjustment module is used to dynamically adjust the filling and drainage status of the buoyancy chamber by monitoring the water pressure difference between the upper and lower surfaces of the sampling aircraft in an underwater environment using an underwater pressure difference balance control method, and at the same time, based on the water velocity vector and inertial measurement data, using a thruster vector cooperative control algorithm to coordinately adjust the inclination angle and speed of the four vector thrusters to generate anti-water flow thrust; The sampling adjustment module is used to brake the vector thruster when the sampling aircraft reaches the preset sampling point, and adjust the posture of the sampling robotic arm through the visual servo control algorithm until the sampling head is aligned with the sampling target, triggering the solenoid valve to perform the sampling action and synchronously record the sampling data.

[0060] Figure 2 The apparatus of the embodiment shown can be used to perform Figure 1 The implementation principles and technical effects of the steps in the method embodiment shown are similar and will not be repeated here.

[0061] An electronic device includes a memory and a processor, wherein a computer program is stored in the memory. When the processor runs the computer program stored in the memory, the processor executes the steps of a water-air multi-rotor sampling aircraft control method as described in any one of the above items.

[0062] like Figure 3 FIG. 1 is a schematic diagram of the hardware structure of an electronic device provided by an embodiment of the present invention. The electronic device 30 includes: a processor 31, a memory 32 and a computer program; The memory 32 is used to store the computer program, which may also be a flash memory. The computer program is, for example, an application program or a functional module for implementing the above method.

[0063] The processor 31 is configured to execute the computer program stored in the memory to implement the various steps performed by the device in the above method. For details, please refer to the relevant description in the above method embodiment.

[0064] Optionally, the memory 32 may be independent or integrated with the processor 31 .

[0065] When the memory 32 is a device independent of the processor 31, the device may further include: The bus 33 is used to connect the memory 32 and the processor 31 .

[0066] A readable storage medium stores a computer program, which, when executed by a processor, is used to implement the steps of a water-air multi-rotor sampling aircraft control method as described in any one of the above.

[0067] The readable storage medium may be a computer storage medium or a communication medium. Communication media include any medium that facilitates the transfer of computer programs from one location to another. Computer storage media may be any available medium that can be accessed by a general-purpose or special-purpose computer. For example, a readable storage medium is coupled to a processor, enabling the processor to read information from and write information to the readable storage medium. Of course, the readable storage medium may also be an integral part of the processor. The processor and the readable storage medium may be located in an application-specific integrated circuit (ASIC). In addition, the ASIC may be located in a user device. Of course, the processor and the readable storage medium may also exist as discrete components in a communication device. The readable storage medium may be a read-only memory (ROM), a random access memory (RAM), a CD-ROM, a magnetic tape, a floppy disk, an optical data storage device, and the like.

[0068] The present invention also provides a program product, which includes execution instructions stored in a readable storage medium. At least one processor of a device can read the execution instructions from the readable storage medium, and at least one processor executes the execution instructions so that the device implements the methods provided in the various embodiments described above.

[0069] In the embodiments of the above-mentioned devices, it should be understood that the processor may be a central processing unit (CPU), other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASICs), etc. The general-purpose processor may be a microprocessor or any conventional processor. The steps of the method disclosed in the present invention may be directly executed by a hardware processor or by a combination of hardware and software modules within the processor.

[0070] Through the introduction of the above embodiments, the present invention adopts a water-air multi-rotor sampling aircraft control method and system, obtains multi-source sensor data of the sampling aircraft in the water-air environment, adopts the Kalman filter fusion algorithm to perform time-space calibration on the multi-source sensor data and constructs an aircraft state space model, and outputs the flight state data of the sampling aircraft, including real-time position, attitude angle, linear velocity vector and angular velocity vector; based on the aircraft state space model, generates the optimal sampling task path through the task planning layer, judges the current flight environment through the medium adaptive layer and automatically switches the flight control mode, and generates a flight execution signal through the execution control layer; in the underwater environment, by monitoring The water pressure difference between the upper and lower surfaces of the sampling aircraft is used to dynamically adjust the filling and drainage status of the buoyancy tank using the underwater pressure difference balance control method. At the same time, based on the water flow velocity vector and inertial measurement data, the thruster vector collaborative control algorithm is used to collaboratively adjust the inclination angle and speed of the four vector thrusters to generate anti-water flow thrust; when the sampling aircraft reaches the preset sampling point, the vector thruster is braked, and the posture of the sampling robotic arm is adjusted through the visual servo control algorithm until the sampling head is aligned with the sampling target, triggering the solenoid valve to execute the sampling action and synchronously record the sampling data, so that the aircraft can be positioned and attitude controlled with high precision, ensuring seamless switching and coordination of the control logic during its water-air cross-media operations.

[0071] The present invention can improve underwater positioning accuracy and reduce attitude angle deviation through pressure difference balance control and thruster vector collaborative algorithm, and then can perform high-precision positioning and attitude control on the sampling aircraft, ensuring that the aircraft maintains a stable attitude in complex underwater environments. The method of the present invention adopts an adaptive control algorithm, which can effectively offset the influence of water flow, and can still keep the aircraft stable in a strong water flow environment, reducing the risk of sampling operations. The present invention can significantly reduce the switching time of the water-air mode, improve the cross-media control efficiency, do not need to rely on complex machine learning algorithms, and reduce the demand for hardware computing resources. The method of the present invention can reduce the sampling position deviation, improve the sampling efficiency, ensure the accuracy of the synchronous collection of water quality parameters, meet the high-precision requirements of environmental monitoring, and improve the pertinence and accuracy of environmental monitoring through depth-attitude dual closed-loop control and sampling collaborative control.

[0072] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the above embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for controlling a water-air multi-rotor sampling aircraft, characterized in that: The control method includes: Acquire multi-source sensor data of a sampling aircraft in a water-air environment, perform spatiotemporal calibration on the multi-source sensor data using a Kalman filter fusion algorithm and construct an aircraft state space model, and output flight state data of the sampling aircraft, including real-time position, attitude angle, linear velocity vector, and angular velocity vector; Based on the aircraft state space model, the mission planning layer generates the optimal sampling mission path, the medium adaptation layer determines the current flight environment and automatically switches the flight control mode, and the execution control layer generates a flight execution signal; In an underwater environment, the water pressure difference between the upper and lower surfaces of the sampling aircraft is monitored, and the filling and drainage status of the buoyancy chamber is dynamically adjusted using an underwater pressure difference balance control method. At the same time, based on the water flow velocity vector and inertial measurement data, a thruster vector cooperative control algorithm is used to coordinately adjust the inclination angle and speed of the four vector thrusters to generate anti-water flow thrust; When the sampling aircraft reaches the preset sampling point, the vector thruster is braked, and the posture of the sampling robotic arm is adjusted through the visual servo control algorithm until the sampling head is aligned with the sampling target, triggering the solenoid valve to perform the sampling action and synchronously record the sampling data.

2. The method for controlling a water-air multi-rotor sampling aircraft according to claim 1, characterized in that: The underwater depth value, the upper and lower surface water pressure values, the water flow velocity vector, and the inertial measurement data are collected respectively through the underwater depth sensor, water pressure sensor, Doppler velocimeter, and inertial measurement unit, and the aerial position data, the aerial height value, and the visual image data are collected respectively through the aerial GPS, barometric altimeter, and visual positioning sensor. The underwater depth value, the upper and lower surface water pressure values, the water flow velocity vector, the inertial measurement data, the aerial position data, the aerial height value, and the visual image data are summarized to generate the multi-source sensor data.

3. The method for controlling a water-air multi-rotor sampling aircraft according to claim 1, characterized in that: The calculation process of the Kalman filter fusion algorithm is as follows: Compute a priori state estimates and the prior state estimate covariance as follows: Then calculate the Kalman gain , posterior state estimation and the posterior state estimate covariance as follows: Where, Represents the prior state estimate at time k, including the real-time position, attitude angle, linear velocity vector and angular velocity vector of the sampled aircraft; Represents the state transfer matrix, which is used to reflect the transfer relationship of the aircraft state vector from time k-1 to time k, and has a mapping relationship with the water velocity vector and inertial measurement data; represents the transpose of the state transition matrix; represents the posterior state estimate at time k-1; Represents the control input matrix, which is used to characterize the influence of the control variable on the aircraft state vector; represents the control vector at time k; represents the prior error covariance at time k, which is used to reflect the uncertainty of the prior state estimation and is related to the sensor noise; represents the posterior error covariance at time k-1; represents the process noise covariance, which is used to characterize the uncertainty of the aircraft state space model; represents the Kalman gain matrix, which is used to balance the weight of the prior state estimate and the sensor observation vector; Represents the observation matrix, which is used to map the aircraft state vector to the sensor observation space; represents the transpose of the observation matrix; represents the observation noise covariance, i.e., the sensor measurement error; represents the sensor observation vector at time k, i.e., multi-source sensor data; represents the posterior state estimate at time k; I represents the identity matrix; represents the posterior error covariance at time k.

4. The method for controlling a water-air multi-rotor sampling aircraft according to claim 1, characterized in that: When the task planning layer uses the fast exploration random tree algorithm to generate the optimal sampling task path, the corresponding optimization objective function is: ,in, They represent the starting time and ending time of path planning respectively; v(k) represents the linear velocity vector of the sampling aircraft at time k; represents the Euclidean norm of the linear velocity vector v(k); a(k) represents the acceleration vector of the sampling aircraft at time k, that is, the differential calculation result of the linear velocity vector v(k) at time k; represents the Euclidean norm of the acceleration vector a(k); represents the angular velocity vector of the sampling aircraft at time k; represents the weight of control speed smoothness; Represents the angular velocity vector The Euclidean norm of ; represents the weight of controlling acceleration smoothness; represents the control attitude stability weight.

5. The method for controlling a water-air multi-rotor sampling aircraft according to claim 2, characterized in that: The method of judging the current flight environment and automatically switching the flight control mode through the medium adaptation layer specifically includes: When the underwater depth value l<0.5m and the aerial altitude value h>1m, the current flight environment is determined to be an aerial flight environment and the flight control mode is automatically switched to the aerial flight control mode; When the underwater depth value l satisfies , determining that the current flight environment is a surface flight environment and automatically switching to a surface flight control mode; When the underwater depth value l>2m, the current flight environment is determined to be an underwater flight environment and the underwater flight control mode is automatically switched; The aerial flight control mode is based on the GPS and the visual positioning sensor, and controls the position and attitude of the sampling aircraft by collaboratively adjusting the six-rotor speed; The surface flight control mode is to close part of the rotors, start the buoyancy tank to keep the aircraft floating, and control the sampling aircraft to move on the surface of the water through the two vector thrusters; The underwater flight control mode is to completely close the rotor, start the four vector thrusters, and adjust the sampling aircraft to a pressure difference balance control state.

6. The method for controlling a water-air multi-rotor sampling aircraft according to claim 1, characterized in that: During the switching process of the flight control mode, a smooth transition algorithm is used to generate a mixed control signal and output it to the sampling aircraft. The calculation formula of the mixed control signal is as follows: Where U(k) represents the mixed control signal at time k; Indicates the control signal before mode switching; Indicates the control signal after mode switching; represents the switching weight factor, ; c represents the control switching rate; Indicates the switching start time.

7. The method for controlling a water-air multi-rotor sampling aircraft according to claim 1, characterized in that: The underwater pressure difference balance control method uses a PID controller to adjust the opening of the filling and drainage pump of the buoyancy tank in real time and dynamically adjust the filling and drainage state of the buoyancy tank until the buoyancy change caused by the filling and drainage of the buoyancy tank is ,in, represents the water pressure difference deviation of the sampling aircraft at time k, that is, the difference between the water pressure difference between the upper and lower surfaces and the pressure difference in the equilibrium state; Indicates PID control parameters.

8. The method for controlling a water-air multi-rotor sampling aircraft according to claim 1, characterized in that: The thruster vector cooperative control algorithm solves the inclination angle of the vector thruster through inverse kinematics and speed ,in, represents the water velocity vector; represents the inertial measurement data; m and n represent the inverse kinematics mapping function in the thruster vector cooperative control algorithm; And the inclination The adjustment range is , the speed The adjustment range is .

9. The method for controlling a water-air multi-rotor sampling aircraft according to claim 1, characterized in that: The adjustment process of the visual servo control algorithm is as follows: Identify the sampled target through HOG feature extraction and SVM classifier; Based on the position of the sampling target, the Lucas-Kanade optical flow algorithm is used to determine the desired position of the sampling head; Calculate the joint control amount of the sampling manipulator based on the image Jacobian matrix ,in, represents the Jacobi pseudoinverse; represents the expected position coordinates of the sampling head; o represents the current position coordinates of the sampling head; Represents the control gain coefficient.

10. A water-to-air multi-rotor sampling aircraft control system, applied to a water-to-air multi-rotor sampling aircraft control method according to any one of claims 1 to 9, characterized in that: The control system includes: a model building module for acquiring multi-source sensor data of a sampling aircraft in a water-air environment, performing spatiotemporal calibration on the multi-source sensor data using a Kalman filter fusion algorithm and constructing an aircraft state space model, and outputting flight state data of the sampling aircraft, including real-time position, attitude angle, linear velocity vector, and angular velocity vector; A multi-layer control module is configured to generate an optimal sampling mission path based on the aircraft state space model through a mission planning layer, determine the current flight environment and automatically switch the flight control mode through a medium adaptation layer, and generate a flight execution signal through an execution control layer; An underwater adjustment module is used to dynamically adjust the filling and drainage status of the buoyancy chamber by monitoring the water pressure difference between the upper and lower surfaces of the sampling aircraft in an underwater environment using an underwater pressure difference balance control method, and at the same time, based on the water velocity vector and inertial measurement data, using a thruster vector cooperative control algorithm to coordinately adjust the inclination angle and speed of the four vector thrusters to generate anti-water flow thrust; The sampling adjustment module is used to brake the vector thruster when the sampling aircraft reaches the preset sampling point, and adjust the posture of the sampling robotic arm through the visual servo control algorithm until the sampling head is aligned with the sampling target, triggering the solenoid valve to perform the sampling action and synchronously record the sampling data.

Images