Unmanned aerial vehicle aerial photography attitude real-time correction system and method based on self-adaptive holder

Abstract

The invention relates to the technical field of unmanned aerial vehicle aerial photography, in particular to an unmanned aerial vehicle aerial photography attitude real-time correction system and method based on a self-adaptive holder, an unmanned aerial vehicle body bears the self-adaptive holder, an attitude detection module is in two-way communication with a data processing module, and the data processing module outputs a control signal to a driving control module; the driving control module directly drives the self-adaptive holder to adjust the posture, and the visual auxiliary module is in one-way communication with the data processing module to provide supplementary deviation correction data. The technical problems that an existing unmanned aerial vehicle aerial photography attitude correction system depends on a single IMU sensor, detection precision is low, response lags behind, the anti-interference capability is weak, sensor redundancy is lacked, visual feedback and attitude prediction are not combined, and high-precision aerial survey requirements such as a 1: 500 proportional scale and the like are difficult to meet are solved.

Description

Technical Field

[0001] This invention relates to the field of UAV aerial photography technology, specifically to a real-time attitude correction system and method for UAV aerial photography based on an adaptive gimbal. Background Technology

[0002] With its advantages of high flexibility, low cost, and high efficiency, UAV aerial photography technology has been widely used in fields such as topographic mapping, urban planning, agricultural monitoring, and disaster assessment. During aerial photography, the stability of the UAV's attitude angles (roll angle, pitch angle, yaw angle) directly determines the geometric accuracy and clarity of the aerial images, which is the foundation for subsequent data processing (such as stitching, modeling, and measurement).

[0003] In existing technologies, drone attitude correction mainly relies on traditional gimbal systems, which detect attitude deviations using a single IMU sensor and employ simple proportional control to drive the motors. However, such systems have the following drawbacks:

[0004] Insufficient detection accuracy: Single IMU sensors are susceptible to interference from fuselage vibrations, and long-term operation will result in cumulative errors, leading to attitude detection errors ≥0.5°, which cannot meet the requirements of high-precision aerial surveys such as 1:500 scale.

[0005] Response lag: A passive correction strategy is adopted, and adjustment is only initiated when the deviation exceeds the threshold. The dynamic response time is ≥0.5s, and overshoot is prone to occur in complex environments such as airflow disturbances.

[0006] Weak anti-interference capability: The transmission gimbal has a fixed shock absorption structure, and its attenuation rate for 5-200Hz wideband vibration is less than or equal to 60%, resulting in an image blur rate of more than 30% due to vibration.

[0007] Lack of redundancy design: There is no backup plan in case of single sensor failure, which may lead to the interruption of aerial photography missions.

[0008] Furthermore, existing methods struggle to handle drift errors during long-endurance operations and are prone to abnormal image overlap (deviation greater than or equal to 5%) in low-altitude aerial photography of complex terrain. Therefore, developing a high-precision, fast-response, and interference-resistant real-time attitude correction system and method for UAV aerial photography has become a pressing technical problem to be solved in this field. Summary of the Invention

[0009] To address the shortcomings of existing technologies, the purpose of this invention is to provide a real-time attitude correction system and method for UAV aerial photography based on an adaptive gimbal. This invention solves the technical problems of existing UAV aerial photography attitude correction systems, which rely on a single IMU sensor, resulting in low detection accuracy, slow response, weak anti-interference capability, lack of sensor redundancy, and failure to combine visual feedback and attitude prediction, making it difficult to meet the high-precision aerial survey requirements such as 1:500 scale.

[0010] The technical solution adopted by this invention to solve its technical problem is:

[0011] A real-time attitude correction system for UAV aerial photography based on an adaptive gimbal includes a UAV body, an adaptive gimbal mounted below the UAV body for mounting aerial photography equipment, and further includes:

[0012] The attitude detection module includes dual six-axis IMU sensors and a high-precision GPS module. The dual six-axis IMU sensors synchronously acquire the real-time attitude angles of the UAV at a set frequency, and the high-precision GPS module acquires the real-time position coordinates of the UAV with a set positioning accuracy.

[0013] The data processing module is connected to the attitude detection module and has a preset attitude deviation threshold. It uses a Kalman filter algorithm to fuse the data collected by the dual six-axis IMU sensors to obtain the filtered attitude angle. The attitude deviation value is calculated based on the filtered attitude angle and the target attitude angle.

[0014] The drive control module is connected to the data processing module and the adaptive gimbal respectively. When the attitude deviation value is greater than the attitude deviation threshold, the correction torque output by the three-axis brushless motor in the adaptive gimbal is determined based on the attitude deviation value, and the three-axis brushless motor is driven according to the correction torque until the attitude deviation value meets the accuracy expectation.

[0015] The adaptive gimbal also includes a three-axis damping bracket, which is equipped with a vibration sensor. The vibration sensor detects the vibration acceleration in real time and sends it to the data processing module. When the data processing module determines that the vibration acceleration is greater than the acceleration threshold, it increases the damping coefficient of the three-axis damping bracket.

[0016] As a preferred embodiment, a further technical solution of the present invention is:

[0017] Preferably, it also includes a vision assistance module, which uses an industrial camera connected to the data processing module to capture landmark images in real time and send them to the data processing module;

[0018] The data processing module calculates the offset between the real-time pixel coordinates of the landmark image and the target pixel coordinates. The specific calculations are as follows:

[0019] ;

[0020] in, Real-time pixel coordinates The target pixel coordinates;

[0021] and offset Converted into visual correction angle The specific calculations are as follows:

[0022] ;

[0023] in, =0.2° / pixel is the pixel-to-angle conversion factor;

[0024] Finally, the visual correction angle is... Superimposed on attitude deviation value The joint bias is obtained. and will Updated to attitude deviation value .

[0025] Preferably, the data processing module uses the ARIMA(2,1,1) model to predict attitude deviations in future time intervals. The expression corresponding to the ARIMA(2,1,1) model is:

[0026] ;

[0027] in, for The first-order difference, Represents the attitude deviation value at time t , , For regression coefficients, The moving average coefficient is... , It is white noise;

[0028] When the predicted attitude deviation When the angle is greater than 0.3°, the drive control module bases its decisions on the predicted attitude deviation value. Determine the correction torque output of the three-axis brushless motor in the adaptive gimbal, and drive the three-axis brushless motor according to the correction torque until the attitude deviation value meets the accuracy expectation, thus achieving pre-compensation.

[0029] Preferably, the data processing module obtains the filtered attitude angle based on the following expression:

[0030] ;

[0031] in, Let [θ1, θ2, θ3]ᵀ be the filtered attitude angle vector at time k, where θ1 is the roll angle, θ2 is the pitch angle, and θ3 is the yaw angle. It is a 3×3 state transition matrix. It is a 3×1 control matrix. This represents the control quantity for the three-axis brushless motor at time k-1. It is a 3×3 Kalman gain matrix. Let k be the observation vector from the dual six-axis IMU sensor. It is a 3×3 observation matrix;

[0032] When the attitude angle difference |θ1_IMU1-θ1_IMU2| collected by the dual six-axis IMU sensors is greater than 0.2°, an IMU fault alarm is triggered, and the system switches to the backup IMU2 to collect attitude angle data.

[0033] Preferably, the process by which the data processing module calculates the attitude deviation value based on the filtered attitude angle and the target attitude angle is represented as follows:

[0034] ;

[0035] in, This is the attitude deviation value. , , These are the target roll angle, target pitch angle, and target yaw angle required for aerial photography.

[0036] Preferably, the process by which the drive module determines the correction torque output by the three-axis brushless motor in the adaptive gimbal based on the attitude deviation value is represented as follows:

[0037] T = Kt × Δθ;

[0038] Where T is the correction torque, and Kt is the motor torque coefficient, taken as 0.5 N·m / °. This represents the attitude deviation value.

[0039] This invention also discloses a real-time attitude correction method for UAV aerial photography based on an adaptive gimbal, which is applied to the above-mentioned system. The specific steps are as follows:

[0040] S1: Attitude Data Acquisition

[0041] The drone's real-time attitude angles are synchronously acquired at a set frequency using dual six-axis IMU sensors, and the drone's real-time position coordinates are acquired at a set positioning accuracy using a high-precision GPS module.

[0042] S2: Attitude Deviation Calculation

[0043] The Kalman filter algorithm is used to fuse real-time attitude angle data to obtain filtered attitude angles; the attitude deviation value is calculated based on the filtered attitude angles and the target attitude angles.

[0044] S3: Deviation Judgment

[0045] Determine if the attitude deviation value is greater than the attitude deviation threshold. If it is, execute S4; otherwise, return to S1.

[0046] S4: Adaptive Gimbal Correction Driver

[0047] The correction torque of the three-axis brushless motor in the adaptive gimbal is determined based on the attitude deviation value, and the three-axis brushless motor is driven according to the correction torque until the attitude deviation value meets the accuracy expectation.

[0048] S5: Damping Adjustment

[0049] The vibration acceleration is detected in real time by the vibration sensor of the adaptive gimbal. When the vibration acceleration is determined to be greater than the acceleration threshold, the damping coefficient of the three-axis vibration damping bracket in the adaptive gimbal is increased.

[0050] Preferred options also include:

[0051] S6: Visual Acquisition and Correction

[0052] Landmark images are captured using an industrial camera, and the offset between the real-time pixel coordinates of the landmark image and the target pixel coordinates is calculated. The specific calculations are as follows:

[0053] ;

[0054] in, Real-time pixel coordinates The target pixel coordinates;

[0055] and offset Converted into visual correction angle The specific calculations are as follows:

[0056] ;

[0057] in, =0.2° / pixel is the pixel-to-angle conversion factor;

[0058] Finally, the visual correction angle is... Superimposed on attitude deviation value The joint bias is obtained. and will Updated to attitude deviation value .

[0059] Preferred options also include:

[0060] S7: Attitude Deviation Prediction and Pre-compensation

[0061] The ARIMA(2,1,1) model is used to predict attitude deviations in future time intervals. The expression corresponding to the ARIMA(2,1,1) model is:

[0062] ;

[0063] in, for The first-order difference, Represents the attitude deviation value at time t , , For regression coefficients, The moving average coefficient is... , It is white noise;

[0064] When the predicted attitude deviation When the angle is greater than 0.3°, the attitude deviation is based on the predicted value. Determine the correction torque output of the three-axis brushless motor in the adaptive gimbal, and drive the three-axis brushless motor according to the correction torque until the attitude deviation value meets the accuracy expectation, thus achieving pre-compensation.

[0065] Preferably, S2 specifically includes:

[0066] The filtered attitude angles are obtained based on the following expression:

[0067] ;

[0068] in, Let [θ1, θ2, θ3]ᵀ be the filtered attitude angle vector at time k, where θ1 is the roll angle, θ2 is the pitch angle, and θ3 is the yaw angle. It is a 3×3 state transition matrix. It is a 3×1 control matrix. This represents the control quantity for the three-axis brushless motor at time k-1. It is a 3×3 Kalman gain matrix. Let k be the observation vector from the dual six-axis IMU sensor. It is a 3×3 observation matrix;

[0069] When the attitude angle difference |θ1_IMU1-θ1_IMU2| collected by the dual six-axis IMU sensors is greater than 0.2°, an IMU fault alarm is triggered, and the system switches to the backup IMU2 to collect attitude angle data.

[0070] The process of calculating the attitude deviation value based on the filtered attitude angle and the target attitude angle is as follows:

[0071] ;

[0072] in, This is the attitude deviation value. , , These are the target roll angle, target pitch angle, and target yaw angle required for aerial photography;

[0073] The process of determining the correction torque output of the three-axis brushless motor in the adaptive gimbal based on the attitude deviation value in S4 is as follows:

[0074] T = Kt × Δθ;

[0075] Where T is the correction torque, and Kt is the motor torque coefficient, taken as 0.5 N·m / °. This represents the attitude deviation value.

[0076] The present invention, which adopts the above technical solution, has the following prominent features compared with the prior art:

[0077] To improve attitude detection accuracy and reliability, a dual-IMU sensor redundancy design combined with a Kalman filter algorithm is used to reduce the attitude detection error from ±0.5° to ±0.1°, and the noise suppression ratio after data fusion is ≥40dB. When the main IMU fails (deviation >0.2°), it can switch to the backup IMU within 10ms to ensure continuous system operation and reduce the task interruption rate to below 0.1%.

[0078] To achieve high-precision real-time correction, a closed-loop control architecture of deviation calculation-drive control-feedback adjustment is adopted, with a correction response time of ≤0.1s, which is 80% shorter than that of traditional systems. Through precise control of the output torque (Kt=0.5N·m / °) of the 3-axis brushless motor, the steady-state attitude deviation can be controlled within ±0.1°, meeting the geometric accuracy requirements of 1:500 scale aerial survey.

[0079] Enhanced anti-interference and vibration reduction capabilities: The adaptive vibration damping bracket increases the attenuation rate of vibrations in the 5-200Hz frequency band to ≥90% by adjusting the damping coefficient (500-800N·s / m), and reduces the image blur rate caused by vibration to below 5%; in complex environments with wind speeds ≤5m / s, the overlap deviation of aerial photography images can be controlled within ±2%.

[0080] The optimized correction strategy and adaptive introduction of the vision-assisted module achieve joint correction through landmark pixel offset conversion (k=0.02° / pixel), which compensates for the accumulated error of the IMU and reduces the attitude drift by 60% during long-endurance (≥30min) operations. The ARIMA model predicts the deviation in the next 0.1s, and the pre-compensation control makes the dynamic overshoot ≤5%, adapting to rapidly changing airflow disturbance scenarios.

[0081] Expanding application scenarios: The system weighs ≤500g and has a maximum load of 1.5kg. It can be used with SLR cameras, multispectral cameras and other equipment. It is suitable for fields with strict requirements for attitude stability, such as high-precision surveying, aerial photography of cultural relics and power line inspection, which significantly improves the quality of aerial photography data and the efficiency of subsequent processing. Attached Figure Description

[0082] Figure 1This is a schematic diagram of the structure of the UAV aerial photography attitude real-time correction system based on an adaptive gimbal in an embodiment of the present invention. Detailed Implementation

[0083] The present invention will be further illustrated below with reference to specific embodiments. The purpose of this illustration is solely to provide a better understanding of the invention. Therefore, the examples given do not limit the scope of protection of the present invention.

[0084] like Figure 1 As shown, this embodiment presents a real-time attitude correction system for UAV aerial photography based on an adaptive gimbal, including a UAV body, an adaptive gimbal mounted below the UAV body, the adaptive gimbal being used to mount aerial photography equipment, and further including:

[0085] The attitude detection module includes dual six-axis IMU sensors and a high-precision GPS module. The dual six-axis IMU sensors synchronously acquire the real-time attitude angles of the UAV at a set frequency (≥100Hz), and the high-precision GPS module acquires the real-time position coordinates of the UAV at a set positioning accuracy (≤1m).

[0086] The data processing module is connected to the attitude detection module and has a preset attitude deviation threshold. It uses a Kalman filter algorithm to fuse the data collected by the dual six-axis IMU sensors to obtain the filtered attitude angle. The attitude deviation value is calculated based on the filtered attitude angle and the target attitude angle.

[0087] The drive control module is connected to the data processing module and the adaptive gimbal respectively. When the attitude deviation value is greater than the attitude deviation threshold (0.5°), the correction torque output by the three-axis brushless motor in the adaptive gimbal is determined based on the attitude deviation value, and the three-axis brushless motor is driven according to the correction torque until the attitude deviation value meets the accuracy expectation (≤0.1°).

[0088] The adaptive gimbal also includes a three-axis damping bracket, which is equipped with a vibration sensor. The vibration sensor detects the vibration acceleration in real time and sends it to the data processing module. When the data processing module determines that the vibration acceleration is greater than the acceleration threshold (0.1g), it increases the damping coefficient of the three-axis damping bracket.

[0089] The implementation also includes a vision assistance module, which uses an industrial camera connected to the data processing module to capture landmark images in real time and send them to the data processing module.

[0090] The data processing module can use the Sobel operator to extract the edge features of preset ground landmarks (such as crosshairs), and then calculate the offset between the real-time pixel coordinates of the landmark image and the target pixel coordinates. The specific calculations are as follows:

[0091] ;

[0092] in, Real-time pixel coordinates The target pixel coordinates;

[0093] and offset Converted into visual correction angle The specific calculations are as follows:

[0094] ;

[0095] in, =0.2° / pixel is the pixel-to-angle conversion factor;

[0096] Finally, the visual correction angle is... Superimposed on attitude deviation value The joint bias is obtained. and will Updated to attitude deviation value .

[0097] In practice, the UAV body carries an adaptive gimbal, the attitude detection module and the data processing module communicate bidirectionally, the data processing module outputs control signals to the drive control module, the drive control module directly drives the adaptive gimbal to adjust the attitude, and the vision assistance module communicates unidirectionally with the data processing module to provide supplementary correction data.

[0098] The drone body uses a quadcopter drone platform, with a maximum takeoff weight of 3.5kg, a flight time of ≥30min, and is equipped with a 2.4GHz wireless data transmission module, which can transmit attitude data and aerial images in real time.

[0099] Adaptive gimbal:

[0100] Mechanical structure: It adopts a 3-axis (roll axis, pitch axis, yaw axis) orthogonal structure, and is made of aviation aluminum alloy. Its weight is ≤500g and its maximum load is 1.5kg (compatible with SLR cameras, multispectral cameras and other aerial photography equipment).

[0101] Drive unit: Built-in 3-axis brushless motor, rated voltage 12V, maximum speed 3000rpm, torque coefficient Kt=0.5N·m / °, response time ≤5ms.

[0102] Vibration damping components: Integrated 3-axis vibration damping bracket, adopting a spring-damping composite vibration damping structure, with an initial damping coefficient c=500N·s / m and a natural frequency f0=10Hz. The damping coefficient can be adjusted to c'=800N·s / m via electromagnetic damping, and the attenuation rate of vibration in the 5-200Hz frequency band is ≥90%.

[0103] Attitude detection module:

[0104] Six-axis IMU sensor: Employs MPU6050 chip (main IMU1) and BMI160 chip (backup IMU2), synchronously acquiring roll angle (θ1), pitch angle (θ2), and yaw angle (θ3), with a sampling frequency of f=100Hz, a measurement range of ±20°, and an accuracy of ±0.1°.

[0105] High-precision GPS module: adopts Beidou + GPS dual-mode positioning, with positioning accuracy ≤1m (single point positioning) and ≤0.1m (differential positioning), update frequency 10Hz, and output position coordinates (X,Y,Z).

[0106] Vibration sensor: A MEMS accelerometer is used, with a measurement range of ±2g and a resolution of 0.001g, where g is 9.8m / s², to detect the vibration acceleration a of the fuselage.

[0107] The data processing module is based on the STM32H743 microprocessor (400MHz).

[0108] Industrial camera: 1920×1080 resolution, 30fps frame rate, 8mm lens focal length, 60° field of view, mounted below the adaptive gimbal, parallel to the optical axis of the aerial photography equipment.

[0109] The data processing module uses an ARIMA(2,1,1) model (autoregressive order p=2, differencing order d=1, moving average order q=1) to predict attitude deviations within 0.1 seconds in the future. With a sampling interval of 0.01s, the expression for the ARIMA(2,1,1) model is:

[0110] ;

[0111] in, for The first-order difference, Represents the attitude deviation value at time t , , For regression coefficients, The moving average coefficient is... , It is white noise;

[0112] When the predicted attitude deviation When the angle is greater than 0.3°, the drive control module bases its decisions on the predicted attitude deviation value. Determine the correction torque output of the three-axis brushless motor in the adaptive gimbal, and drive the three-axis brushless motor according to the correction torque until the attitude deviation value meets the accuracy expectation, thus achieving pre-compensation.

[0113] In practice, the data processing module obtains the filtered attitude angles based on the following expression:

[0114] ;

[0115] in, Let [θ1, θ2, θ3]ᵀ be the filtered attitude angle vector at time k, where θ1 is the roll angle, θ2 is the pitch angle, and θ3 is the yaw angle. It is a 3×3 state transition matrix. It is a 3×1 control matrix. This represents the control quantity for the three-axis brushless motor at time k-1. The 3×3 Kalman gain matrix is ​​updated in real time according to the minimum mean square error criterion. Let k be the observation vector from the dual six-axis IMU sensor. It is a 3×3 observation matrix;

[0116] When the attitude angle difference |θ1_IMU1-θ1_IMU2| collected by the dual six-axis IMU sensors is greater than 0.2°, an IMU fault alarm is triggered, and the system switches to the backup IMU2 to collect attitude angle data.

[0117] In practice, the process by which the data processing module calculates the attitude deviation value based on the filtered attitude angle and the target attitude angle is as follows:

[0118] ;

[0119] in, This is the attitude deviation value. , , These are the target roll angle, target pitch angle, and target yaw angle required for aerial photography.

[0120] In practice, the process by which the drive module determines the correction torque output by the three-axis brushless motor in the adaptive gimbal based on the attitude deviation value is as follows:

[0121] T = Kt × Δθ;

[0122] Where T is the correction torque, and Kt is the motor torque coefficient, taken as 0.5 N·m / °. This represents the attitude deviation value. After obtaining the correction torque, the drive module can use a conventional PID closed-loop control algorithm (proportional coefficient Kp=5.0, integral coefficient Ki=0.1, derivative coefficient Kd=0.5) to adjust the motor speed and achieve attitude closed-loop control.

[0123] Increase the damping coefficient of the triaxial damping bracket. Specifically, the damping coefficient can be adjusted from c=500N·s / m to c'=800N·s / m to attenuate vibrations in the 5-200Hz range until a≤0.1g.

[0124] This invention also discloses a real-time attitude correction method for UAV aerial photography based on an adaptive gimbal, which is applied to the above-mentioned system. The specific steps are as follows:

[0125] S1: Attitude Data Acquisition

[0126] The drone's real-time attitude angles are synchronously acquired at a set frequency using dual six-axis IMU sensors, and the drone's real-time position coordinates are acquired at a set positioning accuracy using a high-precision GPS module.

[0127] S2: Attitude Deviation Calculation

[0128] The Kalman filter algorithm is used to fuse real-time attitude angle data to obtain filtered attitude angles; the attitude deviation value is calculated based on the filtered attitude angles and the target attitude angles.

[0129] S3: Deviation Judgment

[0130] Determine if the attitude deviation value is greater than the attitude deviation threshold. If it is, execute S4; otherwise, return to S1.

[0131] S4: Adaptive Gimbal Correction Driver

[0132] The correction torque of the three-axis brushless motor in the adaptive gimbal is determined based on the attitude deviation value, and the three-axis brushless motor is driven according to the correction torque until the attitude deviation value meets the accuracy expectation.

[0133] S5: Damping Adjustment

[0134] The vibration acceleration is detected in real time by the vibration sensor of the adaptive gimbal. When the vibration acceleration is determined to be greater than the acceleration threshold, the damping coefficient of the three-axis vibration damping bracket in the adaptive gimbal is increased.

[0135] The implementation also includes:

[0136] S6: Visual Acquisition and Correction

[0137] Landmark images are captured using an industrial camera, and the offset between the real-time pixel coordinates of the landmark image and the target pixel coordinates is calculated. The specific calculations are as follows:

[0138] ;

[0139] in, Real-time pixel coordinates The target pixel coordinates;

[0140] and offset Converted into visual correction angle The specific calculations are as follows:

[0141] ;

[0142] in, =0.2° / pixel is the pixel-to-angle conversion factor;

[0143] Finally, the visual correction angle is... Superimposed on attitude deviation value The joint bias is obtained. and will Updated to attitude deviation value .

[0144] The implementation also includes:

[0145] S7: Attitude Deviation Prediction and Pre-compensation

[0146] The ARIMA(2,1,1) model is used to predict attitude deviations in future time intervals. The expression corresponding to the ARIMA(2,1,1) model is:

[0147] ;

[0148] in, for The first-order difference, Represents the attitude deviation value at time t , , For regression coefficients, The moving average coefficient is... , It is white noise;

[0149] When the predicted attitude deviation When the angle is greater than 0.3°, the attitude deviation is based on the predicted value. Determine the correction torque output of the three-axis brushless motor in the adaptive gimbal, and drive the three-axis brushless motor according to the correction torque until the attitude deviation value meets the accuracy expectation, thus achieving pre-compensation.

[0150] In implementation, S2 specifically includes:

[0151] The filtered attitude angles are obtained based on the following expression:

[0152] ;

[0153] in, Let [θ1, θ2, θ3]ᵀ be the filtered attitude angle vector at time k, where θ1 is the roll angle, θ2 is the pitch angle, and θ3 is the yaw angle. It is a 3×3 state transition matrix. It is a 3×1 control matrix. This represents the control quantity for the three-axis brushless motor at time k-1. It is a 3×3 Kalman gain matrix. Let k be the observation vector from the dual six-axis IMU sensor. It is a 3×3 observation matrix;

[0154] When the attitude angle difference |θ1_IMU1-θ1_IMU2| collected by the dual six-axis IMU sensors is greater than 0.2°, an IMU fault alarm is triggered, and the system switches to the backup IMU2 to collect attitude angle data.

[0155] The process of calculating the attitude deviation value based on the filtered attitude angle and the target attitude angle is as follows:

[0156] ;

[0157] in, This is the attitude deviation value. , , These are the target roll angle, target pitch angle, and target yaw angle required for aerial photography;

[0158] The process of determining the correction torque output of the three-axis brushless motor in the adaptive gimbal based on the attitude deviation value in S4 is as follows:

[0159] T = Kt × Δθ;

[0160] Where T is the correction torque, and Kt is the motor torque coefficient, taken as 0.5 N·m / °. This represents the attitude deviation value.

[0161] The technical solution of this invention improves the accuracy and reliability of attitude detection by using a dual IMU sensor redundancy design combined with a Kalman filter algorithm. The attitude detection error is reduced from ±0.5° to ±0.1°, and the noise suppression ratio after data fusion is ≥40dB. When the main IMU fails (deviation >0.2°), it can switch to the backup IMU within 10ms to ensure continuous system operation and reduce the task interruption rate to below 0.1%.

[0162] To achieve high-precision real-time correction, a closed-loop control architecture of deviation calculation-drive control-feedback adjustment is adopted, with a correction response time of ≤0.1s, which is 80% shorter than that of traditional systems. Through precise control of the output torque (Kt=0.5N·m / °) of the 3-axis brushless motor, the steady-state attitude deviation can be controlled within ±0.1°, meeting the geometric accuracy requirements of 1:500 scale aerial survey.

[0163] Enhanced anti-interference and vibration reduction capabilities: The adaptive vibration damping bracket increases the attenuation rate of vibrations in the 5-200Hz frequency band to ≥90% by adjusting the damping coefficient (500-800N·s / m), and reduces the image blur rate caused by vibration to below 5%; in complex environments with wind speeds ≤5m / s, the overlap deviation of aerial photography images can be controlled within ±2%.

[0164] The optimized correction strategy and adaptive introduction of the vision-assisted module achieve joint correction through landmark pixel offset conversion (k=0.02° / pixel), which compensates for the accumulated error of the IMU and reduces the attitude drift by 60% during long-endurance (≥30min) operations. The ARIMA model predicts the deviation in the next 0.1s, and the pre-compensation control makes the dynamic overshoot ≤5%, adapting to rapidly changing airflow disturbance scenarios.

[0165] Expanding application scenarios: The system weighs ≤500g and has a maximum load of 1.5kg. It can be used with SLR cameras, multispectral cameras and other equipment. It is suitable for fields with strict requirements for attitude stability, such as high-precision surveying, aerial photography of cultural relics and power line inspection, which significantly improves the quality of aerial photography data and the efficiency of subsequent processing.

[0166] The above description is merely a preferred embodiment of the present invention and is not intended to limit the scope of the invention. All equivalent changes made based on the description and drawings of the present invention are included within the scope of the present invention.

Claims

1. A real-time attitude correction system for UAV aerial photography based on an adaptive gimbal, comprising a UAV body, an adaptive gimbal mounted below the UAV body, the adaptive gimbal being used to mount aerial photography equipment, characterized in that, Also includes: The attitude detection module includes dual six-axis IMU sensors and a high-precision GPS module. The dual six-axis IMU sensors synchronously acquire the real-time attitude angles of the UAV at a set frequency, and the high-precision GPS module acquires the real-time position coordinates of the UAV with a set positioning accuracy. The data processing module is connected to the attitude detection module and has a preset attitude deviation threshold. It uses the Kalman filter algorithm to fuse the data collected by the dual six-axis IMU sensors to obtain the filtered attitude angle. The attitude deviation value is calculated based on the filtered attitude angle and the target attitude angle. The drive control module is connected to the data processing module and the adaptive gimbal respectively. When the attitude deviation value is greater than the attitude deviation threshold, the correction torque output by the three-axis brushless motor in the adaptive gimbal is determined based on the attitude deviation value, and the three-axis brushless motor is driven according to the correction torque until the attitude deviation value meets the accuracy expectation. The adaptive gimbal also includes a three-axis damping bracket, which is equipped with a vibration sensor. The vibration sensor detects the vibration acceleration in real time and sends it to the data processing module. When the data processing module determines that the vibration acceleration is greater than the acceleration threshold, it increases the damping coefficient of the three-axis damping bracket.

2. The real-time attitude correction system for UAV aerial photography based on an adaptive gimbal according to claim 1, characterized in that, It also includes a vision assistance module, which uses an industrial camera connected to the data processing module to capture landmark images in real time and send them to the data processing module. The data processing module calculates the offset between the real-time pixel coordinates of the landmark image and the target pixel coordinates. The specific calculations are as follows: ； in, Real-time pixel coordinates The target pixel coordinates; and offset Converted into visual correction angle The specific calculations are as follows: ； in, =0.2° / pixel is the pixel-to-angle conversion factor; Finally, the visual correction angle is... Superimposed on attitude deviation value The joint bias is obtained. and will Updated to attitude deviation value .

3. The real-time attitude correction system for UAV aerial photography based on an adaptive gimbal according to claim 1, characterized in that, The data processing module uses the ARIMA(2,1,1) model to predict attitude deviations in future time intervals. The expression corresponding to the ARIMA(2,1,1) model is: ； in, for The first-order difference, Represents the attitude deviation value at time t , , For regression coefficients, The moving average coefficient is... , It is white noise; When the predicted attitude deviation When the angle is greater than 0.3°, the drive control module bases its decisions on the predicted attitude deviation value. Determine the correction torque output of the three-axis brushless motor in the adaptive gimbal, and drive the three-axis brushless motor according to the correction torque until the attitude deviation value meets the accuracy expectation, thus achieving pre-compensation.

4. The real-time attitude correction system for UAV aerial photography based on an adaptive gimbal according to claim 1, characterized in that, The data processing module obtains the filtered attitude angles based on the following expression: ； in, Let [θ1, θ2, θ3] be the filtered attitude angle vector at time k. T θ1 is the roll angle, θ2 is the pitch angle, and θ3 is the yaw angle. It is a 3×3 state transition matrix. It is a 3×1 control matrix. This represents the control quantity for the three-axis brushless motor at time k-1. It is a 3×3 Kalman gain matrix. Let k be the observation vector from the dual six-axis IMU sensor. It is a 3×3 observation matrix; When the attitude angle difference |θ1_IMU1-θ1_IMU2| collected by the dual six-axis IMU sensors is greater than 0.2°, an IMU fault alarm is triggered, and the system switches to the backup IMU2 to collect attitude angle data.

5. The real-time attitude correction system for UAV aerial photography based on an adaptive gimbal according to claim 4, characterized in that, The process by which the data processing module calculates the attitude deviation based on the filtered attitude angle and the target attitude angle is as follows: ； in, This is the attitude deviation value. , , These are the target roll angle, target pitch angle, and target yaw angle required for aerial photography.

6. The real-time attitude correction system for UAV aerial photography based on an adaptive gimbal according to claim 1, characterized in that, The process by which the drive module determines the correction torque output by the three-axis brushless motor in the adaptive gimbal based on the attitude deviation value is as follows: T = Kt × Δθ; Where T is the correction torque, and Kt is the motor torque coefficient, taken as 0.5 N·m / °. This represents the attitude deviation value.

7. A method for real-time attitude correction in UAV aerial photography based on an adaptive gimbal, characterized in that, The specific steps are as follows, when applied to the system described in any one of claims 1 to 6: S1: Attitude Data Acquisition The drone's real-time attitude angles are synchronously acquired at a set frequency using dual six-axis IMU sensors, and the drone's real-time position coordinates are acquired at a set positioning accuracy using a high-precision GPS module. S2: Attitude Deviation Calculation The Kalman filter algorithm is used to fuse real-time attitude angle data to obtain filtered attitude angles; the attitude deviation value is calculated based on the filtered attitude angles and the target attitude angles. S3: Deviation Judgment Determine if the attitude deviation value is greater than the attitude deviation threshold. If it is, execute S4; otherwise, return to S1. S4: Adaptive Gimbal Correction Driver The correction torque of the three-axis brushless motor in the adaptive gimbal is determined based on the attitude deviation value, and the three-axis brushless motor is driven according to the correction torque until the attitude deviation value meets the accuracy expectation. S5: Damping Adjustment The vibration acceleration is detected in real time by the vibration sensor of the adaptive gimbal. When the vibration acceleration is determined to be greater than the acceleration threshold, the damping coefficient of the three-axis vibration damping bracket in the adaptive gimbal is increased.

8. The method for real-time attitude correction of UAV aerial photography based on adaptive gimbal according to claim 7, characterized in that, Also includes: S6: Visual Acquisition and Correction Landmark images are captured using an industrial camera, and the offset between the real-time pixel coordinates of the landmark image and the target pixel coordinates is calculated. The specific calculations are as follows: ； in, Real-time pixel coordinates The target pixel coordinates; and offset Converted into visual correction angle The specific calculations are as follows: ； in, =0.2° / pixel is the pixel-to-angle conversion factor; Finally, the visual correction angle is... Superimposed on attitude deviation value The joint bias is obtained. and will Updated to attitude deviation value .

9. The method for real-time attitude correction of UAV aerial photography based on adaptive gimbal according to claim 7, characterized in that, Also includes: S7: Attitude Deviation Prediction and Pre-compensation The ARIMA(2,1,1) model is used to predict attitude deviations in future time intervals. The expression corresponding to the ARIMA(2,1,1) model is: ； in, for The first-order difference, Represents the attitude deviation value at time t , , For regression coefficients, The moving average coefficient is... , It is white noise; When the predicted attitude deviation When the angle is greater than 0.3°, the attitude deviation is based on the predicted value. Determine the correction torque output of the three-axis brushless motor in the adaptive gimbal, and drive the three-axis brushless motor according to the correction torque until the attitude deviation value meets the accuracy expectation, thus achieving pre-compensation.

10. The method for real-time attitude correction of UAV aerial photography based on adaptive gimbal according to claim 7, characterized in that, S2 specifically includes: The filtered attitude angles are obtained based on the following expression: ； in, Let [θ1, θ2, θ3] be the filtered attitude angle vector at time k. T θ1 is the roll angle, θ2 is the pitch angle, and θ3 is the yaw angle. It is a 3×3 state transition matrix. It is a 3×1 control matrix. This represents the control quantity for the three-axis brushless motor at time k-1. It is a 3×3 Kalman gain matrix. Let k be the observation vector from the dual six-axis IMU sensor. It is a 3×3 observation matrix; When the attitude angle difference |θ1_IMU1-θ1_IMU2| collected by the dual six-axis IMU sensors is greater than 0.2°, an IMU fault alarm is triggered, and the system switches to the backup IMU2 to collect attitude angle data. The process of calculating the attitude deviation value based on the filtered attitude angle and the target attitude angle is as follows: ； in, This is the attitude deviation value. , , These are the target roll angle, target pitch angle, and target yaw angle required for aerial photography; The process of determining the correction torque output of the three-axis brushless motor in the adaptive gimbal based on the attitude deviation value in S4 is as follows: T = Kt × Δθ; Where T is the correction torque, and Kt is the motor torque coefficient, taken as 0.5 N·m / °. This represents the attitude deviation value.

Images