Unmanned aerial vehicle flight control system based on attitude information reconstruction

Abstract

The invention discloses an unmanned aerial vehicle flight control system based on attitude information reconstruction. The system aims to realize high-precision attitude information reconstruction by decomposing and re-fusing attitude information provided by a flight controller and an external fusion system. The specific implementation method is as follows: the system firstly uses pitch angle and roll angle data provided by a flight control processor, combines yaw angle data provided by a visual inertia information fusion system, and adopts a weighted fusion algorithm to process the data to reconstruct complete and accurate attitude information; through a weighted fusion algorithm, the system can distribute appropriate weight for each data source according to the precision and reliability of different sensors, and it is ensured that the reconstruction result has high accuracy. In addition, in order to further improve the stability and precision of the system, the system is also combined with calibration parameters, and errors caused by sensor noise and external environment changes are reduced. The reconstructed attitude information is applied to flight control of the unmanned aerial vehicle, and flight stability and control precision are effectively improved. The system can effectively solve the problem of attitude information floating possibly occurring in the starting process of an external fusion system, and therefore control deviation caused by attitude information floating is avoided. The system is especially suitable for unmanned aerial vehicle flight tasks needing high-precision control, such as vertical take-off, precise positioning and hovering, and can meet the task requirements of strict flight control precision requirements.

Description

Technical Field

[0001] The present invention relates to the field of UAV flight control, and particularly to a UAV flight control method based on attitude information reconstruction, which is applicable to UAV takeoff and high-precision positioning control scenarios. Background Art

[0002] With the rapid development of UAV technology, it is widely used in various tasks, including aerial monitoring, logistics delivery, agricultural spraying, aerial photography, etc. However, the accuracy and stability of UAV flight tasks still face many challenges, especially flight control in complex environments. The flight control system of UAVs usually relies on sensors, such as inertial measurement units (IMUs), accelerometers, and gyroscopes, to obtain flight attitude information for control. In order to obtain more accurate attitude estimation, more and more UAV systems use fusion sensor technologies, such as visual inertial navigation systems and laser inertial navigation systems.

[0003] Existing UAV attitude information acquisition systems have some limitations. First of all, inertial sensors have cumulative errors. As time goes by, the errors will gradually increase, resulting in deviations or "drifts" in attitude information. Especially during long-term flights or high-speed flights, the errors will intensify. These errors affect the flight control accuracy. Especially during vertical takeoff, precise positioning, or long-term hovering, unnecessary flight deviations will be caused, thus affecting the success rate of flight tasks.

[0004] Although the flight controller can provide relatively stable attitude information such as pitch angle and roll angle, the accuracy of its yaw angle is often insufficient to meet the requirements of complex flight tasks. External fusion systems, especially visual inertial navigation systems, although they can provide relatively accurate yaw angle information, during the startup of the fusion system, attitude information may "drift", resulting in deviations in the flight control system.

[0005] The existing attitude information acquisition methods in the prior art often cannot completely solve the problems of imbalance or control deviation during flight, especially during high-speed flights and in complex environments. Therefore, how to efficiently fuse information from different sensors and eliminate errors in attitude information in real time has become a key technology to improve the flight accuracy and stability of UAVs.

[0006] Based on this, a UAV flight control system based on attitude information reconstruction is proposed. By decomposing and reconstructing the attitude information provided by the flight controller and the external fusion system, high-precision attitude information reconstruction is achieved, thereby improving the accuracy and stability of flight control. This method can effectively solve the problem of attitude information drift during the startup of the external fusion system and reduce the resulting control deviation, and is particularly applicable to UAV flight tasks with strict requirements for high-precision control. Summary of the Invention

[0007] Technical problem solved by the present invention: To provide an unmanned aerial vehicle (UAV) flight control system based on attitude information reconstruction, which overcomes the problem of attitude information drift in existing information fusion technologies, and provides an unmanned aerial vehicle flight control system based on attitude information reconstruction. By decomposing and reconstructing the attitude information provided by the flight controller and the external fusion system, the present invention significantly improves the flight accuracy and stability.

[0008] Technical solution of the present invention: An unmanned aerial vehicle flight control system based on attitude information reconstruction, comprising an attitude information reconstruction module, a binocular camera sensing module, a flight control processing module, a visual-inertial information fusion module, a flight control module, a flight execution module, and a communication module. Among them, the binocular camera sensing module and the flight control processing module provide visual data and gyroscope accelerometer data for the visual-inertial information fusion module to perform fusion processing to generate visual-inertial fusion attitude information; the flight control processing module provides pitch angle and roll angle information based on inertial sensors for the attitude information reconstruction module to perform attitude information reconstruction; the attitude information reconstruction module receives the attitude information provided by the flight control processing module and the visual-inertial information fusion module, and reconstructs the pitch angle, roll angle, and yaw angle through a weighted fusion algorithm, and outputs accurate attitude information to the flight control module to control the flight of the unmanned aerial vehicle; the flight control module generates flight control instructions according to the requirements of the flight mission and transmits them to the flight execution module; the communication module ensures efficient data exchange between each module, maintains synchronization and coordination among the system modules, and ensures the efficient operation of the unmanned aerial vehicle control system.

[0009] The said attitude information reconstruction module consists of an attitude information decomposer and an attitude information reconstructor. The attitude information decomposer is used to extract and decompose the attitude information in the flight control processing module and the visual-inertial information fusion module; the attitude information reconstructor is used to perform fusion calculation on the decomposed attitude information and output the reconstructed attitude information to the flight control module.

[0010] The basic principle of the present invention is as follows: In the flight control system of an unmanned aerial vehicle (UAV), the attitude information provided by the flight controller and the external fusion system is often affected by sensor errors and environmental noise, resulting in drift or deviation of the attitude information, thereby affecting flight accuracy and stability. To achieve high-precision flight control, the present invention decomposes and reconstructs the attitude information provided by the flight controller and the external fusion system to eliminate these errors and ensure flight stability. That is, the pitch angle and roll angle information provided by the flight controller, and the yaw angle information provided by the external visual-inertial fusion navigation system are processed through a weighted fusion algorithm. First, these data are decomposed into different attitude angles, namely the yaw angle, pitch angle, and roll angle. Then, using the weighted fusion algorithm, according to the noise characteristics and accuracy requirements of different sensors, the weights of each sensor are dynamically adjusted to accurately reconstruct new attitude information. Through this reconstruction process, the reconstructed attitude information not only eliminates the attitude drift problem that may occur during the startup process of the external fusion system but also eliminates the errors caused by the sensor noise of the flight controller, thereby achieving more precise flight control. The flight control module uses this precise attitude information to generate flight instructions to ensure the flight stability and accuracy of the UAV in high-precision tasks.

[0011] The system schematic diagram is as Figure 1 shown. In the flight control processing module, first, the raw data obtained from the gyroscope and accelerometer are subjected to integration processing, horizon calibration, and magnetometer correction. The purpose of this process is to eliminate the drift error of the sensor and the influence of environmental factors on attitude measurement, so as to obtain relatively accurate attitude information, denoted as P a , and the inertial sensor data in the flight control processing module are fused with the visual data collected by the camera to form a visual-inertial fusion attitude that combines visual information and inertial data, denoted as P a . Both P a and P b are output in the form of quaternions. A quaternion is a mathematical tool used to represent three-dimensional rotations, which can avoid the singularity problem generated by Euler angles near the polar angle, such as 90 degrees. Then, the attitude reconstruction module converts these two quaternions. First, the quaternion of P a is converted into Euler angles, and the pitch angle ψ a and roll angle θ a are extracted from them; then, the quaternion of P b is converted into Euler angles, and the yaw angle φ b is extracted. Finally, through the weighted fusion algorithm, the pitch angle, roll angle, and yaw angle are combined to obtain the final reconstructed P c , and this attitude information will be used as the basis for flight control. In this way, the flight control system can comprehensively use the information from different sensors to perform accurate attitude reconstruction, effectively overcome the limitations of each sensor itself, and achieve high-precision flight control.

[0012] The main principle of the attitude information decomposer is as follows: extract key attitude information from the data of multiple modules. This information includes the accelerometer and gyroscope data provided by the flight control processing module, as well as the data of the visual inertial information fusion module. All data are in quaternion form, describing the attitude of the aircraft. A quaternion is a mathematical tool for representing three-dimensional rotations, which can avoid the singularity problem of Euler angles. In order to extract the actual attitude angles from the quaternion, that is, the pitch angle, roll angle, and yaw angle, the conversion formula from quaternion to Euler angle is used. Specifically, given the quaternion q=(q0,q1,q2,q3), its formula for converting to Euler angles is: pitch angle yaw angle roll angle φ = arcsin[2(q0q2 + q1q3)]. Through this formula, the elements q0, q1, q2, q3 in the quaternion can be converted into Euler angles, obtaining the pitch angle and roll angle provided by the flight control processing module, and the yaw angle provided by the external visual inertial information fusion module. These Euler angles will be sent to the attitude information decomposer for processing. In the above process, the sensor data of the flight controller usually has a certain amount of noise, and these noises may accumulate over time and affect the accuracy of attitude estimation. Therefore, the core role of the decomposer is to independently process the data obtained from different sensors, so as to extract accurate attitude information and provide reliable basic data for subsequent fusion and reconstruction.

[0013] The main principle of the attitude information reconstructor is as follows: fuse and correct the pitch angle, roll angle, and yaw angle information obtained from the flight control processing module and the visual inertial information fusion module to generate more accurate attitude information. These data are first extracted and decomposed into individual attitude angles by the decomposer, and then the reconstructor accurately reconstructs these angles through a weighted fusion algorithm, and finally outputs the synthesized attitude information for the flight control module to use. The core of the weighted fusion algorithm lies in dynamically adjusting the weights according to the signal noise characteristics and measurement accuracy of each sensor. This algorithm is processed according to the following formula: reconstructed yaw angle reconstructed pitch angle reconstructed roll angle where are the reconstructed yaw angle, pitch angle, and roll angle. ψ, θ, φ are the initial yaw angle, pitch angle, and roll angle obtained from the flight controller. are the estimated yaw angle, pitch angle, and roll angle of the external fusion system, and ω1, ω2, ω3, ω4, ω5, ω6 are weight coefficients set according to sensor noise and accuracy. These weight coefficients ω1, ω2, ω3, ω4, ω5, ω6 are set based on the noise characteristics, sensitivity, and experimental calibration data of the sensors. The adjustment of the weights is based on the actual performance of the sensors. That is, during the operation of the system, the data accuracy and stability of the sensors may vary, so the weight values will also be dynamically adjusted to ensure that the final attitude information can minimize errors as much as possible. The goal of the reconstructor is to eliminate errors caused by sensor noise or external interference through precise fusion of data from different sensors, especially the attitude drift phenomenon that occurs when the external system starts. The process of weighted fusion ensures that all data extracted from the flight control processing module and the vision-inertial information fusion module can contribute to the final attitude information, thereby maximizing the accuracy of flight control. Description of the Drawings

[0014] Figure 1 is the schematic diagram of an unmanned aerial vehicle flight control system based on attitude information reconstruction according to the present invention;

[0015] Figure 2 is the schematic diagram of the composition of an unmanned aerial vehicle flight control system based on attitude information reconstruction according to the present invention;

[0016] Figure 3 is the flow chart of the attitude information reconstruction algorithm according to the present invention;

[0017] Figure 4 is the graph of the following situation between the actual flight position and the desired flight position without attitude information reconstruction;

[0018] Figure 5 is the graph of the following situation between the actual flight position and the desired flight position with the attitude information reconstruction module added; Detailed Embodiment

[0019] The flow chart of the attitude information reconstruction algorithm of the present invention is as Figure 3 shown. The attitude information provided by the flight control processing module and the vision-inertial information fusion module is usually represented in the form of quaternions. To convert this information into pitch angle, roll angle, and yaw angle suitable for flight control, the quaternions need to be converted into Euler angles. The specific steps are as follows: First, obtain the quaternion data q fc = (q 0fc , q 1fc , q 2fc , q 3fc ) based on the inertial sensors from the flight control processing module, and obtain the quaternion data q vins = (q 0vins , q 1vins , q2vins , q 3vins ). Next, using the conversion formula from quaternion to Euler angles, the quaternion output by the flight control processing module is converted into the pitch angle θ fc and the roll angle φ fc , as well as the yaw angle ψ of the external visual inertial information fusion module vins . The conversion formulas are as follows: φ fc = arcsin[2(q 0fc q 2fc + q 1fc q 3fc )], These formulas can extract the elements in the quaternion as Euler angles, obtaining the pitch angle and roll angle provided by the flight control processing module, as well as the yaw angle provided by the external visual inertial information fusion module. By decomposing this data, it provides a reliable basis for subsequent attitude information reconstruction.

[0020] When performing the conversion from quaternion to Euler angles, to avoid sign ambiguities, the signs of the quaternions are normalized so that the scalar part q0 of the quaternion is always non - negative, i.e., q0 ≥ 0. If q0 < 0, then all components q0, q1, q2, q3 of the quaternion are inverted to ensure a unified sign standard, thus avoiding the situation where the same rotation has multiple Euler angle representations. The sign normalization of the quaternion is achieved by checking the scalar part q0 of the quaternion. If q0 < 0, then all components q0, q1, q2, q3 of the quaternion are inverted to ensure their signs are consistent, obtaining a standardized quaternion, and then accurate Euler angle calculations can be performed.

[0021] After obtaining the preliminary attitude information provided by the flight control processing module and the external visual inertial information fusion module, the reconstructor uses a weighted fusion algorithm to accurately fuse this data. This algorithm generates the final attitude information by weighted summation based on the noise characteristics and measurement accuracies of each sensor. The specific steps are as follows: First, weight coefficients are set. These weight coefficients are obtained through experimental calibration and are dynamically adjusted based on the performance of different sensors. The weight coefficients ω1, ω2, ω3, ω4, ω5, ω6 are used to adjust the contribution degrees of the yaw angle, pitch angle, and roll angle provided by the flight controller and the external system in the final reconstruction result. Then, the weighted fusion algorithm is used to perform weighted calculations on the attitude information, and the formula is as follows: The new attitude information is used for the flight control module.

[0022] To illustrate the effect of the attitude information reconstruction system, Figure 4 , Figure 5The comparison of the actual flight position and the desired flight position of the Q250 frame unmanned aerial vehicle (UAV) before and after attitude information reconstruction is given, where Figure 4 It takes off and hovers at the coordinates of 0m, 0m, and 1m. Before attitude information reconstruction, there are drifts in the pitch angle and roll angle, and it can be seen from the actual position changes in the x and y directions during the takeoff phase that the aircraft cannot take off vertically. After calculation, the standard deviation in the x direction during the hovering process is 8.53 cm, 12.35 cm in the y direction, and 18.24 cm in the z direction, and there are offset amounts in the x and y directions. Figure 5 It takes off and hovers at the coordinates of 0m, 0m, and 0.5m for 40 seconds, then flies to the coordinates of 1.7m, -0.5m, and 0.7m for 20 seconds and then returns. After attitude information reconstruction, the errors in the x and y directions during the takeoff phase are very small, realizing vertical takeoff, and the flight accuracy is significantly improved. After calculation, the standard deviation in the x direction for the whole process is 3.56 cm, 5.67 cm in the y direction, and 4.88 cm in the z direction, and the standard deviation of the combined position error in the three directions is only 8.28 cm.

[0023] In the example, the attitude information reconstruction module of the present invention uses Intel's UNC as the on-board computer and is implemented through C++ language software programming. In practical applications, NVIDIA computing boards, FPGAs, etc. can also be used. The flight control module uses PD position loop control, and other applicable flight control methods can also be used.

[0024] The parts not elaborated in detail in the present invention belong to the prior art well-known to professionals in the field.

Claims

1. A UAV flight control system based on attitude information reconstruction, characterized in that: The system comprises the following modules: an attitude information reconstruction module (1), a binocular camera sensing module (2), a flight control processing module (3), a visual inertial information fusion module (4), a flight control module (5), a flight execution module (6), and a communication module (7), wherein the binocular camera sensing module (2) and the flight control processing module (3) provide visual data and gyroscope accelerometer data for the visual inertial information fusion module (4) to perform fusion processing to generate visual inertial fusion attitude information; the flight control processing module (3) provides pitch angle and roll angle information based on inertial sensors for the attitude information reconstruction module (1). The attitude information is reconstructed; the attitude information reconstruction module (1) receives the attitude information provided by the flight control processing module (3) and the visual inertial information fusion module (4), reconstructs the pitch angle, roll angle and yaw angle through a weighted fusion algorithm, and outputs accurate attitude information to the flight control module (5) to control the flight of the UAV; the flight control module (5) generates flight control instructions according to the requirements of the flight mission, and transmits them to the flight execution module (6); the communication module (7) ensures efficient data exchange between the modules, maintains synchronization and coordination between the modules of the system, and ensures efficient operation of the UAV control system.

2. The UAV flight control system based on attitude information reconstruction according to claim 1 is characterized in that: The attitude information reconstruction module (4) is composed of an attitude information decomposer (8) and an attitude information reconstructor (9), wherein the attitude information decomposer (8) is used to extract and decompose attitude information from the flight control processing module (2) and the visual inertial information fusion module (3); and the attitude information reconstructor (9) is used to perform fusion calculation on the decomposed attitude information and output the reconstructed attitude information to the flight control module (5).

3. The UAV flight control system based on attitude information reconstruction according to claim 2 is characterized in that: The attitude information decomposer (8) is used to extract and decompose the pitch angle and roll angle information in the flight control processing module (2) and the yaw angle information from the visual inertial information fusion module (3); The attitude information reconstructor (9) comprises the following steps: quaternion to Euler angle conversion: converting the quaternion (q0, q1, q2, q3) from the visual inertial information fusion module (3) into Euler angle representation, and extracting the yaw angle ψ; flight control module quaternion to Euler angle conversion: converting the quaternion (q0, q1, q2, q3) in the flight control processing module (2) into Euler angle representation, and extracting the pitch angle θ and the roll angle φ; weighted fusion: using a weighted fusion algorithm to fuse the yaw angle, the pitch angle and the roll angle into new attitude information, specifically: Among them, ω1, ω2, ω3, ω4, ω5, ω6 are weight parameters set according to the sensor noise characteristics and calibration parameters. are the values ​​of the initial yaw angle, pitch angle and roll angle obtained from the visual inertial information fusion module (3) and the flight control processing module (2).

4. The UAV flight control system based on attitude information reconstruction according to claim 3 is characterized in that: The weighted fusion algorithm sets weight parameters according to the sensor noise characteristics in the flight control processing module (2) and the visual inertial information fusion module (3). The weight parameters can be obtained through experimental calibration, and the calibration process is as follows: using a static calibration method to obtain the sensor noise in the flight control processing module (2) and the visual inertial information fusion module (3); by comparing the experimental data with the actual flight trajectory, adjusting the weight parameters ω1, ω2, ω3, ω4, ω5, ω6, so that the reconstructed attitude information error is minimized.

5. The UAV flight control system based on attitude information reconstruction according to claim 3 is characterized in that: The flight control processing module (2) and the visual inertial information fusion module (3) are based on the attitude information represented by the quaternion output by the inertial sensor and the visual sensor respectively, through the quaternion conversion formula: in, is the rotation angle, v is the unit rotation axis, q=(q0,q1,q2,q3) is the quaternion, and the quaternion is converted into Euler angles ψ, θ, φ using the following formula: φ=arcsin[2(q0q2+q1q3)], where q0, q1, q2, q3 are the quaternion components in the flight control processing module (2) and the visual inertial information fusion module (3), and ψ, θ, φ are the reorganized new attitudes used for flight control.

Images