Ground-unmanned aerial vehicle cooperative positioning method and device suitable for forest environment

Abstract

The invention discloses a ground-unmanned aerial vehicle cooperative positioning method and device suitable for a forest environment, and the method comprises the steps: firstly obtaining laser radar data and integrated navigation data obtained through the scanning of an unmanned aerial vehicle, and obtaining the overall point cloud of a forest; preprocessing, vertical feature extraction, trunk point clustering, trunk point fitting and trunk clustering center point clustering are carried out on the forest overall point cloud, and the tree position of the unmanned aerial vehicle priori map is obtained; then acquiring laser radar data of ground mobile equipment, constructing a local point cloud map through an SLAM algorithm, and performing the same operation on the local point cloud map to obtain the tree position of the local map; and the local map and the three-dimensional point cloud with geographic reference of the unmanned aerial vehicle are fused to solve the positioning problem of the ground mobile surveying and mapping system. According to the method, positioning is realized by using natural tree distribution in a forest environment, a manual reference target does not need to be deployed in the field, availability and uncertainty information of a positioning result is provided, and fusion with other positioning technologies is facilitated.

Description

Technical Field

[0001] The present invention relates to the field of positioning technology, specifically to the field of GNSS positioning in a stationary environment and laser radar mobile mapping technology, more specifically to a mobile mapping equipment and intelligent mechanical high-precision positioning technology and method suitable for a forest environment, and especially to a ground-UAV collaborative positioning method and device suitable for a forest environment. Background Art

[0002] Forests are one of the most important natural ecosystems. The sustainable development of forest resources requires accurate and efficient forest resource inventory, including accurate acquisition of 3D spatial data and extraction of forest structure information, such as diameter at breast height (DBH), tree height, and canopy size. LiDAR technology has become a popular choice for the development of various mobile mapping systems such as unmanned aerial vehicles (UAVs) and ground mobile platforms (such as backpacks, handheld systems, and intelligent forestry machinery) due to its high accuracy and efficiency. These mobile mapping systems can be used to collect 3D point clouds of forests above and below the canopy.

[0003] UAV-based mobile mapping systems typically use Global Navigation Satellite System (GNSS) positioning as a geo-reference solution when operating above the tree canopy, and they have been widely used in forestry resource surveys, such as segmenting individual trees, estimating crown dimensions, and tree height. However, mobile mapping above the tree canopy is not enough to map the structure under the canopy in detail, such as estimating DBH and aboveground biomass (AGB). Therefore, ground mobile platforms such as backpack mobile mapping and intelligent machinery are needed to integrate LiDAR with ground platforms to map and perceive the structure under the canopy. Such ground systems work below the forest canopy, and the GNSS signal attenuation is severe, resulting in reduced positioning accuracy, which is not enough to serve as a position reference for 3D point clouds.

[0004] The navigation and mobile mapping community has long been researching precise positioning technology in GNSS-degraded environments. GNSS and inertial measurement unit (GNSS / INS) integrated positioning is the most widely used solution. The integration of GNSS and inertial sensors can effectively solve the problem of short-term interruptions of GNSS signals and alleviate GNSS positioning anomalies caused by signal interference. However, when the GNSS interruption time continues to increase, the positioning error will continue to accumulate, resulting in its inability to provide accurate position for a long time. LiDAR-based simultaneous localization and mapping (SLAM) is a self-positioning method, but the SLAM method is a relative positioning method without geographic coordinate reference, and the positioning accuracy of SLAM will decrease with the increase of moving distance, and it needs to be calibrated by absolute positioning methods. At present, obtaining accurate positioning information in the environment under the forest canopy is still a challenge faced by ground mobile mapping and intelligent machinery. Summary of the invention

[0005] In order to overcome the problem of inaccurate positioning of mobile surveying and mapping equipment and intelligent machinery in forests, the present invention proposes a forest environment ground-UAV collaborative positioning method based on lidar map matching.

[0006] In order to achieve the above object, the technical solution adopted by the present invention is as follows:

[0007] A first aspect provides a ground-UAV collaborative positioning method suitable for a forest environment, comprising:

[0008] Obtain the lidar data and combined navigation data obtained from the drone scan to obtain the overall point cloud of the forest as the drone prior map;

[0009] The overall forest point cloud is preprocessed with vertical feature extraction, trunk point clustering, trunk point fitting and trunk cluster center point clustering to obtain the tree locations of the UAV prior map;

[0010] Obtain the lidar data of the ground mobile device, build a local point cloud map through the SLAM algorithm, pre-process the local point cloud map and extract vertical features, cluster tree points, fit tree points and cluster the center points of tree clusters to obtain the tree positions of the local map;

[0011] Based on the tree positions of the UAV prior map and the tree positions of the local map, a triangulated network is constructed, and corresponding triangles and corresponding points are found according to the similarity between the triangles contained in the two triangulated networks;

[0012] The coordinate transformation matrix is ​​calculated based on the corresponding triangles found, and according to the coordinate transformation matrix and the distance between the corresponding points, the coordinate transformation matrix with the smallest error is selected as the position of the local map in the UAV prior map.

[0013] In one embodiment, the overall forest point cloud is preprocessed and vertical feature extracted, trunk point clustered, trunk point fitted and trunk cluster center point clustered to obtain the tree positions of the drone prior map, including:

[0014] Downsample and ground filter the overall point cloud of the forest; calculate the verticality characteristics of all points in the overall point cloud of the forest, and select points whose verticality meets the preset conditions to obtain the trunk point cloud;

[0015] The tree trunk point cloud is sliced ​​vertically according to a preset thickness and clustered according to distance to obtain clusters of tree trunk points of different heights and different trees;

[0016] Fit each trunk point cluster and determine whether each trunk point cluster is part of the trunk;

[0017] The center points of the trunk clusters obtained in the trunk point fitting step are projected onto the XOY plane, and the center points are clustered by distance to obtain the tree centers and generate the tree positions of the UAV prior map.

[0018] In one embodiment, each trunk point cluster is fitted to determine whether each trunk point cluster is part of a trunk, including: using a three-point method to determine whether each trunk point cluster is a circle, and determining whether the trunk point cluster is a trunk based on the distance between the point in the trunk point cluster and the circle.

[0019] In one embodiment, searching for corresponding triangles and corresponding points based on similarities between triangles included in two triangulated networks includes:

[0020] Calculate the similarity between a triangle in the local map triangulation network and all triangles in the drone prior map triangulation network;

[0021] According to the similarity, the corresponding triangles of the triangles in the local map triangulation network are selected from the triangulation network of the drone prior map;

[0022] Find corresponding points based on corresponding triangles.

[0023] In one embodiment, after selecting corresponding triangles of triangles in the local map triangulation network from the drone prior map triangulation network according to the size of the similarity, the method further includes: using a nearest triangle similarity ratio method to determine whether there are two or more corresponding triangles, wherein the nearest triangle similarity ratio method determines whether there are two or more corresponding triangles based on the relationship between the ratio of the minimum similarity to the second smallest similarity and a ratio threshold.

[0024] In one embodiment, a coordinate transformation matrix is ​​calculated based on the corresponding triangles found, and according to the coordinate transformation matrix and the distance between the corresponding points, a coordinate transformation matrix with the smallest error is selected as the position of the local map in the prior map of the drone, including:

[0025] According to the three points in each corresponding triangle found in the triangulated network of the UAV prior map, a coordinate transformation matrix between the UAV prior map and the local map is calculated using the singular value decomposition algorithm;

[0026] Filter out the target corresponding points according to whether the distance between the corresponding points meets the distance threshold;

[0027] The coordinate transformation matrix with the largest number of corresponding points is selected as the optimal solution. When there are multiple triangle pairs with the same number of corresponding points, the coordinate transformation matrix corresponding to the triangle pair with the smallest error is selected as the optimal solution. The optimal solution is the position of the local map in the drone's prior map.

[0028] In one embodiment, after calculating the coordinate transformation matrix based on the corresponding triangles found, and selecting the coordinate transformation matrix with the smallest error as the position of the local map in the drone prior map according to the coordinate transformation matrix and the distance between the corresponding points, the method further includes:

[0029] Calculate the average distance between all corresponding points according to the selected coordinate transformation matrix;

[0030] Whether the positioning is successful is determined based on the relationship between the calculated average distance between the corresponding points and the average distance threshold.

[0031] Based on the same inventive concept, the second aspect of the present invention provides a ground-UAV collaborative positioning device suitable for a forest environment, characterized in that it includes:

[0032] The UAV prior map construction module is used to obtain the lidar data and combined navigation data obtained by the UAV scanning to obtain the overall point cloud of the forest as the UAV prior map;

[0033] The tree position generation module of the UAV prior map is used to preprocess the overall forest point cloud and extract vertical features, cluster tree points, fit tree points and cluster tree cluster center points to obtain the tree positions of the UAV prior map;

[0034] The local map tree position generation module is used to obtain the laser radar data of the ground mobile device, build a local point cloud map through the SLAM algorithm, pre-process the local point cloud map and extract vertical features, cluster the tree points, fit the tree points and cluster the center points of the tree clusters to obtain the tree positions of the local map;

[0035] A feature matching module is used to construct a triangulated network based on the tree positions of the UAV prior map and the tree positions of the local map, and find corresponding triangles and corresponding points according to the similarity between the triangles contained in the two triangulated networks;

[0036] The positioning module is used to calculate the coordinate transformation matrix based on the corresponding triangles found, and select the coordinate transformation matrix with the smallest error as the position of the local map in the drone's prior map according to the coordinate transformation matrix and the distance between the corresponding points.

[0037] Based on the same inventive concept, the third aspect of the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the ground-UAV collaborative positioning method suitable for a forest environment as described in the first aspect.

[0038] Based on the same inventive concept, the fourth aspect of the present invention 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, the ground-UAV collaborative positioning method suitable for a forest environment as described in the first aspect is implemented.

[0039] Compared with the prior art, the advantages and beneficial technical effects of the present invention are as follows:

[0040] The present invention proposes a ground-UAV collaborative positioning method suitable for forest environment, which is a global positioning method based on laser radar. It is realized by collaboratively fusing the laser radar point cloud with georeference collected by UAV / helicopter and the ground mobile scanning laser radar point cloud, solving the absolute positioning problem of backpack mobile mapping ground system or intelligent machinery. First, GNSS positioning is used to obtain a three-dimensional point cloud with georeference (forest overall point cloud); and the overall point cloud of the forest is preprocessed and vertical feature extraction, trunk point clustering, trunk point fitting and trunk clustering center point clustering are performed on the forest to obtain the tree position of the UAV prior map, then, the laser radar data of the ground mobile device is obtained, and a local point cloud map is constructed by SLAM algorithm, and the local point cloud map is preprocessed and vertical feature extraction, trunk point clustering, trunk point fitting and trunk clustering center point clustering are performed on the local point cloud map to obtain the tree position of the local map. Finally, the local map is fused with the three-dimensional point cloud with georeference of the UAV, including respectively constructing triangulated networks, finding corresponding triangles and corresponding points according to the similarity between the triangles contained in the two triangulated networks, and selecting the target coordinate transformation matrix, thereby solving the positioning problem of the ground mobile mapping system. The present invention utilizes the natural tree distribution in the forest environment to achieve positioning, without the need to deploy artificial reference targets in the wild, and provides availability and uncertainty information of the positioning results, which is convenient for integration with other positioning technologies. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0042] Figure 1 This is an overall flow chart of a ground-UAV collaborative positioning method applicable to a forest environment according to an embodiment of the present invention;

[0043] Figure 2 A schematic diagram of the process of fitting tree trunk points in the positioning method provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0044] The present invention proposes a global positioning method based on LiDAR, which is achieved by collaboratively fusing the LiDAR point cloud with geo-reference collected by UAV / helicopter and the ground mobile scanning LiDAR point cloud, solving the absolute positioning problem of backpack mobile mapping ground system or intelligent machinery. The UAV-based LiDAR mobile mapping system first uses GNSS positioning to obtain a three-dimensional point cloud with geo-reference; then, a ground mobile scanning system equipped with one or more LiDAR sensors builds a local map. Finally, the local map is fused with the UAV's three-dimensional point cloud with geo-reference to solve the positioning problem of the ground mobile mapping system. The present invention utilizes the natural tree distribution in the forest environment to achieve positioning, without the need to deploy artificial reference targets in the field, and provides the availability and uncertainty information of the positioning results, which is convenient for integration with other positioning technologies.

[0045] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0046] Embodiment 1

[0047] The present invention discloses a ground-UAV collaborative positioning method applicable to a forest environment, comprising:

[0048] S1: Obtain the lidar data and combined navigation data obtained by the drone scanning to obtain the overall point cloud of the forest as the drone prior map;

[0049] S2: Preprocessing and vertical feature extraction, trunk point clustering, trunk point fitting and trunk cluster center point clustering of the overall forest point cloud to obtain the tree positions of the UAV prior map;

[0050] S3: Obtain the lidar data of the ground mobile device, build a local point cloud map through the SLAM algorithm, pre-process the local point cloud map and extract vertical features, cluster tree points, fit tree points and cluster the center points of tree clusters to obtain the tree positions of the local map;

[0051] S4: construct triangulated networks based on the tree positions of the UAV prior map and the tree positions of the local map, and find corresponding triangles and corresponding points according to the similarity between the triangles contained in the two triangulated networks;

[0052] S5: Calculate the coordinate transformation matrix based on the corresponding triangles found, and select the coordinate transformation matrix with the smallest error as the position of the local map in the drone's prior map according to the coordinate transformation matrix and the distance between the corresponding points.

[0053] Specifically, S1 uses the lidar data and combined navigation data obtained by drone scanning to obtain the drone prior map. S2 performs feature extraction and clustering on the point cloud map in S1 to obtain the tree positions of the drone prior map. S3 uses the lidar data of ground mobile equipment to build a local point cloud map, and performs the same operation as the drone prior map in S2 to obtain the tree positions of the local map. S4 performs feature matching based on the tree positions of the two maps to find the corresponding triangles and corresponding points in the two maps. S5 finds the target coordinate transformation matrix based on the corresponding triangles and corresponding points.

[0054] The positioning method of the present invention is described below by using specific examples. Figure 1 The present invention discloses a mobile surveying and mapping device and a real-time high-precision positioning method for intelligent machinery suitable for forest environments. The specific process is as follows:

[0055] 101, taking the lidar data and combined navigation data obtained by the drone scanning, the overall point cloud of the forest is obtained as the drone prior map.

[0056] During the specific implementation process, the drone is equipped with a lidar scanning system and a GNSS / IMU combined navigation system. The combination of the two can obtain the overall point cloud of the forest with geographic coordinates, which serves as a priori map for positioning ground mobile equipment.

[0057] 102, downsampling and ground filtering are performed on the point cloud map; verticality features of all points in the point cloud map are calculated, and points with high verticality are screened to obtain a tree trunk point cloud.

[0058] In the specific implementation, the point cloud map is first uniformly downsampled, the entire point cloud is divided into voxels with a certain side length, and only the centroid point is retained in each voxel to reduce the point cloud density and the number of points. Then the cloth simulation algorithm is used to segment and identify the ground points, and only the non-ground points at a certain height above the ground are retained. The verticality features are calculated for all non-ground points:

[0059] v=1-|n z |

[0060] Among them, v represents the verticality of the point, n z Represents the third component of the normal vector of a point. The closer v is to 1, the closer the local plane where the point is located is to vertical. Based on the assumption that the trunk of a real tree grows vertically, points with v greater than the threshold γ are regarded as trunk points, and the points that meet the conditions are screened to obtain the trunk point cloud.

[0061] 103, the trunk point cloud is sliced ​​in the vertical direction at a certain thickness, and clustered according to the distance to obtain the trunk point clusters of different heights and different trees.

[0062] Specifically, the trunk point cloud is segmented in the vertical direction at a certain thickness and interval to obtain a series of slices with a thickness of h. The slices at each height are segmented into multiple clusters using Euclidean clustering, each of which is a partial trunk of different trees at the same height.

[0063] 104, fitting each trunk point cluster, and further determining whether each cluster is part of the trunk.

[0064] Specifically, the clusters obtained in 103 are not tree trunks due to various reasons such as different tree species and inaccurate normal vector calculation, so fitting is needed to determine. Traditional fitting methods such as RANSAC are time-consuming and not suitable for real-time positioning. The present invention proposes a three-point method to determine whether each cluster is a circle, and then determine whether the cluster is a tree trunk. Please refer to Figure 2 , which is a schematic diagram of a process of fitting tree trunk points in a positioning method provided in an embodiment of the present invention, including:

[0065] 201, calculate the distances between all points in the cluster and obtain the distance set d = {d 12 ,d 23 ,…d ij ,…}, and select the two points p corresponding to the largest value i ,p j ;

[0066] 202, calculate other points in the cluster and point p i ,p j The distance between them is denoted as d i and d j ;

[0067] 203, calculate d i and d j The difference between them and take the absolute value, we get the set Δd = {Δd1, Δd2, … Δd k}, select the point p corresponding to the smallest element in the set Δd k , p k is the third point needed to calculate the circle parameters;

[0068] 204, through point p i 、p j 、p k Compute the parameters of a circle:

[0069] x 2 +y 2 +ax+by+c=0

[0070] 205, calculate the distance between other points and the circle:

[0071]

[0072] Where Δ is the distance from the point to the circle, and a, b, c are the parameters of the circle calculated in 204. Points with a distance less than the threshold are marked as inliers. If the number of inliers is greater than 80% of the number of cluster points, the cluster is considered to be part of the trunk and participates in subsequent calculations (clusters that meet the conditions are calculated in 105, and those that do not meet the conditions are discarded).

[0073] 105, project the center points of the tree trunk clusters to the XOY plane, perform distance clustering on the center points, obtain the tree centers, and generate the tree locations on the drone map.

[0074] Specifically, the cluster centers selected in 104 are projected onto the XOY plane, and Euclidean clustering is performed on these center points on the XOY plane to obtain the centroid of the cluster as the location of the tree.

[0075] 106, read the laser radar data of the ground mobile scanning device, and build a local point cloud map through the SLAM (Simultaneous Localization and Mapping) algorithm, and perform steps 102-105 to obtain the tree positions in the local map.

[0076] Specifically, the ground mobile scanning data is used to construct a local map through the SLAM algorithm. The local map is a point cloud composed of a certain number of laser radar scanning frames whose relative postures are calculated by the SLAM algorithm. Compared with a single-frame laser radar scan, the local map has richer information for tree position extraction. Repeat steps 102-105 for the local map to obtain the tree position distribution of the local map.

[0077] 107, construct a triangulated network among the tree positions on the drone map and the tree positions on the local map, and find corresponding triangles and corresponding points based on triangle similarity.

[0078] Specifically, any three points in the tree location map can construct a triangle, and by traversing all tree location points, the UAV map triangulation network and the local map triangulation network can be constructed. To ensure computational efficiency, when constructing the triangulation network, each point only constructs a triangulation network with points within a certain distance range. According to the scanning distance of the laser radar carried by the ground mobile scanning device, this distance range is set to 50 meters.

[0079] The local map triangulation network is defined as T s , contains all constructed triangles; the drone map triangulation network is defined as T t, including all constructed triangles. For the two triangulated networks obtained, corresponding points are found by triangle similarity.

[0080] Specifically, define the triangle T i and T j The local map triangulation network T s and UAV map triangulation network T t The similarity between them is calculated as follows:

[0081] Sort the three sides of the triangle by length and take the shortest side as the first side to find the corresponding side. Calculate the similarity according to the following formula:

[0082]

[0083] Δ1, Δ2, and Δ3 are the length differences between the three corresponding sides of the triangle, and S is the similarity.

[0084] According to the above formula, calculate the triangle T i T-Network with UAV Map t The similarity of all triangles in , and sort the similarity in ascending order to get the set If the minimum similarity S first Satisfy the maximum threshold S th , then T t Zhong and S first The corresponding triangle is triangle T i The potential corresponding triangle of . The threshold S th Usually determined by the noise level of the point cloud and is generally set to 10 cm.

[0085] Due to the errors between the tree center position and the LiDAR scan, it is possible that t Find multiple T i To ensure the uniqueness of the correspondence and reduce the impact of mismatching, the nearest triangle similarity ratio method is used to determine whether there are two or more corresponding triangles. The correct correspondence is determined by the following formula:

[0086]

[0087] Among them, S first and S second are the minimum similarity and the second minimum similarity, SR th is a threshold greater than 1. If the conditions described by the above formula are not met, then the triangle T i In T t There is no corresponding triangle in ; if the conditions are met, then S first The corresponding triangle is triangle T iIn the UAV map triangulation network T t After determining the corresponding triangle, the corresponding points in the drone map and the local map can be determined.

[0088] 108, the coordinate transformation matrix is ​​calculated according to all corresponding triangles, and the distances between corresponding points are calculated, and the coordinate transformation matrix with the smallest error is selected as the position of the local map in the UAV prior map.

[0089] Specifically, after finding all corresponding triangles between the UAV map and the local map, the singular value decomposition (SVD) algorithm can be used to calculate a transformation matrix between the UAV map and the local map based on the three points in each corresponding triangle. The optimal solution is selected through the geometric consistency test, and the mismatched corresponding triangles can be excluded. For each pair of triangles in {(Tsi, Tti), i = 1, 2…N}, a transformation matrix is ​​calculated by SVD. Using the transformation matrix, the points in the local map are transformed into the UAV map coordinate system. The transformed points are iteratively calculated to check whether there are corresponding points in the UAV map that are close enough, that is,

[0090]

[0091] in, and are the positions of the transformed point and the corresponding point in the drone map, respectively. Δd is the distance threshold of the corresponding point pair. Points that meet the distance threshold are corresponding points, otherwise there are no corresponding points, and this pair of triangles is a wrong match. The transformation matrix with the largest number of corresponding points is selected as the optimal solution. If many triangle pairs generate the same number of corresponding points, the transformation matrix corresponding to the triangle pair with the smallest error is selected. The transformation matrix is ​​the current position of the device in the geographic coordinate system.

[0092] 109, judging whether the positioning is successful according to the number of corresponding points in the local map and the UAV map.

[0093] Specifically, the number of corresponding points found in step 108 is N orresponding The number of tree locations extracted from the local map in step 106 is N extracted , the completeness index Ig can be written as:

[0094]

[0095]

[0096] Where CS is the confidence score; Ig is the completeness of global positioning; Thr is the threshold of CS. After the current position is calculated in step 108, the completeness is calculated using the above formula, which determines whether the positioning result is available. If Ig is 1, the positioning result is available; conversely, if Ig is 0, the positioning result is not available. The confidence score Thr is generally set to 0.75.

[0097] The prior art (publication number CN115480257A) discloses a matching positioning method based on rod-shaped features, which uses a collection vehicle to scan a laser point cloud map in advance and extract rod-shaped features; uses the real-time laser point cloud scanned by the vehicle to extract rod-shaped features, and obtains the vehicle posture through feature matching. The invention concept and specific technical solution of the present invention are different, mainly in the following aspects.

[0098] (1) The laser point cloud maps used are different. The laser point cloud maps and real-time positioning acquisition maps used in the prior art are both collected on-board. The laser point cloud map used in the present invention is the ALS point cloud scanned by the drone, which is different from the real-time laser point cloud obtained during positioning in terms of scanning angle, accuracy, density, etc. From the perspective of practical application, ALS point cloud acquisition is more convenient and has higher accuracy (POS system position and posture can be used).

[0099] (2) The feature extraction methods used are different. The prior art uses the laser radar line number characteristic to filter the point cloud (paragraph

[0013] of the specification). This line number feature is only available for rotating laser radars and cannot be used for solid-state laser radars. The feature extraction method used in the present invention does not require the type of laser radar. The prior art uses the z-axis component of the feature vector to judge and extract the point cloud cluster of the trunk part (paragraphs

[0019] -

[0020] ), while the present invention uses a cylindrical fitting method to judge, which has high computational efficiency; and the center of the trunk is extracted at multiple heights, which can eliminate the influence of bifurcated branches and is more accurate. The rod-shaped feature described in the prior art is a point cloud cluster that meets the feature vector requirements (paragraph

[0023] ), and the present invention directly uses the fitted trunk center as the feature point.

[0100] (3) The feature matching method used is different. The prior art requires an initial posture (paragraph

[0094] of the specification) and requires continuous operation to achieve positioning. The present invention has an independent matching and positioning process each time and does not require an initial posture. The posture calculation method used in the prior art is point-to-point residual optimization, which is relatively time-consuming. The present invention uses a triangulated network matching method, which is more efficient and introduces a mismatch elimination step to improve accuracy.

[0101] In general, the present invention provides a mobile mapping equipment and intelligent mechanical high-precision positioning technology and method suitable for forest environments, which provides positioning references for ground equipment through prior maps collected efficiently by drones. The forest environment ground equipment positioning method proposed by the present invention can solve the problems of low positioning accuracy and poor robustness caused by insufficient GNSS signals in forests, provide stable and high-precision position references in GNSS-free environments, and provide confidence references for each positioning result.

[0102] Embodiment 2

[0103] Based on the same inventive concept, this embodiment discloses a ground-UAV collaborative positioning device suitable for a forest environment, comprising:

[0104] The UAV prior map construction module is used to obtain the lidar data and combined navigation data obtained by the UAV scanning to obtain the overall point cloud of the forest as the UAV prior map;

[0105] The tree position generation module of the UAV prior map is used to preprocess the overall forest point cloud and extract vertical features, cluster tree points, fit tree points and cluster tree cluster center points to obtain the tree positions of the UAV prior map;

[0106] The local map tree position generation module is used to obtain the laser radar data of the ground mobile device, build a local point cloud map through the SLAM algorithm, pre-process the local point cloud map and extract vertical features, cluster the tree points, fit the tree points and cluster the center points of the tree clusters to obtain the tree positions of the local map;

[0107] A feature matching module is used to construct a triangulated network based on the tree positions of the UAV prior map and the tree positions of the local map, and find corresponding triangles and corresponding points according to the similarity between the triangles contained in the two triangulated networks;

[0108] The positioning module is used to calculate the coordinate transformation matrix based on the corresponding triangles found, and select the coordinate transformation matrix with the smallest error as the position of the local map in the drone's prior map according to the coordinate transformation matrix and the distance between the corresponding points.

[0109] Since the device introduced in the second embodiment of the present invention is a device used to implement the ground-UAV collaborative positioning method applicable to the forest environment in the first embodiment of the present invention, based on the method introduced in the first embodiment of the present invention, the person skilled in the art can understand the specific structure and deformation of the device, so it is not repeated here. All devices used in the method in the first embodiment of the present invention belong to the scope of protection of the present invention.

[0110] Embodiment 3

[0111] Based on the same inventive concept, the present invention further provides a computer-readable storage medium on which a computer program is stored. When the program is executed by a processor, the method described in the first embodiment is implemented.

[0112] Since the computer-readable storage medium introduced in the third embodiment of the present invention is the computer-readable storage medium used to implement the ground-UAV collaborative positioning method applicable to the forest environment in the first embodiment of the present invention, based on the method introduced in the first embodiment of the present invention, the person skilled in the art can understand the specific structure and deformation of the computer-readable storage medium, so it is not repeated here. All computer-readable storage media used in the method of the first embodiment of the present invention belong to the scope of protection of the present invention.

[0113] Embodiment 4

[0114] The present invention also provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the method described in Embodiment 1 when executing the program.

[0115] Since the computer device introduced in the fourth embodiment of the present invention is a computer device used to implement the ground-UAV collaborative positioning method applicable to the forest environment in the first embodiment of the present invention, based on the method introduced in the first embodiment of the present invention, the person skilled in the art can understand the specific structure and deformation of the computer device, so it is not repeated here. All computer devices used in the method of the first embodiment of the present invention belong to the scope of protection of the present invention.

[0116] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may 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.

[0117] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0118] Although preferred embodiments of the present invention have been described, additional changes and modifications may be made to these embodiments by those skilled in the art once the basic creative concepts are known. 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. Obviously, those skilled in the art may make various changes and modifications to the embodiments of the present invention without departing from the spirit and scope of the embodiments of the present invention. Thus, if these modifications and variations of the embodiments of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A ground-UAV collaborative positioning method suitable for forest environments, characterized in that: include: Obtain the lidar data and combined navigation data obtained from the drone scan to obtain the overall point cloud of the forest as the drone prior map; The overall forest point cloud is preprocessed with vertical feature extraction, trunk point clustering, trunk point fitting and trunk cluster center point clustering to obtain the tree locations of the UAV prior map; Obtain the lidar data of the ground mobile device, build a local point cloud map through the SLAM algorithm, pre-process the local point cloud map and extract vertical features, cluster tree points, fit tree points and cluster the center points of tree clusters to obtain the tree positions of the local map; Based on the tree positions of the UAV prior map and the tree positions of the local map, a triangulated network is constructed, and corresponding triangles and corresponding points are found according to the similarity between the triangles contained in the two triangulated networks; The coordinate transformation matrix is ​​calculated based on the corresponding triangles found, and according to the coordinate transformation matrix and the distance between the corresponding points, the coordinate transformation matrix with the smallest error is selected as the position of the local map in the UAV prior map.

2. The ground-UAV collaborative positioning method applicable to forest environment as claimed in claim 1, characterized in that: The overall forest point cloud is preprocessed and vertical feature extracted, trunk point clustered, trunk point fitted and trunk cluster center point clustered to obtain the tree positions of the UAV prior map, including: Downsample and ground filter the overall point cloud of the forest; calculate the verticality characteristics of all points in the overall point cloud of the forest, and select points whose verticality meets the preset conditions to obtain the trunk point cloud; The tree trunk point cloud is sliced ​​vertically according to a preset thickness and clustered according to distance to obtain clusters of tree trunk points of different heights and different trees; Fit each trunk point cluster and determine whether each trunk point cluster is part of the trunk; The center points of the trunk clusters obtained in the trunk point fitting step are projected onto the XOY plane, and the center points are clustered by distance to obtain the tree centers and generate the tree positions of the UAV prior map.

3. The ground-UAV collaborative positioning method applicable to forest environment as claimed in claim 2, characterized in that: Fitting each trunk point cluster to determine whether each trunk point cluster is part of the trunk, including: using a three-point method to determine whether each trunk point cluster is a circle, and determining whether the trunk point cluster is a trunk according to the distance between the point in the trunk point cluster and the circle.

4. The ground-UAV collaborative positioning method applicable to a forest environment as claimed in claim 1, characterized in that: Find corresponding triangles and corresponding points based on the similarity between the triangles contained in two triangulated networks, including: Calculate the similarity between a triangle in the local map triangulation network and all triangles in the drone prior map triangulation network; According to the similarity, the corresponding triangles of the triangles in the local map triangulation network are selected from the triangulation network of the UAV prior map; Find corresponding points based on corresponding triangles.

5. The ground-UAV collaborative positioning method applicable to forest environment as claimed in claim 4, characterized in that: After selecting corresponding triangles of triangles in the local map triangulation network from the UAV prior map triangulation network according to the size of the similarity, the method further includes: using a nearest triangle similarity ratio method to determine whether there are two or more corresponding triangles, wherein the nearest triangle similarity ratio method determines whether there are two or more corresponding triangles based on the relationship between the ratio of the minimum similarity to the second minimum similarity and a ratio threshold.

6. The ground-UAV collaborative positioning method applicable to a forest environment as claimed in claim 1, characterized in that: The coordinate transformation matrix is ​​calculated based on the corresponding triangles found, and according to the coordinate transformation matrix and the distance between the corresponding points, the coordinate transformation matrix with the smallest error is selected as the position of the local map in the drone's prior map, including: According to the three points in each corresponding triangle found in the triangulated network of the UAV prior map, a coordinate transformation matrix between the UAV prior map and the local map is calculated using the singular value decomposition algorithm; Filter out the target corresponding points according to whether the distance between the corresponding points meets the distance threshold; The coordinate transformation matrix with the largest number of corresponding points is selected as the optimal solution. When there are multiple triangle pairs with the same number of corresponding points, the coordinate transformation matrix corresponding to the triangle pair with the smallest error is selected as the optimal solution. The optimal solution is the position of the local map in the drone's prior map.

7. The ground-UAV collaborative positioning method applicable to a forest environment as claimed in claim 1, characterized in that: After calculating the coordinate transformation matrix based on the corresponding triangles found, and selecting the coordinate transformation matrix with the smallest error as the position of the local map in the prior map of the drone according to the coordinate transformation matrix and the distance between the corresponding points, the method further includes: Calculate the average distance between all corresponding points according to the selected coordinate transformation matrix; Whether the positioning is successful is determined based on the relationship between the calculated average distance between the corresponding points and the average distance threshold.

8. A ground-UAV collaborative positioning device suitable for forest environments, characterized in that: include: The UAV prior map construction module is used to obtain the lidar data and combined navigation data obtained by the UAV scanning to obtain the overall point cloud of the forest as the UAV prior map; The tree position generation module of the UAV prior map is used to preprocess the overall forest point cloud and extract vertical features, cluster tree points, fit tree points and cluster tree cluster center points to obtain the tree positions of the UAV prior map; The local map tree position generation module is used to obtain the laser radar data of the ground mobile device, build a local point cloud map through the SLAM algorithm, pre-process the local point cloud map and extract vertical features, cluster the tree points, fit the tree points and cluster the center points of the tree clusters to obtain the tree positions of the local map; A feature matching module is used to construct a triangulated network based on the tree positions of the UAV prior map and the tree positions of the local map, and find corresponding triangles and corresponding points according to the similarity between the triangles contained in the two triangulated networks; The positioning module is used to calculate the coordinate transformation matrix based on the corresponding triangles found, and select the coordinate transformation matrix with the smallest error as the position of the local map in the drone's prior map according to the coordinate transformation matrix and the distance between the corresponding points.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the ground-UAV collaborative positioning method applicable to a forest environment is implemented as described in any one of claims 1 to 7.

10. A computer 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, the ground-UAV collaborative positioning method applicable to a forest environment is implemented as described in any one of claims 1 to 7.

Images