Unmanned aerial vehicle automatic battery changing method and system based on 3D visual guidance

Abstract

The invention belongs to the technical field of unmanned aerial vehicle battery replacement, and provides an unmanned aerial vehicle automatic battery replacement method and system based on 3D visual guidance, and the method comprises the steps: constructing a target feature template of each region of a battery compartment and a feed battery based on obtained standard image data; performing multi-stage pyramid matching on the obtained to-be-matched point cloud data and the constructed target feature template, and reconstructing a point cloud data set and standard image data subjected to multi-stage pyramid matching; calculating the pose deviation value of the feed battery, and combining the pose deviation value of the feed battery to obtain the actual position of the feed battery; the target position and posture information are updated according to the feed battery posture deviation value, a mechanical arm is controlled to move to the target position according to the updated posture information, and the feed battery is grabbed and placed at the designated position; and controlling the mechanical arm to grab the fully-stored battery and then return to the specified image data acquisition position, controlling the 3D camera to carry out secondary scanning, repeating the steps, and putting the fully-stored battery into the battery compartment of the unmanned aerial vehicle to complete automatic battery replacement operation.

Description

Technical Field

[0001] This invention belongs to the field of drone battery replacement technology, and particularly relates to an automatic battery swapping method and system for drones based on 3D vision guidance. Background Technology

[0002] The statements in this section are merely background information related to the present invention and do not necessarily constitute prior art.

[0003] Existing drone battery replacement methods require complex analysis to determine the position and attitude of the drone battery, and cannot obtain complete three-dimensional spatial coordinates and attitude information. Furthermore, existing 3D camera-based battery replacement methods require complex information acquisition. For example, when the 3D camera captures an image of the upper surface of the frame fixing the battery module, it also needs to capture a fixed-size reference object and its spatial positional relationship with the upper surface of the frame fixing the battery module. A proportional scaling method is used to calculate the spatial coordinates of the battery module and its frame. The outline of the overhead crane manipulator is captured to create a 3D model, and the spatial coordinates of the manipulator are analyzed and calculated using a 3D intelligent positioning system. These steps increase the complexity of the battery replacement operation and reduce its efficiency. Additionally, after the overhead crane manipulator grabs a fully charged battery, it moves directly to the spatial position where it grabbed the depleted battery for the replacement. For the drone, the slight displacement caused by the battery movement during the manipulator removal can lead to collisions when replacing a fully charged battery with the one that grabbed the depleted battery, affecting the safety and reliability of the battery replacement process. Summary of the Invention

[0004] To address at least one of the technical problems mentioned above, this invention provides a method and system for automatic battery swapping of unmanned aerial vehicles (UAVs) based on 3D vision guidance. This method improves battery swapping efficiency by simplifying the target location calculation method and enhances the safety and reliability of battery swapping by employing repeated scanning operations, thereby achieving the goal of automatic battery swapping for UAVs.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: A first aspect of the present invention provides an automatic battery swapping method for unmanned aerial vehicles (UAVs) based on 3D vision guidance, comprising the following steps: Step 1: Based on the acquired standard image data from multiple locations and angles, construct target feature templates for each area of ​​the battery compartment and the feeding battery; Step 2: Perform multi-level pyramid matching on the acquired point cloud data to be matched and the constructed target feature template, and reconstruct the point cloud dataset after multi-level pyramid matching and standard image data; Step 3: Calculate the battery pose deviation value based on the reconstructed point cloud dataset and standard image data, and combine the battery pose deviation value to obtain the actual position of the battery. Step 4: Update the target position and attitude information based on the battery pose deviation value, control the robotic arm to move to the target position according to the updated attitude information, and pick up the battery and place it in the designated position. Step 5: After the robotic arm picks up the fully charged battery, it returns to the designated location for image data acquisition. The 3D camera then performs a second scan. Steps 1-4 are repeated to place the fully charged battery into the drone's battery compartment and complete the automatic battery swapping operation.

[0006] Furthermore, the target feature templates for each area of ​​the battery compartment and the feeding battery include adaptive pyramid layer number, template block extremely strong binary, template gradient information, and adaptive angle rotation step size.

[0007] Furthermore, the step of performing multi-level pyramid matching on the acquired point cloud data to be matched and the constructed target feature template includes: The image to be matched and the target feature template are matched point-to-point at first level to obtain the first-level matching result. Based on the first-level matching results, the image to be matched and the target feature template are matched in the second-level contour matching to obtain the second-level matching results; Based on the second-level matching results, the image to be matched and the target feature template are subjected to third-level gradient matching to obtain the third-level matching results.

[0008] Furthermore, when performing first-level point-to-point matching between the image to be matched and the target feature template, the top-level search box of the image to be matched pyramid is changed from... Composed of 4 points, the number of contour points within the search box in the image to be matched is calculated using an integral image. ,in ,like If the value is less than the set threshold Q, then the first-level match is successful. To adjust the parameters.

[0009] Furthermore, when performing secondary contour matching on the image to be matched and the target feature template, the extracted block-level strong binary values ​​of the image to be matched and the template block-level strong binary values ​​in the target feature template are matched, and compared with the block-level strong binary detection mode of the template image at each rotation angle. The threshold for the number of inconsistent blocks is set as follows: If the number of inconsistent blocks is less than the threshold If so, the second-level match passes.

[0010] Furthermore, when performing three-level gradient matching between the image to be matched and the target feature template, the template gradient information in the target feature template is converted into a series of points. There is a gradient direction vector for each point. At a certain location in the target image The sub-image has a feature vector ,in and For the vector representation of features, a normalized similarity measure function is used. S As a criterion for whether the third-level matching is successful, if the value calculated by the pixel-level function is greater than the set value... This is considered a success, at a certain location in the target image. The sub-image has a feature vector ,in and For the vector representation of features, a normalized similarity measure function is used. S As a criterion for whether the third-level matching is successful, if the value calculated by the pixel-level function is greater than the set value... Then it is considered a success.

[0011] Furthermore, when calculating the pose deviation value of the battery feeder based on the reconstructed point cloud dataset and the standard dataset, the covariance matrix of the point cloud dataset and the standard dataset is obtained by constructing the matrix based on the reconstructed point cloud dataset and the standard dataset. When the minimum distance between the two point sets meets the set threshold, the rotation matrix and translation vector are obtained by using the SVD decomposition method. The standard position of the battery feeder is translated and rotated by combining the rotation matrix and the translation vector to obtain the actual position of the battery feeder.

[0012] A second aspect of the present invention provides an automatic battery swapping device for unmanned aerial vehicles based on 3D vision guidance, comprising: The battery gripping module is used to construct target feature templates for each region of the battery compartment and the battery itself based on standard image data acquired from multiple position angles. It then performs multi-level pyramid matching on the acquired point cloud data to be matched and the constructed target feature templates, and reconstructs the point cloud dataset after multi-level pyramid matching with the standard image data. Based on the reconstructed point cloud dataset and standard image data, it calculates the battery pose deviation value and uses this value to determine the battery's actual position. Finally, it updates the target position and attitude information according to the battery pose deviation value, controls the robotic arm to move to the target position according to the updated attitude information, and grips the battery, placing it at the designated location. The fully charged battery replacement module controls the robotic arm to grab a fully charged battery and return to the designated image data acquisition position. It then controls the 3D camera to perform a secondary scan, repeating the steps of the battery grabbing module to place the fully charged battery into the drone's battery compartment, thus completing the automatic battery swapping operation.

[0013] A third aspect of the present invention provides a computer-readable storage medium.

[0014] A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the above-described 3D vision-guided automatic battery swapping method for unmanned aerial vehicles.

[0015] A fourth aspect of the present invention provides a computer device.

[0016] A computer device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the steps of the above-described 3D vision-guided automatic battery swapping method for unmanned aerial vehicles.

[0017] Compared with the prior art, the beneficial effects of the present invention are: This invention employs 3D vision-guided technology, using a 3D camera to acquire high-precision point cloud data of a specific area of ​​the drone's battery compartment. This data is then matched with a preset feature template to accurately calculate the battery's position and attitude, improving battery swapping efficiency. It eliminates the need for fixed-size reference objects and the acquisition of 3D models from robotic arms, thus achieving precise calculation of battery position and attitude and solving the problem of low efficiency in existing technologies.

[0018] This invention, through secondary scanning during the robotic arm control process, further ensures the accuracy of the new battery installation position, effectively avoiding the impact problems that may occur in the prior art, and improving the safety and reliability of battery swapping.

[0019] Advantages of additional aspects of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0020] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an improper limitation of the invention.

[0021] Figure 1 This is a flowchart of an automatic battery swapping method for unmanned aerial vehicles based on 3D vision guidance provided in an embodiment of the present invention; Figure 2 This is a schematic diagram of the multi-level pyramid matching process provided in an embodiment of the present invention; Figure 3 This is a schematic diagram of the pose deviation correction process provided in the embodiment of the present invention. Detailed Implementation

[0022] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0023] It should be noted that the following detailed description is illustrative and intended to provide further explanation of the invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0024] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the scope of exemplary embodiments according to the invention. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Furthermore, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0025] Example 1 like Figure 1 As shown, this embodiment provides an automatic battery swapping method for drones based on 3D vision guidance, including the following steps: Step 1: Based on the acquired image data from multiple standard position angles, construct target feature templates for each area of ​​the battery compartment and the feeding battery; In this embodiment, constructing the target feature templates for each area of ​​the battery compartment and the feeding battery includes preprocessing the image, building an image pyramid, and extracting data features such as the coordinate information of the pyramid image contour points, the block-level strong binary detection mode, and the normalized gradient vector. Specifically, the steps include the following: Step 101: Acquire image data from multiple standard position angles; In this embodiment, an industrial camera is mounted on the robotic arm. Based on multiple preset standard position angles, the mounted camera is moved to a preset position, and the camera obtains a standard target image and retains the current position and posture information of the robotic arm.

[0026] Step 102: Preprocess the image data obtained from multiple standard position angles; In this embodiment, the acquired standard image is subjected to preprocessing operations such as grayscale conversion, grayscale stretching, blurring, and edge enhancement.

[0027] Step 103: Extract template feature parameter data based on the preprocessed image data, including adaptive pyramid layer number, template block strong binary, template gradient information and adaptive angle rotation step size; In this embodiment, the adaptive pyramid layer number is determined based on the contour point features of the standard image; the maximum pyramid layer number of the standard image in the initial stage is... Optimal number of layers M′ The calculation formula is: , in, and For the width and height of the model, For the set threshold number of contour points, For the first Number of points in the pyramid outline of the layer image For the (i+1)th, Number of points in the pyramid outline of the layer image.

[0028] When performing strong binary computation on template blocks, the standard image H is divided into... a × b Given image patches of a certain size, calculate the average pixel grayscale value for each patch. and the average pixel grayscale value of the standard image H c The calculation formula for block-level strong binary detection mode is: , When calculating template gradient information, the gradient direction, grayscale, or gradient value of the edge points is obtained. For standard images Perform surface fitting to calculate the coefficients of the second-order surface equation and determine the template gradient information, expressed as: , When calculating the adaptive angle rotation step size, the angle is determined based on the distance between the farthest edge point and the centroid. This controls the number of standard images searched during matching and is used to calculate the search angle step size for image rotation.

[0029] The adaptive angle step size calculation method based on the cosine theorem is as follows: , in, Indicates the centroid of the contour image. This represents the point farthest from the centroid of the contour image. Point representation The position of the point after rotation; Step 2: Perform multi-level pyramid matching on the acquired point cloud data to be matched and the constructed target feature template, and reconstruct the point cloud dataset after multi-level pyramid matching and standard image data; In this embodiment, the first-level matching is point-to-point matching, which serves as a coarse filter for the second and third-level matching, quickly eliminating obviously inconsistent regions and reducing subsequent computation. The second-level matching is contour matching, which is based on block-level strong binary matching in the target feature template. During matching, it is compared with the block-level strong binary detection mode of the template image at various rotation angles. The third-level matching is based on gradient matching in the target feature template, ultimately obtaining multi-level matching results.

[0030] like Figure 2As shown, the specific steps include the following: Step 201: Perform first-level point-to-point matching between the point cloud data to be matched and the target feature template; When constructing a pyramid image first-level matching operation, the top-level search box of the pyramid of the image to be matched is... Composed of 4 points, the number of contour points within the search box in the image to be matched is calculated using an integral image. ,in ,like If the value is less than the set threshold Q, then the first-level match is successful. To adjust the parameters.

[0031] Step 202: Based on the first-level point-to-point matching results, perform second-level contour matching between the image to be matched and the target feature template; Based on pixel blocks of a set size, such as 8×8, block-level strong binary values ​​are extracted from the image to be matched. Secondary contour matching is then performed between the extracted block-level strong binary values ​​of the image to be matched and the template block-level strong binary values ​​in the target feature template. Template images for each rotation angle are obtained according to an adaptive rotation step size, and compared with the block-level strong binary detection mode of the template image at each rotation angle. A threshold for the number of inconsistent blocks is set as follows. If the number of inconsistent blocks is less than the threshold If so, the second-level match passes.

[0032] Step 204: Based on the secondary contour matching results, perform tertiary gradient matching between the image to be matched and the target feature template; The template gradient information in the target feature template is converted into a series of points. There is a gradient direction vector for each point. ; At a certain location in the target image The sub-image has a feature vector Using a normalized similarity measure function S As a criterion for whether the third-level matching is successful, if the value calculated by the pixel-level function is greater than the set value... This is considered a success, and is represented as: , in, The gradient direction of the image to be matched. , It is a vector representation of the features.

[0033] Step 205: Obtain the point cloud dataset and standard dataset obtained through first-level point-to-point matching, second-level contour matching and third-level gradient matching, and reconstruct the point cloud dataset and standard dataset. Specifically, obtain the point cloud dataset that matches the two-dimensional image data. Standard dataset The centroid is calculated to remove the centered coordinates, that is, the coordinate systems of the source and target data point sets are reconstructed with the centroid as the origin of the three-dimensional coordinate system. ,in, The centroid of the standard point cloud dataset, For the centroid of the point cloud dataset, For the standard point cloud i The original three-dimensional coordinates of the points For the target point cloud i The original three-dimensional coordinates of each point.

[0034] Step 3: Calculate the battery pose deviation value based on the reconstructed point cloud dataset and the standard dataset, and combine the battery pose deviation value to obtain the actual position of the battery. like Figure 3 As shown, in this embodiment, the pose deviation value is calculated based on the iterative nearest point algorithm, using the source data point set. and target data point set Constructing a matrix based on The covariance matrices of datasets P and Q are obtained; when the minimum distance between two point sets satisfies a set threshold... At that time, the rotation matrix is ​​obtained by SVD decomposition. Translation vector Combine rotation matrix and translation vector to determine the standard position of the battery. Perform translation and rotation transformations to determine the actual position. .

[0035] Step 4: Update the target position and attitude information based on the battery pose deviation value, control the robotic arm to move to the target position according to the updated attitude information, and pick up the battery and place it in the designated position. Step 5: After the robotic arm picks up the fully charged battery, it returns to the designated location for image data acquisition. The 3D camera then performs a second scan. Steps 1-4 are repeated to place the fully charged battery into the drone's battery compartment and complete the automatic battery swapping operation.

[0036] After the robotic arm grasps the fully charged battery, it returns to the photo-taking position. The 3D camera performs a second scan of the drone's battery compartment to acquire new point cloud data. Steps 1-4 are repeated to identify the target area of ​​the battery compartment based on the second scanned point cloud data and calculate the placement posture deviation value of the fully charged battery. The target position and attitude information are updated according to the placement posture deviation value, and the robotic arm is controlled to place the fully charged battery into the drone's battery compartment according to the updated posture, completing the automatic battery swapping operation.

[0037] Example 2 This embodiment provides an automatic battery swapping system for drones based on 3D vision guidance, including: The battery gripping module is used to construct target feature templates for each region of the battery compartment and the battery itself based on standard image data acquired from multiple position angles. It then performs multi-level pyramid matching on the acquired point cloud data to be matched and the constructed target feature templates, and reconstructs the point cloud dataset after multi-level pyramid matching with the standard image data. Based on the reconstructed point cloud dataset and standard image data, it calculates the battery pose deviation value and uses this value to determine the battery's actual position. Finally, it updates the target position and attitude information according to the battery pose deviation value, controls the robotic arm to move to the target position according to the updated attitude information, and grips the battery, placing it at the designated location. The fully charged battery replacement module controls the robotic arm to grab a fully charged battery and return to the designated image data acquisition position. It then controls the 3D camera to perform a secondary scan, repeating the steps of the battery grabbing module to place the fully charged battery into the drone's battery compartment, thus completing the automatic battery swapping operation.

[0038] It should be noted that the specific implementation of the 3D vision-guided automatic battery swapping system for drones in this embodiment of the invention is similar to the specific implementation of the 3D vision-guided automatic battery swapping method for drones in this embodiment of the invention. For details, please refer to the description in the method section. To reduce redundancy, it will not be repeated here.

[0039] Example 3 This embodiment provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps in the above-described 3D vision-guided automatic battery swapping method for unmanned aerial vehicles.

[0040] Example 4 This embodiment provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, it implements the steps in the above-described automatic battery swapping method for UAVs based on 3D vision guidance.

[0041] Example 5 This embodiment provides a program product, which is a computer program product including a computer program. The computer program is characterized in that, when executed by a processor, it implements the steps in the above-described 3D vision-guided automatic battery swapping method for unmanned aerial vehicles.

[0042] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of hardware embodiments, software embodiments, or embodiments combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage and optical storage) containing computer-usable program code.

[0043] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0044] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0045] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0046] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. The storage medium can be a magnetic disk, optical disk, read-only memory (ROM), or random access memory (RAM), etc.

[0047] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for automatic battery replacement of a UAV based on 3D vision guidance, characterized in that, The method comprises the following steps: Step 1: based on the acquired standard image data of multiple position angles, constructing a target feature template of each region of the battery compartment and the feeding battery; Step 2: performing multi-level pyramid matching on the acquired to-be-matched point cloud data and the constructed target feature template, and reconstructing the point cloud data set and the standard image data after multi-level pyramid matching; Step 3: calculating a feeding battery pose deviation value based on the reconstructed point cloud data set and the standard image data, and obtaining the actual position of the feeding battery in combination with the feeding battery pose deviation value; Step 4: updating target position and attitude information according to the feeding battery pose deviation value, controlling the mechanical arm to move to the target position according to the updated attitude information, and placing the feeding battery at a specified position after grabbing the feeding battery; Step 5: controlling the mechanical arm to return to the image data specified acquisition position after grabbing the fully charged battery, controlling the 3D camera to perform secondary scanning, and repeating steps 1-4 to put the fully charged battery into the unmanned aerial vehicle battery compartment to complete the automatic battery replacement operation. 2.The 3D vision guidance based automatic battery swapping method for UAV according to claim 1, wherein, The target feature template of each region of the battery compartment and the feeding battery comprises adaptive pyramid layer number, template block extremely strong binary, template gradient information and adaptive angle rotation step. 3.The 3D vision guidance based automatic battery swapping method for UAV according to claim 1, wherein, The multi-level pyramid matching on the acquired to-be-matched point cloud data and the constructed target feature template comprises: performing one-level point-to-point matching on the to-be-matched image and the target feature template to obtain one-level matching results; performing two-level contour matching on the to-be-matched image and the target feature template based on the one-level matching results to obtain two-level matching results; performing three-level gradient matching on the to-be-matched image and the target feature template based on the two-level matching results to obtain three-level matching results. 4.The 3D vision guidance based automatic battery swapping method for UAV according to claim 3, wherein, When the to-be-matched image and the target feature template are subjected to one-stage point-to-point matching, the search frame of the top layer of the to-be-matched image pyramid is composed of 4 points, and the number of contour points in the search frame of the to-be-matched image is calculated by using an integral graph , wherein , if the value is less than a set threshold Q, one-stage matching is successful, wherein is an adjustment parameter. When the to-be-matched image and the target feature template are subjected to one-stage point-to-point matching, the search frame of the top layer of the to-be-matched image pyramid is composed of 4 points, and the number of contour points in the search frame of the to-be-matched image is calculated by using an integral graph , wherein , if the value is less than a set threshold Q, one-stage matching is successful, wherein 5.The 3D vision guidance based automatic battery swapping method for UAV according to claim 3, wherein, When the to-be-matched image and the target feature template are subjected to the second-level contour matching, the extracted block-level strong binary of the to-be-matched image and the template block-level strong binary in the target feature template are subjected to contour matching, and the block-level strong binary detection mode of the template image of each rotation angle is compared, a threshold value of the number of inconsistent blocks is set as , and if the number of inconsistent blocks is less than the threshold value , the second-level matching is passed. 6.The 3D vision guidance based automatic battery swapping method for UAV according to claim 3, wherein, When performing three-level gradient matching between the image to be matched and the target feature template, the template gradient information in the target feature template is converted into a series of points. There is a gradient direction vector for each point. At a certain location in the target image The sub-image has a feature vector ,in and For the vector representation of features, a normalized similarity measure function is used. S As a criterion for whether the third-level matching is successful, if the value calculated by the pixel-level function is greater than the set value... This is considered a success, at a certain location in the target image. The sub-image has a feature vector ,in and For the vector representation of features, a normalized similarity measure function is used. S As a criterion for whether the third-level matching is successful, if the value calculated by the pixel-level function is greater than the set value... Then it is considered a success.

7. The 3D vision guidance based automatic battery swapping method for UAV according to claim 1, wherein, When the minimum distance of the two point sets meets the set threshold value, a rotation matrix and a translation vector are obtained by SVD decomposition method, the standard position of the feeding battery is translated and rotated in combination with the rotation matrix and the translation vector, and the actual position of the feeding battery is obtained. The method comprises:

8. An unmanned aerial vehicle automatic battery changing system based on 3D vision guidance, characterized in that, a feeding battery grabbing module, which is used for constructing a target feature template of each region of the battery compartment and the feeding battery based on the acquired standard image data of multiple position angles, performing multi-level pyramid matching on the acquired to-be-matched point cloud data and the constructed target feature template, reconstructing the point cloud data set and the standard image data after multi-level pyramid matching, calculating a feeding battery pose deviation value based on the reconstructed point cloud data set and the standard image data, obtaining the actual position of the feeding battery in combination with the feeding battery pose deviation value, and updating target position and attitude information according to the feeding battery pose deviation value. a fully charged battery replacement module, which is used for controlling the mechanical arm to return to the image data specified acquisition position after grabbing the fully charged battery, controlling the 3D camera to perform secondary scanning, repeating the steps of the feeding battery grabbing module, and putting the fully charged battery into the unmanned aerial vehicle battery compartment to complete the automatic battery replacement operation. ​ 9. A computer readable storage medium having stored thereon a computer program, characterized in that, The program is executed by the processor to implement the steps in the 3D vision guidance-based automatic battery replacement method for a UAV according to any one of claims 1-7.

10. A computer device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, The processor executes the program to implement the steps in the 3D vision guidance-based automatic battery replacement method for a UAV according to any one of claims 1-7.

Images