Unmanned aerial vehicle docking cone target detection and positioning method based on laser imaging radar

Abstract

This invention relates to the field of target detection and localization technology, and discloses a method for detecting and localizing UAV docking cone targets based on laser imaging radar. The method includes: coordinate system calibration: calibrating the internal parameters of the laser imaging radar based on a ring target; calculating the pose of the ring target in the laser imaging radar coordinate system using a direct linear transformation method; optimizing the pose using bundle adjustment; setting the UAV and ring target coordinate systems to a translation-only relationship; calculating the rotation and translation matrices of the laser imaging radar and UAV coordinate systems to establish their pose relationship; target detection: acquiring intensity and distance images of the cone target using the laser imaging radar; combining this with a target detection algorithm to train a detection model in a complex environment; pose calculation: calculating the centroid coordinates of the cone target and outputting the position information of the cone target in the UAV coordinate system. This invention can improve the detection and localization capabilities of UAVs docking cone targets.

Description

Technical Field

[0001] This invention relates to the field of target detection and positioning technology, and in particular to a method for UAV docking cone target detection and positioning based on laser imaging radar. Background Technology

[0002] With the rapid development of drone technology, its application in many fields is becoming increasingly widespread. However, during the recovery process, the autonomous and precise docking of drones with conical targets faces significant challenges. Especially in complex environments (such as low light, strong light, and obstructions), achieving high-precision, real-time docking presents an even greater challenge.

[0003] Traditional imaging systems, such as visible light and infrared imaging, as well as triangulation ranging techniques, can only provide two-dimensional images that can be directly recognized by the human eye. However, when processing these images, it is impossible to obtain the spatial relationships between the image content. This is especially problematic for UAV docking based on laser imaging radar, which requires stable tracking of the target, centroid marking, and 3D attitude inversion. Using traditional imaging systems and triangulation ranging techniques often necessitates additional algorithms to process the image for subsequent recognition, significantly increasing computational complexity and making the task extremely difficult. In particular, for triangulation ranging, the unavoidable shadow effect creates a significant visual blind spot in target ranging, severely reducing the system's alignment accuracy. Summary of the Invention

[0004] To address the aforementioned issues, this invention proposes a method for detecting and locating UAV docking cone targets based on laser imaging radar, which can improve the detection and location capabilities of UAVs and cone targets during aerial docking.

[0005] The technical solution adopted in this invention is as follows:

[0006] A method for UAV docking cone target detection and localization based on laser imaging radar includes:

[0007] Coordinate system calibration: The internal parameters of the laser imaging radar are calibrated based on the ring target. The pose of the ring target in the laser imaging radar coordinate system is calculated by the direct linear transformation method. After optimization by the bundle adjustment method, the coordinate system of the UAV and the ring target is set to a translation-only relationship. Combining GPS and UAV size data, the rotation and translation matrices of the coordinate system of the laser imaging radar and the UAV are calculated to form the pose relationship between the two.

[0008] Target detection: Intensity and distance images of the cone-shaped target are acquired by laser imaging radar, and target detection algorithms are combined to train the model in a complex environment to generate a cone-shaped target detection model;

[0009] Pose calculation: Combining the calibrated internal parameters of the laser imaging radar with the intensity and range images of the cone-shaped target, a 3D point cloud of the cone-shaped target in the laser imaging radar coordinate system is obtained; after transformation to the UAV coordinate system, it is registered with the sampled point cloud of the cone-shaped target detection model through the consistency point drift method; the centroid coordinates of the cone-shaped target are calculated, and the position information of the cone-shaped target in the UAV coordinate system is output.

[0010] Furthermore, in the coordinate system calibration, the internal parameters of the laser imaging radar are calibrated based on the ring target, including: taking pictures of the ring target with reflective markings by the laser imaging radar to obtain ring target images at different angles, and detecting the ring target in the ring target images. The detection methods include sub-pixel contour detection, ellipse fitting and clustering methods.

[0011] Furthermore, the coordinate system calibration, which calibrates the internal parameters of the laser imaging radar based on the ring target, also includes: after detecting the ring target, calibrating the internal parameters of the laser imaging radar. The calibration process involves three coordinate systems: the laser imaging radar, the normalized plane, and the world coordinate system, and the transformation relationship between the three needs to be solved.

[0012] Furthermore, the coordinate system calibration, which calibrates the internal parameters of the laser imaging radar based on the ring target, also includes: after detecting the ring target, calculating the minimum bounding rectangle containing the center points based on the image coordinates of all center points in the ring target; and expanding the size of the minimum bounding rectangle to a set multiple so that the corresponding ring target can still be found in the area when the target moves, thereby realizing ring target tracking.

[0013] Furthermore, in the coordinate system calibration, setting the UAV and the ring target coordinate system to a translation-only relationship includes: keeping the UAV coordinate system and the ring target coordinate system in the same spatial position, so that there is only a translation relationship between the two coordinate systems and no change in angle.

[0014] Furthermore, in the coordinate system calibration, the rotation and translation matrices of the laser imaging radar and the UAV coordinate system are calculated by combining GPS and UAV size data to establish their pose relationship, including:

[0015] Calculate the rotation matrix of the laser imaging radar relative to the ring target, and then obtain the rotation matrix of the laser imaging radar relative to the UAV coordinate system;

[0016] Calculate the translation matrix between the laser imaging radar and the UAV coordinate system by combining GPS positioning information and UAV size data;

[0017] The rotation matrix and translation matrix together constitute the pose relationship between the UAV and the laser imaging radar coordinate system, thus completing the calibration of the laser imaging radar and the UAV coordinate system.

[0018] Furthermore, in the pose calculation, by combining the calibrated internal parameters of the laser imaging radar and the intensity and range images of the cone-shaped target, a three-dimensional point cloud of the cone-shaped target in the laser imaging radar coordinate system is obtained, including:

[0019] The target, a cone-shaped object, is captured by a laser imaging radar to obtain distance and intensity images;

[0020] Based on the image information of the cone-shaped target and the internal parameters of the laser imaging radar, the three-dimensional point cloud data of the cone-shaped target in the laser imaging radar coordinate system is obtained:

[0021]

[0022] in, This represents the three-dimensional spatial coordinates of the cone-shaped target in the laser imaging radar coordinate system; This represents the depth information of the point cloud obtained from the distance image; Represents pixel coordinates; c x The x-coordinate of the intersection point of the optical axis and the image plane. c y The ordinate of the point where the optical axis intersects the image plane; f x and f y This represents the internal parameters of the laser imaging radar.

[0023] Furthermore, in the pose calculation, based on the calibration relationship between the laser imaging radar and the UAV coordinate system, the 3D point cloud of the cone-shaped target in the laser imaging radar coordinate system is transformed to the UAV coordinate system. The transformation method includes:

[0024]

[0025] in, This represents the three-dimensional coordinates of the cone-shaped target's three-dimensional point cloud in the UAV coordinate system; It is the rotation matrix of the laser imaging radar to the UAV coordinate system. It is the translation matrix of the laser imaging radar to the UAV coordinate system; This represents the three-dimensional spatial coordinates of the cone-shaped target in the laser imaging radar coordinate system.

[0026] Furthermore, in the pose calculation, the registration of the sampled point cloud of the cone-shaped target detection model with the consistency point drift method includes: registering the sampled point cloud of the cone-shaped target detection model with the consistency point drift method to obtain real cone-shaped target point cloud data that meets the constraints.

[0027] Furthermore, in the pose calculation, the coordinates of the centroid of the cone-shaped target are calculated, and the position information of the cone-shaped target in the UAV coordinate system is output, including:

[0028]

[0029] in, This represents the centroid coordinates of the cone-shaped target in the UAV coordinate system. This represents the distance to the cone-shaped target in the UAV coordinate system. This represents the azimuth deviation of the cone-shaped target in the UAV coordinate system. The pitch deviation of the cone-shaped target in the UAV coordinate system; This represents the three-dimensional spatial coordinates of the cone-shaped target in the laser imaging radar coordinate system; N is the number of samples in the point cloud data.

[0030] The beneficial effects of this invention are as follows:

[0031] Compared to traditional methods for detecting and locating conical targets, this invention utilizes laser imaging radar to acquire depth information from a three-dimensional scene. This provides not only two-dimensional grayscale information of the target but also more detailed three-dimensional information: specifically, the distance of the spatial target from the Time-of-Flight (TOF) measurement system. This depth information enables computers to easily perform operations that are more difficult than those based on two-dimensional images, thereby improving the UAV's ability to detect and locate conical targets during aerial docking. Attached Figure Description

[0032] Figure 1 A flowchart illustrating the process of calculating the pose of the cone sleeve target.

[0033] Figure 2 This is a schematic diagram of the coordinate system calibration between a laser imaging radar and an unmanned aerial vehicle (UAV).

[0034] Figure 3 This is a schematic diagram of a laser imaging radar, a normalized plane, and a world coordinate system.

[0035] Figure 4 This is a schematic diagram illustrating the acquisition of the pose information of the cone-shaped target in the UAV coordinate system. Detailed Implementation

[0036] To provide a clearer understanding of the technical features, objectives, and effects of the present invention, specific embodiments are now described. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention; that is, the described embodiments are only a part of the embodiments of the invention, not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0037] Example 1

[0038] The autonomous and precise docking of UAVs with target cones during recovery presents significant challenges. This is especially true in complex environments (such as low light, strong light, and obstructions), where achieving high-precision, real-time docking is even more difficult. Using traditional imaging systems and triangulation ranging techniques often requires additional algorithms to process the images for subsequent identification, significantly increasing computational complexity and making the task extremely difficult. In particular, for triangulation ranging, the unavoidable shadow effect creates severe visual blind spots in target ranging, severely reducing the system's alignment accuracy.

[0039] In contrast, lidar imaging radar, also known as a Time-of-Flight (TOF) depth camera, acquires depth information of a three-dimensional scene by using a CCD / CMOS imaging array combined with active infrared laser modulation technology. This not only provides two-dimensional grayscale information about the target, but more importantly, it provides more detailed three-dimensional information: the distance of the spatial target from the TOF measurement system. This depth information allows computers to easily perform operations that are more difficult with two-dimensional images, such as target segmentation, recognition, tracking, and feature point marking, making it particularly suitable for high-precision, real-time UAV docking missions.

[0040] Based on this, this embodiment provides a method for UAV docking cone target detection and positioning based on laser imaging radar, including:

[0041] Coordinate system calibration: The internal parameters of the laser imaging radar are calibrated based on the ring target. The pose of the ring target in the laser imaging radar coordinate system is calculated by the direct linear transformation method. After optimization by the bundle adjustment method, the coordinate system of the UAV and the ring target is set to a translation-only relationship. Combining GPS and UAV size data, the rotation and translation matrices of the coordinate system of the laser imaging radar and the UAV are calculated to form the pose relationship between the two.

[0042] Target detection: Intensity and distance images of the cone-shaped target are acquired by laser imaging radar. The target detection algorithm is then used to train the model in complex environments (such as low light, strong light, occlusion, etc.) to generate a cone-shaped target detection model.

[0043] Pose calculation: Combining the calibrated internal parameters of the laser imaging radar with the intensity and range images of the cone-shaped target, a 3D point cloud of the cone-shaped target in the laser imaging radar coordinate system is obtained; after transformation to the UAV coordinate system, it is registered with the sampled point cloud of the cone-shaped target detection model through the consistency point drift method; the centroid coordinates of the cone-shaped target are calculated, and the position information of the cone-shaped target in the UAV coordinate system is output.

[0044] Specifically, the method of this embodiment can be implemented by the following steps:

[0045] (1) The internal parameters of the laser imaging radar were calibrated by combining Zhang Zhengyou's planar calibration method with a ring target.

[0046] (2) Place the UAV coordinate system and the ring target coordinate system in the same spatial position so that there is only a translation relationship between the two coordinate systems and no change in angle.

[0047] (3) Calculate the pose relationship between the circular target and the laser imaging radar coordinate system, and combine GPS data and UAV size information to calibrate the pose relationship between the laser imaging radar and the UAV.

[0048] (4) A large number of intensity and distance images of the cone-shaped targets were acquired by laser imaging radar and divided into two parts: training dataset and validation dataset. The cone-shaped targets in the training dataset were trained using a target detection algorithm (e.g., YoloV11) and a training model was generated. Then, the cone-shaped targets in the validation dataset were detected by the detection network to extract the pose data of the cone-shaped targets.

[0049] (5) Intensity image and distance image data are fused, and noise interference such as reflections is suppressed through operations such as adaptive image enhancement and adaptive threshold segmentation to adapt to complex scenes. Training and prediction are performed based on occlusion model data to ensure high positioning accuracy in partially occluded scenes.

[0050] (6) Using the target image of the cone and the calibration parameters of the laser imaging radar, calculate the three-dimensional spatial point cloud data of the cone in the coordinate system of the laser imaging radar. Combine the pose relationship between the UAV and the coordinate system of the laser imaging radar, and transform the three-dimensional spatial point cloud of the cone to the coordinate system of the UAV.

[0051] (7) Use CPD technology to register the point cloud of the cone target in the UAV coordinate system with the random sampled point cloud data of the actual model. After registration, calculate the coordinates of the centroid of the cone target in the UAV coordinate system and output the position information (distance, azimuth deviation, pitch deviation) of the centroid of the cone target in the UAV coordinate system.

[0052] Example 2

[0053] This embodiment is based on embodiment 1:

[0054] This embodiment provides a method for UAV docking cone target detection and localization based on laser imaging radar, including laser imaging radar and UAV coordinate system calibration, target detection, and UAV pose calculation steps, as detailed below. Figure 1 As shown.

[0055] Preferably, the coordinate system calibration between the laser imaging radar and the UAV includes the following steps:

[0056] (1) such as Figure 2 As shown, a ring target with reflective markings was captured using a laser imaging radar to obtain images of the ring target from different angles; subpixel contour detection, ellipse fitting, and clustering methods were used to detect the ring target in the images. Figure 2 In the diagram, (x, y, z) represents the three-dimensional coordinates of the cone-shaped target in the UAV's coordinate system. L ,Y L Z L (X) represents the three-dimensional coordinates of the cone-shaped target in the lidar coordinate system. C ,Y C Z C ) represents the three-dimensional coordinates of the cone-shaped target in the circular target coordinate system.

[0057] (2) After detecting the ring target in the image, the internal parameters of the laser imaging radar are calibrated. The calibration process involves three coordinate systems: the laser imaging radar, the normalized plane, and the world coordinate system. Their relationship is as follows: Figure 3 As shown. The calibration process for laser imaging radar requires solving the transformation relationship between these three elements. For any point in three-dimensional space... Projecting this onto a two-dimensional image plane, where P'(x,y) are the pixel coordinates of the two-dimensional image plane, according to the coordinate transformation relationship, we have:

[0058]

[0059] in, Figure 3 middle The coordinates of point P in the laser imaging radar coordinate system are C (X) C ,Y C Z C The coordinates of point P in the normalized plane coordinate system are P g And P g =(Xc / Zc,Yc / Zc,1); Let (u0, v0) be the pixel coordinates, (u0, v0) be the center pixel, and s represent the non-perpendicularity factor between the u-axis and v-axis. This indicates the focal length of the laser imaging radar (in mm). and These represent the phase element dimensions in the x and y directions, respectively. , The unit is pixels. , These are the rotation and translation matrices, respectively. In the above equation, ... , The matrix composed of these parameters is the internal parameter matrix of the laser imaging radar, abbreviated as: ;Depend on , The matrix formed by equal parameters is called the extrinsic parameter, abbreviated as . Therefore, the above formula can be written as:

[0060]

[0061] in, This represents the rotation vector matrix. This represents the final projection matrix from a 3D point in the world coordinate system to the pixel plane of a 2D image.

[0062] During the manufacturing process or prolonged use, the imaging plane of a laser imaging radar may not be perfectly parallel to the lens, resulting in image distortion. Therefore, distortion correction is necessary. The relationship between image points before and after correction is as follows:

[0063]

[0064] in, and These represent image points on the normalized plane before and after distortion correction, respectively. Indicates radial distortion; Indicates tangential distortion; All of these are constants. The calculated distortion-corrected planar coordinates need to be converted into two-dimensional image pixel plane points to achieve image distortion correction, i.e.:

[0065]

[0066] Based on Zhang Zhengyou's calibration method, multiple images are taken of the circular target at different positions to extract elliptical points and obtain the internal parameters of the laser imaging radar. Then, maximum likelihood estimation is used to solve for the distortion parameters of the laser imaging radar and optimize the internal and external parameters. The optimization method is as follows:

[0067]

[0068] in, This indicates the internal parameters of the laser imaging radar; These are the distortion parameters to be solved and optimized; It is the first The transformation relationship between the images in the world coordinate system and the laser imaging radar coordinate system; Indicates the first The first calibration plate image The actual pixel coordinates of the center point; This represents the coordinates of the reprojection point obtained through internal, external, and distortion parameters.

[0069] (3) After detecting the ring target, calculate the minimum bounding rectangle containing its center point based on the image coordinates of all center points in the ring target. Increase the size of the minimum bounding rectangle by a certain factor so that a suitable ring target can still be found in the area when the target moves, thus realizing the tracking of the ring target.

[0070] (4) Based on the size information of the ring target, the pose information of the ring target in the laser imaging radar coordinate system is calculated using the Direct Linear Transform (DLT) algorithm, and then the rotation and translation matrix is ​​optimized using the Bundle Adjustment optimization algorithm (i.e., the bundle adjustment method).

[0071] (5) Keep the UAV coordinate system and the ring target coordinate system in the same spatial position so that there is only a translation relationship between the two coordinate systems and no change in angle.

[0072] (6) Calculate the rotation matrix of the laser imaging radar relative to the ring target, and then obtain the rotation matrix of the laser imaging radar relative to the UAV coordinate system. Calculate the translation matrix between the laser imaging radar and the UAV coordinate system using GPS positioning information and UAV size data. The rotation matrix and translation matrix together constitute the pose relationship between the UAV and the laser imaging radar coordinate system, completing the calibration of the laser imaging radar and UAV coordinate system.

[0073] Preferably, the target detection in this embodiment includes the following steps:

[0074] (1) The laser imaging radar emits laser pulses (e.g., 850nm wavelength infrared laser pulses) toward the cone target, measures the phase difference between the reflected light and the emitted light, converts the phase error into a distance value, and generates a distance map and a grayscale map.

[0075] (2) Collect a large amount of cone-shaped target image data, use image labeling tools to annotate the image dataset, and convert the labeled files into training files in a specific format. Use a deep learning algorithm for target detection (e.g., the YOLOv11 model) to configure the cone-shaped target detection training environment. After the environment is configured, train the model to obtain a model that can be used for target detection. After training, use the optimal model to verify cone-shaped target detection in complex environments (e.g., low light, strong light, occlusion, etc.).

[0076] Preferably, the UAV pose calculation in this embodiment includes the following steps:

[0077] (1) In order to obtain the three-dimensional coordinates of the centroid of the cone-shaped target in the UAV coordinate system, it is necessary to sample and acquire the point cloud data of the cone-shaped target, such as Figure 4 As shown. Figure 4In the diagram, P1 and Pn represent the 1st and nth point cloud data points of the cone-shaped target in the lidar coordinate system, respectively, and Po represents the centroid of these n points. By acquiring range and intensity images of the cone-shaped target through lidar imaging, and based on the target image information and the lidar's internal parameters, the 3D point cloud data of the cone-shaped target in the lidar coordinate system can be obtained. The calculation process is as follows:

[0078]

[0079] in, This represents the three-dimensional spatial coordinates of the cone-shaped target in the laser imaging radar coordinate system; This represents the depth information of the point cloud obtained from the distance image; Represents pixel coordinates; c x The x-coordinate of the intersection point of the optical axis and the image plane. c y The ordinate of the point where the optical axis intersects the image plane; f x and f y This represents the internal parameters of the laser imaging radar.

[0080] (2) Based on the calibration relationship between the laser imaging radar and the UAV coordinate system, the three-dimensional point cloud of the cone-shaped target in the laser imaging radar coordinate system can be transformed to the UAV coordinate system. The transformation method is as follows:

[0081]

[0082] in, This represents the three-dimensional coordinates of the cone-shaped target's three-dimensional point cloud in the UAV coordinate system. and These are the rotation and translation matrices of the laser imaging radar to the UAV coordinate system.

[0083] (3) The three-dimensional point cloud data of the cone-shaped target obtained after conversion may contain a lot of noisy data. At this time, the CPD matching algorithm (consistent point drift method) is used to register it with the actual random sampling point cloud data of the cone-shaped target in order to obtain the real point cloud data of the cone-shaped target that meets the constraints.

[0084] (4) By using the registered point cloud data to calculate the centroid of the cone-shaped target, the real-time pose information of the cone-shaped target in the UAV coordinate system can be obtained. The centroid is calculated as follows:

[0085]

[0086] in, This represents the centroid coordinates of the cone-shaped target in the UAV coordinate system, where N is the number of point cloud data samples. At this point, the centroid coordinates of the cone-shaped target are in the UAV coordinate system. The distance (along the UAV's longitudinal axis), azimuth deviation (horizontal offset), and pitch deviation (vertical offset) of the cone-shaped target in the UAV coordinate system are respectively... , , .

[0087] The above description is merely a preferred embodiment of the present invention. It should be understood that the present invention is not limited to the forms disclosed herein and should not be construed as excluding other embodiments. It can be used in various other combinations, modifications, and environments, and can be altered within the scope of the concept described herein through the above teachings or related technologies or knowledge. Modifications and variations made by those skilled in the art that do not depart from the spirit and scope of the present invention should be within the protection scope of the appended claims.

[0088] It should be noted that, for the sake of simplicity, the foregoing method embodiments are described as a series of actions. However, those skilled in the art should understand that this application is not limited to the described order of actions, as some steps may be performed in other orders or simultaneously according to this application. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are preferred embodiments, and the actions and modules involved are not necessarily essential to this application.

Claims

1. A method for UAV docking cone target detection and positioning based on laser imaging radar, characterized in that, include: Coordinate system calibration: The internal parameters of the laser imaging radar are calibrated based on the ring target. The pose of the ring target in the laser imaging radar coordinate system is calculated by the direct linear transformation method. After optimization by the bundle adjustment method, the coordinate system of the UAV and the ring target is set to a translation-only relationship. Combining GPS and UAV size data, the rotation and translation matrices of the coordinate system of the laser imaging radar and the UAV are calculated to form the pose relationship between the two. Target detection: Intensity and distance images of the cone-shaped target are acquired by laser imaging radar, and target detection algorithms are combined to train the model in a complex environment to generate a cone-shaped target detection model; Pose calculation: Combining the calibrated internal parameters of the laser imaging radar and the intensity and range images of the cone-shaped target, a 3D point cloud of the cone-shaped target in the laser imaging radar coordinate system is obtained; after being transformed to the UAV coordinate system, it is registered with the sampled point cloud of the cone-shaped target detection model through the consistency point drift method; Calculate the centroid coordinates of the cone-shaped target and output the position information of the cone-shaped target in the UAV coordinate system.

2. The method for UAV docking cone target detection and positioning based on laser imaging radar according to claim 1, characterized in that, In the coordinate system calibration, the internal parameters of the laser imaging radar are calibrated based on the ring target, including: taking pictures of the ring target with reflective markings by the laser imaging radar to obtain ring target images at different angles, and detecting the ring target in the ring target images. The detection methods include sub-pixel contour detection, ellipse fitting and clustering methods.

3. The method for UAV docking cone target detection and positioning based on laser imaging radar according to claim 2, characterized in that, The coordinate system calibration, which calibrates the internal parameters of the laser imaging radar based on the ring target, also includes: after detecting the ring target, calibrating the internal parameters of the laser imaging radar. The calibration process involves three coordinate systems: the laser imaging radar, the normalized plane, and the world coordinate system, and the transformation relationship between the three needs to be solved.

4. The method for UAV docking cone target detection and positioning based on laser imaging radar according to claim 2, characterized in that, The coordinate system calibration, which calibrates the internal parameters of the laser imaging radar based on the ring target, also includes: after detecting the ring target, calculating the minimum bounding rectangle containing the center points based on the image coordinates of all the center points in the ring target; expanding the size of the minimum bounding rectangle to a set multiple so that the corresponding ring target can still be found in the area when the target moves, thereby realizing ring target tracking.

5. The method for UAV docking cone target detection and positioning based on laser imaging radar according to claim 1, characterized in that, In the coordinate system calibration, the coordinate system of the UAV and the circular target is set to a translation-only relationship, including: keeping the UAV coordinate system and the circular target coordinate system in the same spatial position, so that there is only a translation relationship between the two coordinate systems and no change in angle.

6. The method for UAV docking cone target detection and positioning based on laser imaging radar according to claim 1, characterized in that, In the coordinate system calibration, the rotation and translation matrices of the laser imaging radar and the UAV coordinate system are calculated by combining GPS and UAV size data to establish their pose relationship, including: Calculate the rotation matrix of the laser imaging radar relative to the ring target, and then obtain the rotation matrix of the laser imaging radar relative to the UAV coordinate system; Calculate the translation matrix between the laser imaging radar and the UAV coordinate system by combining GPS positioning information and UAV size data; The rotation matrix and translation matrix together constitute the pose relationship between the UAV and the laser imaging radar coordinate system, thus completing the calibration of the laser imaging radar and the UAV coordinate system.

7. A method for UAV docking cone target detection and positioning based on laser imaging radar according to claim 1, characterized in that, In the pose calculation, by combining the calibrated internal parameters of the laser imaging radar and the intensity and range images of the cone-shaped target, a three-dimensional point cloud of the cone-shaped target in the laser imaging radar coordinate system is obtained, including: The target, a cone-shaped object, is captured by a laser imaging radar to obtain distance and intensity images; Based on the image information of the cone-shaped target and the internal parameters of the laser imaging radar, the three-dimensional point cloud data of the cone-shaped target in the laser imaging radar coordinate system is obtained: in, This represents the three-dimensional spatial coordinates of the cone-shaped target in the laser imaging radar coordinate system; This represents the depth information of the point cloud obtained from the distance image; Represents pixel coordinates; c x The x-coordinate of the intersection point of the optical axis and the image plane. c y The ordinate of the point where the optical axis intersects the image plane; f x and f y This represents the internal parameters of the laser imaging radar.

8. A method for UAV docking cone target detection and positioning based on laser imaging radar according to claim 1, characterized in that, In the pose calculation, based on the calibration relationship between the laser imaging radar and the UAV coordinate system, the 3D point cloud of the cone-shaped target in the laser imaging radar coordinate system is transformed to the UAV coordinate system. The transformation method includes: in, This represents the three-dimensional coordinates of the cone-shaped target's three-dimensional point cloud in the UAV coordinate system; It is the rotation matrix of the laser imaging radar to the UAV coordinate system. It is the translation matrix of the laser imaging radar to the UAV coordinate system; This represents the three-dimensional spatial coordinates of the cone-shaped target in the laser imaging radar coordinate system.

9. A method for UAV docking cone target detection and positioning based on laser imaging radar according to claim 1, characterized in that, In the pose calculation, the point cloud of the cone target detection model is registered with the consistency point drift method, including: registering the point cloud of the cone target detection model with the consistency point drift method to obtain real point cloud data of the cone target that meets the constraints.

10. A method for UAV docking cone target detection and positioning based on laser imaging radar according to claim 1, characterized in that, In the pose calculation, the coordinates of the centroid of the cone-shaped target are calculated, and the position information of the cone-shaped target in the UAV coordinate system is output, including: in, This represents the centroid coordinates of the cone-shaped target in the UAV coordinate system. This represents the distance to the cone-shaped target in the UAV coordinate system. This represents the azimuth deviation of the cone-shaped target in the UAV coordinate system. The pitch deviation of the cone-shaped target in the UAV coordinate system; This represents the three-dimensional spatial coordinates of the cone-shaped target in the laser imaging radar coordinate system; N is the number of samples in the point cloud data.

Images