An autonomous dynamic landing control system for a cross dual-rotor unmanned aerial vehicle

Abstract

This invention belongs to the field of autonomous control and intelligent flight technology for unmanned aerial vehicles (UAVs), specifically relating to an autonomous dynamic landing control system for a cross-rotor UAV. The system includes the following steps: acquiring UAV flight status data and target landing platform status data; calculating the relative pose data between the UAV coordinate system and the platform coordinate system based on the UAV flight status data and target landing platform status data; predicting the platform motion status data and landing window time within a preset time period; generating a dynamic landing trajectory point set; calculating position compensation and attitude compensation, and generating thrust distribution commands and attitude control commands; and outputting commands to the cross-rotor motor drive unit to execute the landing action. This invention can solve the problems in existing technologies where sudden changes in noise from a single sensor observation in dynamic scenarios lead to dispersion of the filter covariance, and where static control parameters cannot solve for aerodynamic coupling interference in cross-rotor aircraft.

Description

Technical Field

[0001] This invention belongs to the field of autonomous control and intelligent flight technology for unmanned aerial vehicles (UAVs), specifically relating to an autonomous dynamic landing control system for a cross-rotor UAV. Background Technology

[0002] With the rapid development of unmanned aerial vehicle (UAV) technology, it has been widely applied in logistics transportation, power line inspection, emergency rescue, forest fire prevention, marine monitoring, and low-altitude economy. After completing a mission, UAVs typically need to perform autonomous return and precise landing; therefore, autonomous landing technology has become a crucial technology affecting flight safety and mission efficiency. Cross-rotor UAVs, due to their high lift efficiency, strong wind resistance, excellent hovering performance, and large payload capacity, have significant advantages in complex environments. However, in practical applications, UAVs often need to complete autonomous landing tasks on mobile platforms, in confined spaces, in complex terrain, or in unstructured environments. Affected by factors such as gusts of wind, target platform movement, environmental obstruction, and sensor errors, traditional autonomous landing methods are prone to problems such as trajectory deviation, attitude instability, and landing failure. Most existing UAV autonomous landing systems use preset descent trajectories or fixed control parameters, which have limited adaptability to environmental changes and target platform status variations, making it difficult to meet the high-precision autonomous landing requirements in complex environments.

[0003] Problems with existing technology: In scenarios involving high-frequency maneuvering of the target platform or environmental obstruction, a single sensor is prone to the dispersion of the filter covariance due to sudden changes in observation noise, resulting in the failure of relative pose calculation. At the same time, static control parameters cannot calculate the aerodynamic coupling interference matrix before and after the cross rotor in real time, leading to thrust distribution distortion and trajectory tracking error accumulation, which in turn causes landing attitude instability. Summary of the Invention

[0004] The purpose of this invention is to provide an autonomous dynamic landing control system for a cross-rotor unmanned aerial vehicle (UAV) that can solve the problems in the prior art where sudden changes in noise from a single sensor in dynamic scenarios lead to dispersion of the filter covariance, and where static control parameters cannot solve for aerodynamic coupling interference in cross-rotor aircraft.

[0005] The specific technical solution adopted by this invention is as follows: An autonomous dynamic landing control method for a cross-rotor unmanned aerial vehicle (UAV) includes the following steps: Acquire drone flight status data and target landing platform status data; Based on the UAV flight status data and the target landing platform status data, the relative pose data between the UAV coordinate system and the platform coordinate system is calculated. Based on the historical sequence of the target landing platform status data and the current time data, predict the platform motion status data and landing window time within a preset future time period; Based on the relative pose data, platform motion state data, and landing window time, a dynamic landing trajectory point set is generated; Based on the dynamic landing trajectory point set and the current UAV attitude, calculate the position compensation amount and attitude compensation amount, and generate thrust distribution command and attitude control command. The thrust distribution command and attitude control command are output to the cross-rotor motor drive unit to control the UAV to perform landing actions.

[0006] According to another aspect of the present invention, acquiring target landing platform status data includes: The platform collects image data through a visual recognition unit, point cloud data through a laser ranging unit, and radar echo data through a millimeter-wave radar unit. The platform image data, point cloud data, and radar echo data are time-stamped and aligned with the spatial coordinate system to obtain multi-source heterogeneous sensing data. The extended Kalman filter algorithm is used to estimate the state of multi-source heterogeneous sensing data, and the output includes target landing platform state data including platform position, platform attitude and platform motion velocity.

[0007] According to another aspect of the present invention, the process of state estimation using the extended Kalman filter algorithm further includes: Calculate the measurement covariance matrix of the data from each sensor; When the trace of the measurement covariance matrix of any sensor data exceeds a preset abnormal threshold, it is determined that the corresponding sensor has malfunctioned. In the event of a sensor failure, the observation weight of the failed sensor is reset to zero, and the state estimation equation is updated based on the observation data of the remaining normal sensors to maintain the continuous output of target landing platform state data.

[0008] According to another aspect of the present invention, calculating the position compensation amount and the attitude compensation amount includes: Calculate the relative position deviation between the UAV's expected position and the platform's predicted position, and the relative attitude deviation between the UAV's expected attitude and the platform's predicted attitude; A position feedforward compensation term is constructed based on the relative position deviation, and an attitude feedforward compensation term is constructed based on the relative attitude deviation. The position feedforward compensation term is superimposed on the original position control command to obtain the position compensation amount, and the attitude feedforward compensation term is superimposed on the original attitude control command to obtain the attitude compensation amount.

[0009] According to another aspect of the present invention, predicting platform motion state data and landing window time within a preset future time period includes: Input historical sequence and current time data into a pre-trained motion trend analysis model to output the platform's predicted position, predicted posture and predicted speed within a preset future time period; Calculate the covariance of the prediction residuals and calculate the prediction confidence based on the covariance of the prediction residuals; When the prediction confidence level is higher than the preset confidence threshold, the spatial range that allows the UAV to land safely is calculated based on the platform's predicted position and predicted attitude, and the corresponding time interval is determined as the landing window time.

[0010] According to another aspect of the present invention, it further includes: When the predicted confidence level is lower than or equal to a preset confidence threshold, a degraded landing process is triggered. During the degraded landing process, dynamic trajectory updates are stopped, the current platform position and attitude are used as the static landing target, and hovering and waiting instructions or avoidance trajectory instructions to fly to the preset safe landing area are generated.

[0011] According to another aspect of the present invention, generating a dynamic landing trajectory point set includes: Starting from the current location of the drone and ending at the platform's predicted location within the landing window, construct the initial landing path. Based on the maximum flight speed, maximum acceleration, and aerodynamic constraints of the UAV's cross rotors, the initial landing path is optimized using kinematic constraints to obtain a set of candidate trajectory points. Collision detection is performed on the candidate trajectory point set to remove trajectory points that interfere with obstacles in the environment, and the remaining trajectory points are arranged in time sequence to generate a dynamic landing trajectory point set.

[0012] According to another aspect of the present invention, after generating the dynamic landing trajectory point set, the method further includes: Real-time calculation of the tracking error between the actual position of the UAV and the expected position at the corresponding moment in the dynamic landing trajectory point set; When the Euclidean distance of the tracking error exceeds the preset replanning threshold, or when a new obstacle is detected encroaching on the current trajectory safety corridor, the trajectory replanning process is triggered. In the trajectory replanning process, B-spline curves are used to smooth the regenerated trajectory point set to ensure that the replanned trajectory remains continuous in the dimensions of position, velocity, and acceleration.

[0013] According to another aspect of the present invention, generating thrust distribution instructions includes: Obtain the aerodynamic interference coupling matrix between the front and rear rotors of the cross-rotor UAV; Based on the attitude control commands and position compensation amounts, calculate the expected total thrust and expected three-axis torque of the UAV. By using the inverse matrix of the aerodynamic interference coupling matrix, the total thrust and the desired torque of the three axes are decoupled and calculated to obtain the desired speed of the front rotor and the desired speed of the rear rotor. Based on the desired rotational speeds of the front rotor and the rear rotor, thrust distribution commands are generated and output to the front rotor motor and the rear rotor motor, respectively.

[0014] According to another aspect of the present invention, an autonomous dynamic landing control system for a cross-rotor unmanned aerial vehicle is also provided, comprising: The data acquisition module is used to acquire UAV flight status data and target landing platform status data; The pose calculation module calculates the relative pose data between the UAV coordinate system and the platform coordinate system based on the UAV flight status data and the target landing platform status data. The status prediction module is used to predict the platform motion status data and landing window time within a preset time period based on the historical sequence of the target landing platform status data and the current time data. The trajectory planning module generates a dynamic landing trajectory point set based on the relative pose data, platform motion state data, and landing window time. The compensation and control module calculates the position compensation amount and attitude compensation amount based on the dynamic landing trajectory point set and the current UAV attitude, and generates thrust distribution command and attitude control command. The command output module is used to output the thrust distribution command and attitude control command to the cross-rotor motor drive unit to control the UAV to perform landing actions.

[0015] According to another aspect of the present invention, an electronic device is also provided, the electronic device including a memory and a processor; the memory is used to store a program; the processor executes the program to implement the method described in any one of the foregoing.

[0016] According to another aspect of the present invention, a computer-readable storage medium is also provided, the storage medium storing a computer program that, when executed by a processor, implements the method described in any one of the preceding embodiments.

[0017] According to another aspect of the present invention, a computer program product is also provided, including a computer program that, when executed by a processor, implements the method described in any one of the preceding embodiments.

[0018] The technical effects achieved by this invention are as follows: This invention utilizes a multi-source heterogeneous sensor data timestamp alignment and extended Kalman filter fusion mechanism to ensure that the covariance matrix of the target landing platform's position and attitude state estimation remains convergent when observation noise changes abruptly, thereby achieving high-precision real-time calculation of the relative pose between the UAV and the mobile platform in complex environments. Simultaneously, by introducing a feedforward compensation mechanism based on platform motion prediction and a cubic B-spline trajectory smoothing algorithm, the dynamic landing trajectory point set remains continuous in the dimensions of position, velocity, and acceleration, thereby eliminating dynamic tracking errors and mechanical resonance when the cross-rotor UAV approaches the mobile platform at high speed.

[0019] This invention employs an aerodynamic interference coupling matrix containing off-diagonal elements and its pseudo-inverse matrix for thrust distribution calculation. This ensures that the additional pitch and yaw moments generated by the front rotor downwash on the rear rotor are precisely canceled out during the control command generation stage, thereby achieving attitude stability control of the cross-rotor UAV during dynamic landing. Furthermore, by using a long short-term memory network to extract temporal features from the platform's historical state sequence, the covariance of the predicted platform motion state within a preset time period is reduced, thereby improving the accuracy of landing window time assessment.

[0020] This invention, by monitoring the trace of the measurement covariance matrix in real time during the extended Kalman filter update step, enables the system to adaptively reset the corresponding observation weights to zero and rely on the remaining sensors to maintain state estimation when a single sensor experiences a soft failure. This avoids the divergence in relative pose calculation caused by single-point sensor failure. Simultaneously, by setting a main control processor occupancy threshold to trigger a model degradation strategy and a motor physical maximum thrust threshold to trigger a thrust redistribution mechanism, the system can automatically switch to a low-computing-power-consumption model or redistribute the thrust deficit under extreme conditions such as limited edge computing resources or rotor load saturation. This ensures the hard real-time performance and flight safety of the autonomous dynamic landing control loop. Attached Figure Description

[0021] Figure 1 This is a block diagram of the overall system structure of the present invention; Figure 2 This is a flowchart of the target platform perception module in this invention; Figure 3 This is a schematic diagram of the working principle of the relative pose estimation module in this invention; Figure 4 This is a flowchart of the target platform state prediction process in this invention; Figure 5 This is a schematic diagram of the flight status perception structure of the cross-rotor UAV in this invention; Figure 6 This is a flowchart of the dynamic landing trajectory planning process in this invention; Figure 7This is a schematic diagram of the relative motion compensation between the UAV and the target platform in this invention; Figure 8 This is a schematic diagram of the autonomous dynamic landing control principle in this invention. Detailed Implementation

[0022] To make the objectives and advantages of this invention clearer, the invention will be specifically described below with reference to embodiments. It should be understood that the following text is merely used to describe one or more specific embodiments of the invention and does not strictly limit the scope of protection specifically claimed by the invention.

[0023] It should be noted that the terms "first," "second," etc., used in this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0024] According to an embodiment of the present invention, a method embodiment of an autonomous dynamic landing control method for a cross-rotor unmanned aerial vehicle is provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0025] In one optional embodiment, an autonomous dynamic landing control system (hereinafter referred to as the system) for a cross-rotor unmanned aerial vehicle is provided, which is deployed on the airborne flight control computing platform of the cross-rotor unmanned aerial vehicle.

[0026] As a preferred approach, the flight control computing platform adopts a heterogeneous multi-core architecture, including an ARM application processor as the main controller and an FPGA (Field Programmable Gate Array) chip as a coprocessor.

[0027] Further, refer to the appendix Figure 1 and attached Figure 5The sensor array in the perception layer adopts a distributed spatial layout. GNSS antennas, forward-looking vision cameras, IMU inertial measurement units, barometric altimeters, and laser rangefinders (downward-looking) are deployed on the left and front of the UAV to acquire global positioning and high-frequency rigid body motion data. Rotor speed sensors (front / rear), accelerometers (fuselage), magnetometers / temperature and humidity sensors, millimeter-wave radar (forward-looking), and vibration sensors (various parts of the fuselage) are deployed on the right side of the fuselage and at the center of gravity to capture the aerodynamic coupling high-frequency flutter and body attitude unique to cross-rotor aircraft. Data from multiple heterogeneous sensors are aggregated to the flight control computing platform via CAN-FD bus, SPI high-speed interface, and MIPI video interface.

[0028] Furthermore, the high-frequency raw data acquired by the sensor is directly written into the FPGA's on-chip BRAM (Block RAM) via the DMA (Direct Memory Access) controller, bypassing the CPU cache and thus ensuring microsecond-level data throughput latency. The FPGA is responsible for performing low-level image preprocessing, point cloud filtering, and fast timestamp alignment, and then transmits the processed feature vectors to the ARM host's SRAM via the AXI bus for use by the upper-level state prediction network and trajectory planning module.

[0029] See attached document Figure 1 As an optional embodiment, the overall system structure is divided into a perception and estimation layer, a decision and planning layer, and an autonomous control and execution layer.

[0030] The perception and estimation layer includes: The flight status perception module includes an IMU inertial measurement unit, a GNSS positioning unit, a barometric altimeter, a flight status detection unit, a vibration status detection unit, and a rotor speed detection unit, which are used to output UAV status data. The target platform perception module includes a visual recognition unit, a laser ranging unit, a millimeter-wave radar unit, and a platform status detection unit, which is used to output platform status data; The relative pose estimation module includes a coordinate transformation unit, a relative position calculation unit, a relative attitude calculation unit, and an error estimation unit, which is used to output relative pose data. The target platform status prediction module includes a historical status database, a motion trend analysis unit, a status prediction unit, and a landing window evaluation unit, which are used to output prediction results.

[0031] The decision-making and planning level includes: The dynamic landing trajectory planning module includes a landing target generation unit, an optimal trajectory planning unit, an obstacle avoidance unit, and a landing window matching unit, which are used to output the optimal landing trajectory.

[0032] The autonomous control and execution layer includes: The autonomous control decision module includes a thrust distribution unit, an attitude control unit, a velocity control unit, a trajectory tracking unit, and an adaptive gain adjustment unit, which are used to output control commands. The flight execution module includes a motor drive unit, a rotor actuator, and a flight control output unit.

[0033] As described above, the external environment and flight status inputs are processed by the perception and estimation layer to generate relative pose and platform predicted state (solid line data flow), which are then input to the decision and planning layer for trajectory generation. Finally, the autonomous control and execution layer calculates the thrust distribution command (dashed line control flow) to drive the cross-rotor UAV to perform a dynamic landing mission.

[0034] As an optional embodiment, refer to the appendix Figures 1 to 8 An autonomous dynamic landing control method for a cross-rotor unmanned aerial vehicle (UAV) includes the following steps: S1. Acquire UAV flight status data and target landing platform status data; S2. Based on the UAV flight status data and the target landing platform status data, calculate the relative pose data between the UAV coordinate system and the platform coordinate system; S3. Based on the historical sequence of target landing platform status data and the current data, predict the platform motion status data and landing window time within a preset time period in the future; S4. Based on relative pose data, platform motion state data, and landing window time, generate a dynamic landing trajectory point set; S5. Based on the dynamic landing trajectory point set and the current UAV attitude, calculate the position compensation amount and attitude compensation amount, and generate thrust distribution command and attitude control command. S6. Output thrust distribution commands and attitude control commands to the cross-rotor motor drive unit to control the UAV to perform landing actions.

[0035] As an optional embodiment, refer to the appendix Figure 2 Regarding the aforementioned multi-source heterogeneous sensing data, when performing target platform sensing, the system performs timestamp alignment and spatial coordinate system unification on the collected visual, laser, and radar raw data, and then uses the extended Kalman filter (EKF) algorithm for state estimation.

[0036] Furthermore, the specific workflow of the target platform perception module includes: Start; Sensor initialization; Vision / LiDAR / radar data acquisition; Target platform identification and localization; Platform attitude estimation; Platform motion velocity estimation; Platform boundary and size detection; Data fusion and filtering; Output platform status information; End.

[0037] The final output platform status information includes: platform location. Platform posture Platform speed Platform Boundaries and platform size .

[0038] As an optional embodiment, refer to the appendix Figure 3 At the relative pose estimation level, the system is responsible for establishing the UAV coordinate system. (include (axis) and platform coordinate system (include Mapping between axes.

[0039] Input status includes drone status (location) ,attitude ) and platform status (location) ,attitude ).

[0040] Furthermore, the working principle of the relative pose estimation module includes: Coordinate transformation calculation; relative position calculation; relative pose calculation; error estimation and compensation; output relative pose; end. In the calculation, the relative pose calculated by the system is denoted as... Where, relative position is represented as The relative attitude is represented by Euler angles. .

[0041] As an optional embodiment, refer to the appendix Figure 7 To address the tracking lag issue during dynamic landing, the system employs feedforward compensation based on platform motion prediction. Input data includes platform motion... and platform motion prediction .

[0042] like Figure 7 As shown, the system first calculates the relative pose error: Position error:

[0043] Attitude error:

[0044] Motion compensation calculations were then performed to obtain the position compensation amount. With attitude compensation amount The final compensated control command is as follows: Position commands:

[0045] Attitude commands:

[0046] (Note: In engineering implementation, when the attitude is represented by Euler angles and the deviation is small, attitude compensation can be approximated by addition; if quaternions are used, it can be achieved by quaternion multiplication.) Through the above compensation, the actual flight trajectory (uncompensated) is corrected to the compensated trajectory, closely matching the UAV's desired trajectory (it should be noted that the trajectory in the figure is only used to represent the trajectory path and does not represent the actual trajectory position).

[0047] As an optional embodiment, refer to the appendix Figure 4 At the target platform state prediction level, the system uses a motion trend analysis model to infer the future state of the platform.

[0048] The target platform status prediction process specifically includes: start; historical status data acquisition; real-time status data input; data preprocessing and feature extraction; motion trend analysis; status prediction model calculation; landing window evaluation; output prediction results; end.

[0049] Furthermore, the state prediction model employs a temporal prediction architecture based on a Long Short-Term Memory (LSTM) network. The output prediction results include: future position. Future posture Future Speed Landing window and confidence level .when When the confidence level exceeds the preset confidence threshold, the system combines platform boundaries. With size Calculate the effective projected area that allows for safe contact of the drone's landing gear, and assign the corresponding time interval. It has been identified as the final landing window.

[0050] As an optional embodiment, refer to the appendix Figure 6 At the dynamic trajectory planning level, after generating the initial landing path, the system needs to perform kinematic constraint optimization and smoothing. The dynamic landing trajectory planning process specifically includes: Start; Receive relative pose and prediction results; Generate landing propulsion target point; Environmental perception and obstacle detection; Trajectory optimization calculation; Trajectory feasibility check; Landing window matching; Output optimal landing trajectory; End.

[0051] Furthermore, the system uses cubic B-spline curves to parameterize the trajectory point set. The final output includes: landing trajectory. Trajectory point set Trajectory Time and landing method By optimizing the control points, the generated trajectory minimizes the jerk while meeting the limitations of the UAV's maximum speed, maximum acceleration, and the tilt angle specific to cross-rotor rotors, thereby ensuring the smoothness of the motor output.

[0052] As an optional embodiment, refer to the appendix Figure 8 At the autonomous control and thrust distribution level, considering the unique aerodynamic coupling characteristics of the cross-rotor UAV, the thrust distribution unit in the autonomous control decision module performs inverse matrix decoupling calculations. The closed-loop process of the autonomous dynamic landing control principle specifically includes: Perception and estimation: UAV state perception, platform state perception, relative pose estimation, state prediction; Decision-making and planning: landing trajectory planning, landing timing decision-making, and control strategy generation; Autonomous control: thrust distribution control, attitude stabilization control, velocity control, trajectory tracking control, and adaptive gain adjustment; Execution and feedback: motor drive, rotor actuator, flight status update, control effect feedback.

[0053] The execution results are fed back to the perception and estimation stage, forming a closed-loop control.

[0054] Furthermore, the expected total thrust is defined as The desired torque across the three axes is Because the downwash airflow from the front rotor significantly interferes with the rear rotor's aerodynamics, the system constructs an aerodynamic interference coupling matrix containing off-diagonal elements. .

[0055] The pseudo-inverse formula for the thrust distribution matrix is ​​as follows:

[0056]

[0057] in, This represents the total demand control torque and thrust vector. Represents the Moore-Penrose pseudo-inverse of the coupling matrix; This represents the calculated basic thrust command vectors for the front and rear rotors; This represents the aerodynamic coupling compensation correction amount. The final output... This is the final control command issued to the flight control output unit and the motor drive unit.

[0058] As an optional implementation, a degradation process is implemented for sensor data packet loss or timeout: If continuous packet loss of visual image data or radar point cloud data is detected, and the packet loss duration exceeds a preset timeout threshold (e.g., 50 milliseconds), the inference of the state prediction network is paused to save computational resources. An extended Kalman filter is invoked, and based on the state vector and system state transition matrix of the last valid moment before packet loss, the platform motion state at the current moment is recursively estimated. The dynamic trajectory planning module is downgraded to a static hovering mode, and a feedforward compensation torque is calculated based on the recursively estimated platform state to maintain basic flight stability.

[0059] Furthermore, to address computing bottlenecks caused by limited edge computing resources or high loads: when the utilization rate of the flight control main CPU exceeds 85% for three consecutive control cycles, model degradation is triggered. The system automatically switches the time-series prediction model containing a Long Short-Term Memory (LSTM) network to a Gated Recurrent Unit (GRU) network with fewer parameters, or directly calls the "Platform State-Control Gain" three-dimensional lookup table generated offline based on historical platform motion data for interpolation calculations, sacrificing a slight amount of prediction accuracy to ensure that the control loop meets the hard real-time requirements of over 500Hz.

[0060] Furthermore, if any rotor thrust value in the thrust distribution control command exceeds the motor's physical maximum thrust threshold (e.g., PWM duty cycle reaches 95% or more), thrust redistribution is triggered. The rotor thrust value exceeding the motor's physical maximum thrust threshold is truncated to the motor's physical maximum thrust threshold, and the thrust deficit is calculated. Based on the lever arm ratio of the cross-rotor UAV, the thrust deficit is proportionally distributed to the other rotor that is not saturated, and the aerodynamic coupling compensation coefficient of the other rotor is adjusted synchronously, and the thrust distribution control command is updated.

[0061] As an optional implementation, taking a cross-rotor UAV performing a resupply mission on a ship at sea as an example, when the UAV approaches the ship's deck, which is swaying with the waves, the onboard millimeter-wave radar penetrates the sea fog to obtain the precise distance to the deck, and the forward-looking vision camera uses a polarizing filter to suppress strong reflections from the sea surface. The FPGA coprocessor completes the timestamp alignment of multi-source data, and the LSTM model on the ARM main controller quickly predicts the future... Within a given moment, the deck will experience a "non-safe window" with a roll angle greater than 5 degrees, and the landing window that subsequently appears will be precisely locked in. Based on this, the system generates a "follow-up descent" B-spline trajectory with vertical velocity synchronization. During the thrust distribution phase, considering the surge in rear rotor load caused by the ship's wake, the system proactively compensates for the rear rotor's thrust using the inverse of the aerodynamic interference coupling matrix. If the rear rotor reaches its physical limits, the system immediately triggers a thrust redistribution mechanism, transferring the lost thrust to the front rotor and correcting the yaw moment. Through this coordinated data flow and hardware execution, the UAV accurately captures landing windows in adverse sea conditions, achieving a high success rate of autonomous dynamic landing.

[0062] As an optional embodiment, refer to the appendix Figure 1 It also provides an autonomous dynamic landing control system for a cross-rotor unmanned aerial vehicle, including: The data acquisition module is used to acquire UAV flight status data and target landing platform status data; The pose calculation module is used to calculate the relative pose data between the UAV coordinate system and the platform coordinate system based on the UAV flight status data and the target landing platform status data. The status prediction module is used to predict the platform motion status data and landing window time within a preset time period based on the historical sequence of target landing platform status data and the current data. The trajectory planning module is used to generate a dynamic landing trajectory point set based on relative pose data, platform motion state data, and landing window time. The compensation and control module is used to calculate the position compensation and attitude compensation based on the dynamic landing trajectory point set and the current UAV attitude, and to generate thrust distribution commands and attitude control commands. The command output module is used to output thrust distribution commands and attitude control commands to the cross-rotor motor drive unit to control the UAV to perform landing actions.

[0063] This application also provides an autonomous dynamic landing control device for a cross-rotor unmanned aerial vehicle (UAV), comprising: an airborne flight control computing platform, which adopts a heterogeneous multi-core architecture, including an ARM application processor as the main controller and an FPGA chip as a coprocessor; a multi-source sensor array, connected to the FPGA chip via a direct memory access controller, for collecting UAV flight status data and target landing platform status data and writing them into the on-chip block random access memory of the FPGA chip; and a cross-rotor motor drive unit, which is communicatively connected to the airborne flight control computing platform; wherein, the FPGA chip is configured to perform low-level signal filtering and timestamp alignment, and the ARM application processor is configured to execute the method of any of the foregoing embodiments, and convert the generated thrust allocation command into a pulse width modulation waveform and output it to the cross-rotor motor drive unit.

[0064] This application also provides a computer-readable storage medium deployed in the non-volatile memory of a cross-rotor unmanned aerial vehicle (UAV). The computer-readable storage medium stores computer instructions for causing the computer to execute the method of any of the foregoing embodiments. Furthermore, during the execution of the method, the computer instructions are configured to synchronously write multi-source heterogeneous sensing data, relative pose data, and thrust allocation instructions into the black box data recording area of ​​the non-volatile memory for offline calibration of aerodynamic coupling parameters and data backtracking after landing.

[0065] It should be noted that the above modules can be implemented by software or hardware. For the latter, they can be implemented in the following ways, but are not limited to: all the above modules are located in the same processor; or, the above modules are located in different processors in any combination.

[0066] The above description is merely a preferred embodiment of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention. Structures, devices, and operating methods not specifically described or explained in this invention are implemented according to conventional methods in the art unless otherwise specified or limited.

Claims

1. A method for autonomous dynamic landing control of a cross-rotor unmanned aerial vehicle, characterized in that, Includes the following steps: Acquire drone flight status data and target landing platform status data; Based on the UAV flight status data and the target landing platform status data, the relative pose data between the UAV coordinate system and the platform coordinate system is calculated. Based on the historical sequence of the target landing platform status data and the current time data, predict the platform motion status data and landing window time within a preset future time period; Based on the relative pose data, platform motion state data, and landing window time, a dynamic landing trajectory point set is generated; Based on the dynamic landing trajectory point set and the current UAV attitude, calculate the position compensation amount and attitude compensation amount, and generate thrust distribution command and attitude control command. The thrust distribution command and attitude control command are output to the cross-rotor motor drive unit to control the UAV to perform landing actions.

2. The autonomous dynamic landing control method for a cross-rotor unmanned aerial vehicle according to claim 1, characterized in that, Obtain target landing platform status data, including: The platform collects image data through a visual recognition unit, point cloud data through a laser ranging unit, and radar echo data through a millimeter-wave radar unit. The platform image data, point cloud data, and radar echo data are time-stamped and aligned with the spatial coordinate system to obtain multi-source heterogeneous sensing data. The extended Kalman filter algorithm is used to estimate the state of multi-source heterogeneous sensing data, and the output includes target landing platform state data including platform position, platform attitude and platform motion velocity.

3. The autonomous dynamic landing control method for a cross-rotor unmanned aerial vehicle according to claim 2, characterized in that, The process of using the extended Kalman filter algorithm for state estimation also includes: Calculate the measurement covariance matrix of the data from each sensor; When the trace of the measurement covariance matrix of any sensor data exceeds a preset abnormal threshold, it is determined that the corresponding sensor has malfunctioned. In the event of a sensor failure, the observation weight of the failed sensor is reset to zero, and the state estimation equation is updated based on the observation data of the remaining normal sensors to maintain the continuous output of target landing platform state data.

4. The autonomous dynamic landing control method for a cross-rotor unmanned aerial vehicle according to claim 1, characterized in that, Calculate the position compensation and attitude compensation amounts, including: Calculate the relative position deviation between the UAV's expected position and the platform's predicted position, and the relative attitude deviation between the UAV's expected attitude and the platform's predicted attitude; A position feedforward compensation term is constructed based on the relative position deviation, and an attitude feedforward compensation term is constructed based on the relative attitude deviation. The position feedforward compensation term is superimposed on the original position control command to obtain the position compensation amount, and the attitude feedforward compensation term is superimposed on the original attitude control command to obtain the attitude compensation amount.

5. The autonomous dynamic landing control method for a cross-rotor unmanned aerial vehicle according to claim 1, characterized in that, Predict platform motion status data and landing window time within a preset future time period, including: Input historical sequence and current time data into a pre-trained motion trend analysis model to output the platform's predicted position, predicted posture and predicted speed within a preset time period in the future. Calculate the covariance of the prediction residuals and calculate the prediction confidence based on the covariance of the prediction residuals; When the prediction confidence level is higher than the preset confidence threshold, the spatial range that allows the UAV to land safely is calculated based on the platform's predicted position and predicted attitude, and the corresponding time interval is determined as the landing window time.

6. The autonomous dynamic landing control method for a cross-rotor unmanned aerial vehicle according to claim 5, characterized in that, Also includes: When the predicted confidence level is lower than or equal to a preset confidence threshold, a degraded landing process is triggered. During the degraded landing process, dynamic trajectory updates are stopped, the current platform position and attitude are used as the static landing target, and hovering and waiting instructions or avoidance trajectory instructions to fly to the preset safe landing area are generated.

7. The autonomous dynamic landing control method for a cross-rotor unmanned aerial vehicle according to claim 1, characterized in that, Generate a dynamic landing trajectory point set, including: Starting from the current location of the drone and ending at the platform's predicted location within the landing window, construct the initial landing path. Based on the maximum flight speed, maximum acceleration, and aerodynamic constraints of the UAV's cross rotors, the initial landing path is optimized using kinematic constraints to obtain a set of candidate trajectory points. Collision detection is performed on the candidate trajectory point set to remove trajectory points that interfere with obstacles in the environment, and the remaining trajectory points are arranged in time sequence to generate a dynamic landing trajectory point set.

8. The autonomous dynamic landing control method for a cross-rotor unmanned aerial vehicle according to claim 7, characterized in that, After generating the dynamic landing trajectory point set, the following is also included: Real-time calculation of the tracking error between the actual position of the UAV and the expected position at the corresponding moment in the dynamic landing trajectory point set; When the Euclidean distance of the tracking error exceeds the preset replanning threshold, or when a new obstacle is detected encroaching on the current trajectory safety corridor, the trajectory replanning process is triggered. In the trajectory replanning process, B-spline curves are used to smooth the regenerated trajectory point set to ensure that the replanned trajectory remains continuous in the dimensions of position, velocity, and acceleration.

9. The autonomous dynamic landing control method for a cross-rotor unmanned aerial vehicle according to claim 1, characterized in that, Generate thrust distribution instructions, including: Obtain the aerodynamic interference coupling matrix between the front and rear rotors of the cross-rotor UAV; Based on the attitude control commands and position compensation amounts, calculate the expected total thrust and expected three-axis torque of the UAV. By using the inverse matrix of the aerodynamic interference coupling matrix, the total thrust and the desired torque of the three axes are decoupled and calculated to obtain the desired speed of the front rotor and the desired speed of the rear rotor. Based on the desired rotational speeds of the front rotor and the rear rotor, thrust distribution commands are generated and output to the front rotor motor and the rear rotor motor, respectively.

10. An autonomous dynamic landing control system for a cross-rotor unmanned aerial vehicle, operating the method described in any one of claims 1 to 9, characterized in that, include: The data acquisition module is used to acquire UAV flight status data and target landing platform status data; The pose calculation module calculates the relative pose data between the UAV coordinate system and the platform coordinate system based on the UAV flight status data and the target landing platform status data. The status prediction module is used to predict the platform motion status data and landing window time within a preset time period based on the historical sequence of the target landing platform status data and the current time data. The trajectory planning module generates a dynamic landing trajectory point set based on the relative pose data, platform motion state data, and landing window time. The compensation and control module calculates the position compensation amount and attitude compensation amount based on the dynamic landing trajectory point set and the current UAV attitude, and generates thrust distribution command and attitude control command. The command output module is used to output the thrust distribution command and attitude control command to the cross-rotor motor drive unit to control the UAV to perform landing actions.

Images