A method and device for positioning agricultural machinery hangar based on point cloud registration

Abstract

This invention provides a method and apparatus for agricultural machinery hangar positioning based on point cloud registration, relating to the field of indoor navigation and positioning. The method includes: using an improved Scan Context++ algorithm to perform first-order coarse ring key matching and second-order sector key matching on real-time hangar point cloud frames acquired by an inclined-mounted agricultural machinery lidar and a priori 3D point cloud map, obtaining candidate keyframes and the rotation angles of the real-time hangar point cloud frames and candidate keyframe point clouds, and generating an initial registration matrix with a normal distribution transformation based on this. Using the initial registration matrix as the initial value, the collected real-time hangar positioning point cloud frame data is continuously registered with the priori 3D point cloud map to output the agricultural machinery pose, and re-initialization is performed when positioning is interrupted due to dynamic scenes. This invention enables indoor positioning initialization and continuous positioning functions without external information assistance other than lidar.

Description

Technical Field

[0001] This invention relates to the field of indoor navigation and positioning, specifically to a method and apparatus for locating agricultural machinery hangars based on point cloud registration. Background Technology

[0002] Precise positioning technology is crucial for realizing the functions and tasks of intelligent unmanned operation systems. In recent years, with the development of lidar technology, lidar point cloud registration has become an important method for unmanned system positioning. By comparing the set of points in three-dimensional space with a pre-established point cloud map, extracting the inherent structural features of the point cloud, and establishing the conversion relationship between real-time point cloud frames and the point cloud map, accurate estimation of the pose of the unmanned system can be achieved. This technology has advantages such as being unaffected by lighting conditions and being able to handle complex scenes, and has broad application prospects in the field of unmanned agricultural machinery hangar positioning. Addressing the problems of indoor GNSS satellite signal loss, dynamic changes in the hangar scene, and blurred images due to mechanical vibration during agricultural machinery positioning in hangars, this study proposes a point cloud registration-based method for agricultural machinery hangar positioning, which is of great significance for improving the automation and intelligence level of unmanned agricultural machinery operations.

[0003] Point cloud positioning technology relies on sensor characteristics, scene environment, and carrier positioning requirements. It requires comprehensive consideration of internal and external factors such as the scanning field of view of agricultural machinery lidar, hangar structural features, and the motion characteristics of the agricultural machinery carrier to design a reasonable and effective agricultural machinery hangar positioning method. Furthermore, the environment within the hangar changes during operation, such as the movement of agricultural implements and the relocation of stacked items, making the adaptability of point cloud positioning technology to environmental changes need improvement. Moreover, how to achieve effective initialization of positioning within the hangar without prior pose is also a key issue. Therefore, there is an urgent need to propose an agricultural machinery hangar positioning method and device based on point cloud registration to address the technical deficiencies in initial positioning, continuous positioning, and environmental adaptability within agricultural machinery hangars, thereby improving the reliability and robustness of agricultural machinery hangar positioning. Summary of the Invention

[0004] To address the aforementioned technical problems, this invention provides a method and apparatus for agricultural machinery hangar positioning based on point cloud registration. It offers an indoor positioning method applicable to low-speed unmanned agricultural machinery under indoor satellite signal rejection conditions, enabling indoor positioning initialization and continuous positioning functions without external information assistance other than lidar.

[0005] To achieve the above objectives, the present invention adopts the following technical solution:

[0006] A method for locating agricultural machinery hangars based on point cloud registration includes the following steps:

[0007] S1. Using the improved Scan Context++ algorithm, first-order coarse ring key matching and second-order sector key matching are performed on the hangar real-time point cloud frame obtained by the tilted agricultural machinery lidar and the prior 3D point cloud map to obtain candidate key frames and the rotation angle between the hangar real-time point cloud frame and the candidate key frame point cloud, and the initial registration matrix of normal distribution transformation is generated accordingly.

[0008] S2. Using the initial registration matrix as the initial value, continuously register the collected real-time positioning point cloud frame data of the hangar with the prior three-dimensional point cloud map, output the agricultural machinery pose, and return to S1 to reinitialize when the positioning is interrupted due to the dynamic scene.

[0009] The present invention also provides an agricultural machinery hangar positioning device based on point cloud registration, for implementing the above method, comprising the following modules:

[0010] The initial registration module uses the improved Scan Context++ algorithm to perform first-order coarse ring key matching and second-order sector key matching on the hangar real-time point cloud frame acquired by the tilted agricultural machinery lidar and the prior 3D point cloud map to obtain candidate key frames and the rotation angle between the hangar real-time point cloud frame and the candidate key frame point cloud, and generates an initial registration matrix with normal distribution transformation based on this.

[0011] The continuous registration module uses the initial registration matrix as the initial value to continuously register the collected real-time positioning point cloud frame data of the hangar with the prior 3D point cloud map, outputs the agricultural machinery pose, and re-initializes the module when the positioning is interrupted due to dynamic scenes.

[0012] The present invention also provides an electronic device, including 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 method for locating agricultural machinery hangars based on point cloud registration.

[0013] The present invention also provides a non-transitory computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the steps of the above-described method for locating agricultural machinery hangars based on point cloud registration.

[0014] Beneficial effects:

[0015] This invention combines the advantages of high precision and high reliability of second-order positioning to achieve the functions of initialization and continuous positioning of agricultural machinery in hangars under satellite signal rejection conditions. It can also respond to dynamic changes in the hangar environment and has strong robustness and environmental adaptability. Attached Figure Description

[0016] Figure 1 This is a flowchart of a method for locating agricultural machinery hangars based on point cloud registration according to the present invention;

[0017] Figure 2 A schematic diagram of a hangar point cloud collected by agricultural machinery lidar;

[0018] Figure 3 A visualization diagram of a hangar point cloud scan context descriptor; wherein, (a) is the real-time hangar point cloud frame scan context obtained by the present invention, and (b) is the real-time hangar point cloud frame scan context of the original method;

[0019] Figure 4 A schematic diagram showing the positioning results of the agricultural machinery hangar;

[0020] Figure 5 This is a schematic diagram of an agricultural machinery hangar positioning device based on point cloud registration according to the present invention. Detailed Implementation

[0021] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention. Furthermore, the technical features involved in the various embodiments of this invention described below can be combined with each other as long as they do not conflict with each other.

[0022] This invention provides a method and apparatus for locating agricultural machinery hangars based on point cloud registration. It achieves precise hangar positioning by registering point cloud data acquired by a lidar scanner mounted on the agricultural machinery with a pre-established 3D point cloud map. Specifically, the lidar point cloud positioning module loads the pre-established hangar point cloud map data and receives the lidar scanned hangar point cloud frames to be registered. The registration result between the point cloud frames and the point cloud map is the positioning result of the agricultural machinery on the hangar point cloud map. This invention is applicable to the initial positioning of a surround-view scanning lidar installed at an angle on the top of the agricultural machinery.

[0023] like Figure 1 As shown, the agricultural machinery hangar positioning method based on point cloud registration of the present invention is an improvement on the Scan Context++ open-source algorithm to achieve point cloud positioning initialization of agricultural machinery lidar hangars, specifically including the following steps:

[0024] S1. Using the improved Scan Context++ algorithm, perform first-order coarse ring key matching and second-order fine sector key matching (including coarse and fine column offset search) on the real-time point cloud frames of the hangar acquired by the tilted agricultural machinery lidar and the prior 3D point cloud map. Obtain candidate keyframes and the rotation angles between the real-time hangar point cloud frames and the candidate keyframe point clouds, and generate an initial registration matrix based on the normal distribution transformation, including:

[0025] S1.1 Loads a pre-constructed prior 3D point cloud map based on agricultural machinery lidar, and loads the keyframe point clouds and keyframe poses of the prior 3D point cloud map. The prior 3D point cloud map is built using SLAM algorithm based on hangar point clouds and inertial navigation data collected from agricultural machinery motion. Keyframes are representative frames in the prior 3D point cloud map construction process. When building the prior 3D point cloud map, the poses of the keyframes in the point cloud map are simultaneously saved, referred to as keyframe poses. The point clouds scanned by the agricultural machinery lidar corresponding to the keyframes are also saved, referred to as keyframe point clouds. An improved scan context descriptor is calculated for each keyframe point cloud, further generating circular keys, and a kd-tree of the keyframe point cloud circular keys is built.

[0026] The agricultural machinery lidar that constructs the prior 3D point cloud map and collects the real-time point cloud frames of the hangar for positioning initialization is the same device. Since the agricultural machinery lidar is installed at an angle, the real-time point cloud frames of the hangar collected by the agricultural machinery lidar need to be converted to the same horizontal body coordinate system as the key frame point cloud and key frame pose of the prior 3D point cloud map to perform tilt correction, so as to ensure that the vertical distribution of the point cloud is effective, and thus the scan context descriptor is not distorted.

[0027] S1.2 receives the real-time point cloud frame of the hangar collected by the agricultural machinery lidar for positioning initialization, transforms it to the horizontal body coordinate system, calculates the improved scan context descriptor of the real-time point cloud frame of the hangar after coordinate system transformation, further generates a ring key, searches for several nearest neighbors of the ring key of the real-time point cloud frame of the hangar in the kd-tree established in S1.1, and then completes the first-order coarse search of the candidate key frame.

[0028] Improve the scan context descriptor of the hangar point cloud frame by dividing the point cloud radially into equal parts. The ring is divided into equal parts along the circumference. By selecting the average elevation of the point cloud within each sector and the area composed of the annulus and the sector as the real value of that area, the boundary features of the hangar wall are effectively preserved. This avoids extracting only the maximum value, i.e., the top features of the hangar, which would cause the scan context descriptor of the keyframe point cloud and the real-time point cloud frame of the hangar to degrade the features of the partial initial position of the hangar inside the hangar with respect to the axis of symmetry of the collected hangar top point cloud in the rotation direction.

[0029] Real values ​​of the region The calculation formula is:

[0030] ;

[0031] In the formula, Indicates the first The ring and the first A region composed of fan-shaped areas This represents the real value of the region. For this region, in the horizontal body coordinate system, the first... A point cloud coordinate vector, It is a function that returns the z-coordinate value of the point cloud coordinate vector. The number of point clouds in the region. This represents the summation of all point cloud function values.

[0032] S1.3 performs a second-order precise search for candidate keyframes, including:

[0033] Generate the real-time point cloud frame of the hangar and the sector key of the point cloud of each candidate keyframe, and compare the distance between the real-time point cloud frame of the hangar and the point cloud of each candidate keyframe in turn.

[0034] S1.3.1 Compare the L2 norm of the sector keys of the real-time point cloud frame of the hangar with the sector keys of the candidate key frames after column traversal offset, and search for the coarse column offset value.

[0035] Point cloud data collected by tilted agricultural machinery lidar lacks 360-degree rotation invariance in the horizontal direction of the point cloud descriptor. To strengthen rotation constraints, improve solution speed, and further prevent false detections of rotation angles, the ScanContext++ method is improved by adding constraints when searching for coarse column offset values. The column offset traversal range of the fan keys of candidate keyframe point clouds is set to 0~90 degrees and 270~360 degrees. The number of column offsets... for:

[0036] ;

[0037] In the formula, The number of sectors;

[0038] S1.3.2 The fan keys of the candidate keyframe point cloud are extended to the left and right of the center image by the column offset coarse value. The precise distance between the real-time point cloud frame and the candidate keyframe point cloud after column offset is obtained by using the cosine similarity computer library. The column offset corresponding to the shortest distance can be converted into the rotation angle between the real-time point cloud frame of the hangar and the candidate keyframe point cloud. This shortest distance is the distance between the real-time point cloud frame of the hangar and the candidate keyframe point cloud.

[0039] Column offset Rotation angles of the converted real-time point cloud frames and candidate keyframe point clouds in the hangar. The calculation formula is:

[0040] ;

[0041] In the formula, It is a function of converting angles to radians.

[0042] After implementing S1.3.1 and S1.3.2 on each candidate keyframe point cloud, i.e., calculating the distance between the real-time point cloud frame of the hangar and each candidate keyframe point cloud, several candidate keyframes that meet the set distance threshold are selected as second-order candidate keyframes.

[0043] S1.4 sets the prior 3D point cloud map as the target point cloud for the initial registration of the normal distribution transformation. The real-time point cloud frame of the hangar, after voxel filtering and transformation to the horizontal body coordinate system, is used as the source point cloud. The rotation angles of the real-time point cloud frame of the hangar obtained in S1.3 and several second-order candidate keyframes are applied to the poses of the corresponding second-order candidate keyframes as initial values ​​for the initial registration of the normal distribution transformation. The point cloud registration probability and transformation matrix calculated for each initial value are obtained. The maximum value of the point cloud registration probability is selected. If it is greater than the preset initial registration probability threshold, the corresponding transformation matrix is ​​the initial value for continuous positioning registration.

[0044] Specifically, the rotation angles of the real-time point cloud frame of the hangar and several second-order candidate keyframes are applied to the pose of the corresponding second-order candidate keyframes as the initial values ​​for point cloud registration. The calculation method is as follows:

[0045] ;

[0046] In the formula, The rotation angle between the real-time point cloud frame of the hangar and a certain second-order candidate keyframe. Let be the homogeneous matrix of the pose of the second-order candidate keyframe.

[0047] S2. Using the initial registration matrix as the initial value, continuously register the collected real-time positioning point cloud frame data of the hangar with the prior 3D point cloud map, output the agricultural machinery pose, and return to S1 for re-initialization when the positioning is interrupted due to dynamic scenes, including:

[0048] S2.1 Normal distribution transformation continuous positioning uses a priori 3D point cloud map as the target point cloud and real-time positioning point cloud frame data of the hangar in the horizontal machine coordinate system after voxel filtering and transformation collected by the agricultural machinery positioning lidar as the source point cloud. The transformation matrix obtained in S1.4 is used as the initial value for point cloud registration. The point cloud registration probability and transformation matrix are output. If the point cloud registration probability is greater than the preset initial positioning success threshold, the initialization is successful. The transformation matrix is ​​the initial positioning result of the agricultural machinery on the priori 3D point cloud map.

[0049] Specifically, the source point cloud for continuous positioning using normal distribution transformation is a point cloud collected by agricultural machinery for continuous positioning of the hangar. It includes, but is not limited to, the real-time point cloud frame data of the lidar scan used in S1 to construct the prior three-dimensional point cloud map of the hangar and for positioning initialization, and is transferred to the same horizontal machine coordinate system.

[0050] When the continuous positioning initialization of the normal distribution transformation in S2.1 is successful, considering the characteristics of the agricultural machinery moving at low speed in the hangar, the initial value of the positioning registration for the next frame is obtained based on the uniform motion model. Then, the normal distribution transformation positioning operation for the next frame is carried out in sequence, and the continuous positioning results are output. When the initialization fails or the positioning fails due to the dynamic change of the environment, the transformation matrix obtained in S1.4 is used as the initial value of the registration for re-initialization.

[0051] Specifically, the initial value for positioning and registration in the next frame is obtained based on the uniform motion model. The calculation formula is:

[0052] ;

[0053] In the formula, This is the localization result from the previous frame. This is the location result for the current frame.

[0054] Example:

[0055] The length, width and height are approximately The experiment was conducted in a rectangular hangar. The agricultural machinery lidar was installed on top of the machinery at a downward tilt of 30 degrees, at a height of approximately 3.5 meters. The core parameters were set as follows: 20 rings, 60 sector shapes, maximum radius of 40 meters, lidar height of 4 meters, distance threshold of 0.8 meters, and preset initial registration probability threshold of 4.8.

[0056] A point cloud frame collected by agricultural machinery lidar in the hangar, such as Figure 2 As shown, the set of white scattered points in the figure represents the real-time point cloud frame of the hangar when rotated to the horizontal body coordinate system.

[0057] A visualization diagram of the scan context descriptor generated based on the real-time point cloud frame of the hangar is shown below. Figure 3 As shown, Figure 3 The horizontal axis represents a sector sequence, and the vertical axis represents a circular sequence. The value of each colored block is a real value of the region composed of the circular sector. Figure 3 (b) is the scan context descriptor of the real-time point cloud frame of the hangar obtained by the method proposed in this invention, which can be seen from the... Figure 3Compared with the original method in (a) which uses the scan context descriptor obtained from the real-time point cloud frame of the hangar, the improved method adds column offset feature constraint of fan-shaped feature, which effectively reduces the detection ambiguity of rotation angle. Table 1 shows the initialization process and results of agricultural machinery hangar positioning, which shows the second-order candidate keyframes and rotation angles searched based on the real-time point cloud frame data of the hangar. The registration probability is then obtained by using the pose and rotation angle of the second-order candidate keyframes as the initial value of the normal distribution transformation initial registration module. It can be seen that ScanContext++ has rotation angle detection ambiguity in multiple keyframes, which makes the registration probability not meet the preset initial registration probability threshold, making it difficult to provide the initial value of the normal distribution continuous positioning module, and mismatch may occur in the 255th keyframe. The improved ScanContext++ method achieves effective detection of rotation angle in multiple second-order candidate keyframes, which alleviates the mismatch problem caused by the lack of rotation invariance of the descriptor of the tilted installation of lidar agricultural machinery and the degradation of symmetry features in the hangar scene. It has a certain environmental adaptability to point cloud scene structure transformation and enhances the robustness of initial positioning.

[0058] Table 1 Initialization data for agricultural machinery hangar positioning

[0059]

[0060] The positioning results of agricultural machinery in the hangar are as follows Figure 4 As shown, it can realize positioning initialization (obtaining the initial positioning point) and output the continuous positioning results of the agricultural machinery along the movement trajectory.

[0061] like Figure 5 As shown, the present invention also provides an agricultural machinery hangar positioning device based on point cloud registration, used to implement the above method, comprising the following modules:

[0062] The initial registration module uses the improved Scan Context++ algorithm to perform first-order coarse ring key matching and second-order sector key matching on the hangar real-time point cloud frame acquired by the tilted agricultural machinery lidar and the prior 3D point cloud map to obtain candidate key frames and the rotation angle between the hangar real-time point cloud frame and the candidate key frame point cloud, and generates an initial registration matrix with normal distribution transformation based on this.

[0063] The continuous registration module uses the initial registration matrix as the initial value to continuously register the collected real-time positioning point cloud frame data of the hangar with the prior 3D point cloud map, outputs the agricultural machinery pose, and re-initializes the module when the positioning is interrupted due to dynamic scenes.

[0064] The present invention also provides an electronic device, including 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 method for locating agricultural machinery hangars based on point cloud registration.

[0065] The present invention also provides a non-transitory computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the steps of the above-described method for locating agricultural machinery hangars based on point cloud registration.

[0066] 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 a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code. The solutions in the embodiments of the present invention can be implemented using various computer languages, such as the object-oriented programming language Java and the interpreted scripting language JavaScript.

[0067] 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 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0068] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0069] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0070] Although the preferred embodiments of the present invention have been described, those skilled in the art may make additional changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention.

[0071] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for locating agricultural machinery hangars based on point cloud registration, characterized in that, Includes the following steps: S1. Using the improved Scan Context++ algorithm, first-order coarse ring key matching and second-order sector key matching are performed on the hangar real-time point cloud frame obtained by the tilted agricultural machinery lidar and the prior 3D point cloud map to obtain candidate key frames and the rotation angle between the hangar real-time point cloud frame and the candidate key frame point cloud, and the initial registration matrix of normal distribution transformation is generated accordingly. The improved Scan Context++ algorithm uses the average elevation of point clouds within the ring and sector division regions as the region value, preserving the boundary features of the hangar walls; in the second-order sector key fine matching, the sector key is traversed within the range of 0–90° and 270°–360° to search for coarse values ​​of column offset, thereby constraining the offset range and further determining the rotation angle; S2. Using the initial registration matrix as the initial value, continuously register the collected real-time positioning point cloud frame data of the hangar with the prior three-dimensional point cloud map, output the agricultural machinery pose, and return to S1 to reinitialize when the positioning is interrupted due to the dynamic scene. Continuous registration uses a uniform motion model to predict the initial values ​​for localization and registration in the next frame. The calculation formula is as follows: ； Where, This is the localization result from the previous frame. This is the location result for the current frame.

2. The method for locating agricultural machinery hangars based on point cloud registration according to claim 1, characterized in that, In step S1, the real-time point cloud frames of the hangar collected by the agricultural machinery lidar are converted to the same horizontal machine coordinate system as the key frame point cloud and key frame pose of the prior 3D point cloud map.

3. The method for locating agricultural machinery hangars based on point cloud registration according to claim 1, characterized in that, In step S1, the same agricultural machinery lidar is used to acquire the real-time point cloud frame of the hangar and the prior point cloud map.

4. The method for locating agricultural machinery hangars based on point cloud registration according to claim 1, characterized in that, In S2, the determination of positioning interruption includes triggering re-initialization when the registration probability is lower than a threshold.

5. A positioning device for agricultural machinery hangars based on point cloud registration, characterized in that, Includes the following modules: The initial registration module utilizes an improved Scan Context++ algorithm to perform first-order circular key coarse matching and second-order sector key fine matching on the real-time point cloud frames of the hangar acquired by the tilted agricultural machinery lidar and the prior 3D point cloud map. This yields candidate keyframes and the rotation angles between the real-time point cloud frames of the hangar and the candidate keyframe point clouds, generating an initial registration matrix with a normal distribution transformation. The improved Scan Context++ algorithm uses the average elevation of the point clouds within the circular and sector division regions as the region value, preserving the boundary features of the hangar walls. In the second-order sector key fine matching, the column offset coarse values ​​are searched by traversing the sector keys within the ranges of 0–90° and 270°–360°, thereby constraining the offset range and further determining the rotation angle. The continuous registration module uses the initial registration matrix as the initial value to continuously register the collected real-time positioning point cloud frame data of the hangar with the prior three-dimensional point cloud map, outputs the agricultural machinery pose, and re-initializes the data when the positioning is interrupted due to dynamic scenes. Continuous registration uses a uniform motion model to predict the initial values ​​for localization and registration in the next frame. The calculation formula is as follows: ； In the formula, This is the localization result from the previous frame. This is the location result for the current frame.

6. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the steps of the agricultural machinery hangar positioning method based on point cloud registration as described in any one of claims 1 to 4.

7. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the agricultural machinery hangar positioning method based on point cloud registration as described in any one of claims 1 to 4.

Images