Method, system and equipment for testing loading volume of carriage of mining dump truck and storage medium

Abstract

The invention discloses a mining dump truck carriage loading volume testing method, system and device and a storage medium, and relates to the technical field of engineering machinery, the method comprises the steps of carriage point cloud data acquisition and carriage point cloud data processing, and specifically comprises the following steps: S1, point cloud preprocessing; s2, point cloud registration processing; s3, denoising the point cloud; s4, carrying out point cloud voxelization; s5, carrying out voxel gridding; s6, carrying out grid projection processing; and S7, carrying out volume integration. A multi-rotor unmanned aerial vehicle system carries a laser load to scan and reconstruct compartment materials, so that high-efficiency scanning and rebuilding of compartment materials of mining dump trucks (30-400 tons) with different sizes and heights at any position of a mining area can be met, and the scanning time of a single truck is less than 3 minutes; the problem that in the prior art, due to the fact that a mining dump truck is large in truck body width and variable in driving route, a huge movable gantry beam is not suitable for being built is solved, and meanwhile the mining dump truck is not limited to volume calculation of the types of carried materials.

Description

Technical Field

[0001] The invention relates to the technical field of mining dump trucks, in particular to a method for testing the loading volume of a mining dump truck compartment. Background Art

[0002] Mining dump trucks, as crucial transport machinery in large-scale open-pit mining and earthwork construction projects, are responsible for the short-distance transport of materials such as ore and earth. However, the actual loading volume of mining dump truck compartments has been difficult to accurately measure. For systems using loading scales, the loading weight can be converted. Otherwise, the loading volume must be estimated based on the excavator's bucket capacity and the number of buckets loaded. These methods all have significant statistical errors. Efficient and accurate measurement of the loading volume can provide accurate statistical data for project production management and provide data support for determining whether dump trucks are overloaded, underloaded, or unevenly loaded.

[0003] The prior art, with publication number CN116843742B, discloses a method and system for calculating the volume of piled materials after point cloud registration for vehicles carrying black coal. The method includes using a camera or multiple two-dimensional lasers to obtain a three-dimensional point cloud model of the target object, performing three-dimensional surface reconstruction on the three-dimensional point cloud model, and restoring the discrete three-dimensional point cloud model into a three-dimensional model with a continuous surface contour. The volume of the target object can be calculated based on the data of the three-dimensional model combined with the volume integral formula. However, the above-mentioned prior art only calculates the volume of vehicles carrying black coal, making it inapplicable to the transportation of other types of materials. At the same time, due to the large body width of mining dump trucks, generally 7 to 10 meters, and the body height generally 6 to 8 meters, there will be variable driving routes, which are not suitable for the construction of large mobile gantry beams. In addition, the characteristics of the mining dump truck compartment make it unapplicable in actual applications and the construction cost is relatively high. Summary of the Invention

[0004] The purpose of the present invention is to provide a method for testing the loading volume of a mining dump truck compartment, so as to solve the problems in the current market raised by the above background technology.

[0005] To achieve the above-mentioned object, the present invention provides the following technical solutions: a method for testing the loading volume of a mining dump truck compartment, comprising: acquiring compartment point cloud data and processing the compartment point cloud data;

[0006] The acquisition of carriage point cloud data specifically includes the following:

[0007] Calibrate the inertial navigation system of the laser payload on the UAV;

[0008] The drone, equipped with a laser payload, flies above the side of a stationary, unloaded mining dump truck. The laser radar performs a circular flight scan with the truck compartment as the center. The drone then pans forward, backward, and left and right at a constant speed above the truck compartment, collecting point cloud data of the unloaded truck compartment.

[0009] Using the same setting parameters, a circular flight scan and a uniform translation scan were performed on the fully loaded mining dump truck compartment to complete the point cloud data collection of the loaded mining dump truck compartment.

[0010] The processing of carriage point cloud data specifically includes the following:

[0011] S1 point cloud preprocessing: cropping the point cloud data of empty and loaded carriages;

[0012] S2 point cloud registration processing: According to the structure of the carriage, at least three pairs of coordinate points with the same name are selected on the point clouds of the empty carriage and the full carriage respectively;

[0013] S3 point cloud denoising: crop and denoise the point clouds of empty carriages and registered fully loaded carriages;

[0014] S4 point cloud voxelization: voxelize the processed empty and loaded carriage point cloud data;

[0015] S5 voxel meshing: 3D surface reconstruction is performed on the empty and fully loaded point cloud voxels to generate a 3D mesh. The isosurface where TSDF=0 is extracted is the 3D surface mesh.

[0016] S6 mesh projection processing: vertically project each triangulated mesh of the fully loaded 3D surface mesh onto the unloaded 3D surface mesh, and use the plane equation interpolation method to calculate the elevation value projected onto the unloaded 3D surface mesh;

[0017] S7 Volume Integration: Grid projection processing is used to subdivide the complex material model into multiple small triangular prisms. The total volume of the material is obtained by calculating the volume of each triangular prism and summing them up.

[0018] Preferably, when calibrating the inertial navigation system of the UAV laser payload; the laser radar field of view FOV is 70°×75°, the laser radar scanning mode is non-repetitive scanning, and the sampling frequency is 240kHz; and / or

[0019] The drone is flown to a point 23 meters above the side of a stationary, unloaded mining dump truck and performs a circular flight scan with a radius of 40 meters and a pan / tilt angle of -25° to -60°, centered on that point; and / or

[0020] When the drone moves at a constant speed forward, backward, left and right above the carriage, its flight altitude is 30 meters vertically, and the pitch angle of the gimbal is -90°.

[0021] Preferably, S1 point cloud pre-processing and cropping is specifically as follows: only retaining the complete car and the loaded material part of the point cloud data, taking the brim or the side panel anti-smashing beam of the empty car as the horizontal plane, performing a rotation transformation on the point cloud of the empty car, so that the transformed empty car is in a horizontal state;

[0022] The coordinate points with the same name selected in the S2 point cloud registration process include: brim corner points, front and rear corner points of side panels, edge points of stone retaining panels and / or hanging points;

[0023] The cropping in S3 point cloud denoising is to crop the overlapping empty and loaded carriages at the same time. After cropping, the edges of the empty and loaded carriages are consistent, and a spatially distributed denoising algorithm is used to remove drift points, isolated points, and mixed point noise data.

[0024] The S4 point cloud voxelization operation specifically includes: setting the voxel resolution to 0.2 meters, voxelizing the point cloud data using a distance field-based algorithm, and adaptively adjusting the voxel resolution using an octree algorithm driven by the elevation gradient |G|;

[0025] For voxels with |G| ≥ 2, a 3-level octree algorithm is used to adjust the voxel resolution. For voxels with 1.2 ≤ |G| < 3, a 2-level octree algorithm is used to adjust the voxel resolution. For voxels with 0.6 ≤ |G| < 1.2, a 1-level octree algorithm is used to adjust the voxel resolution.

[0026] Preferably, the calculation of the S2 point cloud registration process is specifically as follows:

[0027] Given the full and empty carriage structures, select three pairs of coordinate points with the same name P = {p i} and Q={q j}, use the following formula to solve the orthogonal rotation matrix R and translation matrix t:

[0028] Q=R·P+t

[0029] Where R is the orthogonal rotation matrix, t is the translation matrix, P is the coordinate point of the fully loaded carriage, and Q is the coordinate point of the empty carriage;

[0030] In order to solve the optimal rotation matrix R and translation matrix t, the objective function is used to continuously iterate until the registration error is minimized to less than 0.1, as shown in the formula:

[0031]

[0032] Substitute the optimal rotation matrix R and translation matrix t and the full-loaded carriage point cloud coordinates P into Q = R·P+t to calculate the registered full-loaded carriage point cloud data.

[0033] Preferably, the calculation of the S6 grid projection process is specifically as follows:

[0034] Assume that a vertex of the fully loaded triangulated network is P(x, y, z), and its projection onto the unloaded triangulated network is P'(x, y, z'). P' is located within the triangular surface ABC of the unloaded triangulated network, and the plane equation can be constructed:

[0035] ax+by+cz+d=0

[0036] Substitute the coordinates of the three vertices of the triangle to solve the parameters a, b, c, and d of the plane equation, and substitute the x and y coordinates of the projection point P' into the plane equation to solve the elevation value of the point:

[0037]

[0038] Preferably, the calculation of S7 volume integral is as follows:

[0039] The three vertex coordinates of the fully loaded triangulated network are projected onto the surface of the empty carriage to form a triangle. The area A of the triangle on the empty carriage surface is calculated using vector cross product:

[0040]

[0041] Where, and are the two adjacent side vectors projected onto the surface of the empty carriage;

[0042] The plane equation interpolation method is used to calculate the elevation values ​​of the three points projected onto the empty 3D surface grid, thereby obtaining the elevation difference between the loaded and empty carriages:

[0043] Use the triangular prism volume formula to calculate the volume V of a single prism p :

[0044]

[0045] Where Δz i is the elevation difference between the fully loaded and empty carriages;

[0046] The total volume of the material is obtained by summing up the volumes of all triangular prisms:

[0047]

[0048] A mining dump truck compartment loading volume testing system, used for the above-mentioned mining dump truck compartment loading volume testing method, characterized by comprising: a multi-rotor drone system, a laser payload integrated laser radar, a surveying camera, a three-axis gimbal and an inertial navigation system;

[0049] Multi-rotor UAV system, the multi-rotor UAV system has RTK positioning function;

[0050] Laser payload integrated laser radar, used to transmit laser beams and receive echoes to obtain three-dimensional information of mining dump trucks;

[0051] A mapping camera is used to render the three-dimensional information of the mining dump truck and restore its true color;

[0052] Three-axis gimbal for adjusting the attitude angle of the LiDAR; and

[0053] Inertial navigation system, used to obtain the real-time attitude angle of the lidar;

[0054] Among them, the three-axis gimbal performs dynamic compensation based on the real-time posture to achieve the expected posture angle of the lidar.

[0055] Preferably, the angular jitter of the three-axis gimbal is ≤0.02°.

[0056] A computer device includes a memory and a processor. The memory stores a computer program. When the processor executes the computer program, the steps of the above-mentioned method for testing the loading volume of a mining dump truck compartment are realized.

[0057] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of the above-mentioned method for testing the loading volume of a mining dump truck compartment.

[0058] Compared with the prior art, the present invention has the following beneficial effects:

[0059] The present invention uses a multi-rotor drone system equipped with a laser payload to scan and reconstruct carriage materials. It can meet the needs of mining dump trucks of different sizes and heights (30-400 tons) for high-efficiency scanning and reconstruction of carriage materials at any location in the mining area, and the scanning time for a single vehicle is less than 3 minutes. This solves the problem in the prior art that due to the large body width of mining dump trucks, there will be variable driving routes and it is not suitable for the construction of huge mobile gantry beams. At the same time, it is not limited to volume calculations of the types of transported materials.

[0060] By adopting a larger laser radar field of view angle and non-repetitive scanning method, combined with the drone's circling and straight-line flight methods, a complete scan of blind spots such as the car side panels and materials can be achieved, avoiding incomplete point cloud scanning of the car and material surfaces, which affects the accuracy of subsequent material volume calculations.

[0061] Through the point cloud registration optimization algorithm, the empty and fully loaded carriages are highly aligned after registration. After registration, the root mean square error of the point clouds of the main structural parts of the fully loaded and empty carriages is less than 0.02 meters, ensuring the accuracy of the subsequent calculation of the material volume using the projection method.

[0062] An octree algorithm driven by elevation gradients is used to adaptively adjust the voxel resolution, achieving high-precision voxel processing of empty and loaded carriages. This reduces the amount of data in flat areas and improves the modeling accuracy of features such as carriage baffles and sharp materials.

[0063] Using an empty carriage as a reference surface, the average elevation difference of the projected triangular prism is calculated through triangulation projection of the fully loaded carriage and plane equation interpolation method, which improves the volume calculation accuracy of each tiny triangular prism and ultimately achieves a total material volume error rate of less than 1%. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] The accompanying drawings, as part of this disclosure, are intended to provide a further understanding of the disclosure. The exemplary embodiments of the disclosure and their descriptions are intended to explain the disclosure and do not constitute undue limitations thereon. Obviously, the drawings described below are merely examples, and those skilled in the art can derive other drawings based on these drawings without inventive effort.

[0065] Figure 1 This is a flow chart of the carriage point cloud data acquisition method for the loading volume test method of a mining dump truck carriage of the present invention;

[0066] Figure 2 Schematic diagram of the process of calculating material volume using point cloud data in the method for testing the loading volume of a mining dump truck compartment according to the present invention;

[0067] Figure 3 Schematic diagram of the point cloud registration results of an empty compartment and a fully loaded compartment of a mining dump truck compartment loading volume test method of the present invention;

[0068] Figure 4 Schematic diagram of point cloud voxelization results based on distance field of the method for testing the loading volume of a mining dump truck compartment according to the present invention;

[0069] Figure 5 This is a schematic diagram of the triangular mesh results of the mining dump truck compartment loading volume testing method based on the marching cube algorithm of the present invention. DETAILED DESCRIPTION

[0070] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only 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 making creative efforts are within the scope of protection of the present invention.

[0071] See also Figure 1-Figure 5 The method for testing the loading volume of a mining dump truck compartment includes: obtaining compartment point cloud data and processing the compartment point cloud data.

[0072] Combine Figure 1 As shown in the figure, the acquisition of carriage point cloud data specifically includes the following:

[0073] First, the inertial navigation system of the laser payload on the UAV is calibrated. The laser radar field of view (FOV) is 70°×75°, the laser radar scanning mode is non-repetitive scanning, and the sampling frequency is at least 240kHz.

[0074] The drone flew to a point 23 meters above the stationary, unloaded mining dump truck and performed a circular flight scan with a radius of 40 meters and a gimbal pitch angle of -25° to -60° with this point as the center. The drone then moved forward and backward and left and right at a constant speed 30 meters vertically above the mining dump truck to scan and collect point cloud data of the unloaded mining dump truck compartment. It should be noted that the gimbal pitch angle was -90° at this time.

[0075] Using the same setting parameters, a circular flight scan and a uniform translation scan are performed on the fully loaded mining dump truck compartment to complete the point cloud data collection of the loaded mining dump truck compartment, avoiding incomplete point cloud scanning data due to obstructions by materials and the compartment.

[0076] Combine Figure 2 As shown in the figure, the processing of the carriage point cloud data specifically includes the following:

[0077] S1 point cloud preprocessing: cropping the point cloud data of empty and fully loaded carriages.

[0078] Only the complete car and the loaded material part of the point cloud data are retained, and the brim or side panel anti-smashing beam of the empty car is regarded as the horizontal plane. The point cloud of the empty car is rotated and transformed so that the transformed empty car is in a horizontal state.

[0079] S2 point cloud registration processing: According to the structure of the carriage, at least three pairs of coordinate points with the same name are selected on the point clouds of the empty carriage and the full carriage respectively;

[0080] It should be noted that the selected coordinate points with the same name include: brim corner points, front and rear corner points of side panels, edge points of stone retaining boards and / or lifting points.

[0081] The calculation of point cloud registration processing is as follows:

[0082] Given the fully loaded and unloaded carriage structures, select three pairs of coordinate points P = {pi} and Q = {qj} with the same name, and use the following formula to solve the orthogonal rotation matrix R and translation matrix t:

[0083] Q=R·P+t

[0084] Where R is the orthogonal rotation matrix, t is the translation matrix, P is the coordinate point of the fully loaded carriage, and Q is the coordinate point of the empty carriage.

[0085] In order to solve the optimal rotation matrix R and translation matrix t, the objective function is used to continuously iterate until the registration error is minimized to less than 0.1, as shown in the formula:

[0086]

[0087] Substitute the optimal rotation matrix R and translation matrix t and the full-loaded carriage point cloud coordinates P into Q = R·P+t to calculate the registered full-loaded carriage point cloud data. The registration results of the empty and full carriage point clouds are as follows: Figure 3 shown.

[0088] S3 point cloud denoising: crop and denoise the point clouds of empty carriages and registered fully loaded carriages.

[0089] It should be noted that the cropping is to crop the overlapping empty and loaded carriages at the same time. After cropping, the edges of the empty and loaded carriages are consistent, and a spatially distributed denoising algorithm is used to remove noise data such as drift points, isolated points and mixed points.

[0090] S4 point cloud voxelization: voxelize the processed empty and full carriage point cloud data. The point cloud voxelization results based on the distance field are as follows: Figure 4 shown.

[0091] The operations specifically include setting the voxel resolution to 0.2 meters, voxelizing the point cloud data using a distance field-based algorithm, and adaptively adjusting the voxel resolution using an octree algorithm driven by the elevation gradient |G| to improve the modeling accuracy of features such as car body panels and sharp materials.

[0092] For voxels with |G| ≥ 2, a 3-level octree algorithm is used to adjust the voxel resolution. For voxels with 1.2 ≤ |G| < 3, a 2-level octree algorithm is used to adjust the voxel resolution. For voxels with 0.6 ≤ |G| < 1.2, a 1-level octree algorithm is used to adjust the voxel resolution.

[0093] S5 voxel meshing: 3D surface reconstruction is performed on the empty and fully loaded point cloud voxels to generate a 3D mesh. The isosurface where TSDF=0 is extracted is the 3D surface mesh.

[0094] S6 grid projection processing: Figure 5 As shown in the figure, each triangulated mesh of the fully loaded 3D surface mesh is vertically projected onto the unloaded 3D surface mesh, and the elevation value projected onto the unloaded 3D surface mesh is calculated using the plane equation interpolation method. The specific calculation of the mesh projection process is as follows:

[0095] Assume that a vertex of the fully loaded triangulated network is P(x, y, z), and its projection onto the unloaded triangulated network is P'(x, y, z'). P' is located within the triangular surface ABC of the unloaded triangulated network, and the plane equation can be constructed:

[0096] ax+by+cz+d=0

[0097] Substitute the coordinates of the three vertices of the triangle to solve the parameters a, b, c, and d of the plane equation, and substitute the x and y coordinates of the projection point P' into the plane equation to solve the elevation value of the point:

[0098]

[0099] S7 Volume Integral: Grid projection is used to subdivide the complex material model into multiple small triangular prisms. The total volume of the material is obtained by calculating the volume of each triangular prism and summing them up. The calculation of volume integral is as follows:

[0100] The three vertex coordinates of the fully loaded triangulated network are projected onto the surface of the empty carriage to form a triangle. The area A of the triangle on the empty carriage surface is calculated using vector cross product:

[0101]

[0102] Where, and are the two adjacent side vectors of the triangle projected onto the surface of the empty carriage.

[0103] The plane equation interpolation method is used to calculate the elevation values ​​of the three points projected onto the empty 3D surface grid, thereby obtaining the elevation difference between the loaded and empty carriages:

[0104] Use the triangular prism volume formula to calculate the volume V of a single prism p :

[0105]

[0106] Where Δz i It is the elevation difference between a fully loaded and empty carriage.

[0107] The total volume of the material is obtained by summing up the volumes of all triangular prisms:

[0108]

[0109] In one embodiment, the above steps are used to complete the measurement and calculation of the volume of materials loaded in the dump truck compartment. The acquisition of the fully loaded compartment points takes less than 3 minutes. After completing one scan and preprocessing, the point cloud data of the empty compartment can be used multiple times to align with the point cloud data of different loaded compartments and calculate the material volume. The material volume error rate is less than 1%.

[0110] A mining dump truck compartment loading volume testing system, used for the above-mentioned mining dump truck compartment loading volume testing method, comprising: a multi-rotor drone system, a laser payload integrated lidar, a mapping camera, a three-axis gimbal, and an inertial navigation system;

[0111] Multi-rotor UAV system, the multi-rotor UAV system has RTK positioning function;

[0112] Laser payload integrated laser radar, used to transmit laser beams and receive echoes to obtain three-dimensional information of mining dump trucks;

[0113] A mapping camera is used to render the three-dimensional information of the mining dump truck and restore its true color;

[0114] Three-axis gimbal for adjusting the attitude angle of the LiDAR; and

[0115] Inertial navigation system, used to obtain the real-time attitude angle of the lidar;

[0116] Among them, the three-axis gimbal performs dynamic compensation based on the real-time posture to achieve the expected posture angle of the lidar.

[0117] As a further explanation: the angular jitter of the three-axis gimbal in the above system is ≤0.02°.

[0118] A computer device includes a memory and a processor. The memory stores a computer program. When the processor executes the computer program, the steps of the above-mentioned method for testing the loading volume of a mining dump truck compartment are implemented.

[0119] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of the above-mentioned method for testing the loading volume of a mining dump truck compartment.

[0120] Although the embodiments of the present invention have been shown and described above, it will be understood that the above embodiments are illustrative and are not to be construed as limitations on the present invention. A person skilled in the art may change, modify, replace and modify the above embodiments within the scope of the present invention.

Claims

1. A method for testing the loading volume of a mining dump truck compartment, characterized in that: include: Acquisition and processing of carriage point cloud data; The acquisition of carriage point cloud data specifically includes the following: Calibrate the inertial navigation system of the laser payload on the UAV; The drone, equipped with a laser payload, flies above the side of a stationary, unloaded mining dump truck. The laser radar performs a circular flight scan with the truck compartment as the center. The drone then pans forward, backward, and left and right at a constant speed above the truck compartment, collecting point cloud data of the unloaded truck compartment. Using the same setting parameters, a circular flight scan and a uniform translation scan were performed on the fully loaded mining dump truck compartment to complete the point cloud data collection of the loaded mining dump truck compartment. The processing of carriage point cloud data specifically includes the following: S1 point cloud preprocessing: cropping the point cloud data of empty and loaded carriages; S2 point cloud registration processing: According to the structure of the carriage, at least three pairs of coordinate points with the same name are selected on the point clouds of the empty carriage and the fully loaded carriage; S3 point cloud denoising: crop and denoise the point clouds of empty carriages and registered fully loaded carriages; S4 point cloud voxelization: voxelize the processed empty and fully loaded carriage point cloud data; S5 voxel meshing: 3D surface reconstruction is performed on the empty and fully loaded point cloud voxels to generate a 3D mesh. The isosurface where TSDF=0 is extracted is the 3D surface mesh. S6 mesh projection processing: vertically project each triangulated mesh of the fully loaded 3D surface mesh onto the unloaded 3D surface mesh, and use the plane equation interpolation method to calculate the elevation value projected onto the unloaded 3D surface mesh; S7 Volume Integration: Grid projection is used to subdivide complex material models into multiple small triangular prisms. The total volume of the material is obtained by calculating the volume of each triangular prism and summing them up.

2. The method for testing the loading volume of a mining dump truck compartment according to claim 1, wherein: When calibrating the UAV's inertial navigation system for laser payloads; the laser radar field of view is 70°×75°, the laser radar scanning mode is non-repetitive scanning, and the sampling frequency is 240kHz; and / or The drone is flown to a point 23 meters above the side of a stationary, unloaded mining dump truck and performs a circular flight scan with a radius of 40 meters and a pan / tilt angle of -25° to -60°, centered at that point; and / or When the drone moves at a constant speed forward, backward, left and right above the carriage, its flight altitude is 30 meters vertically, and the pitch angle of the gimbal is -90°.

3. The method for testing the loading volume of a mining dump truck compartment according to claim 1 or 2, characterized in that: The S1 point cloud pre-processing and cropping process is as follows: only the complete car and the loaded material part of the point cloud data are retained, the brim or side panel anti-smashing beam of the empty car is regarded as the horizontal plane, and the point cloud of the empty car is rotated to make the transformed empty car horizontal; The coordinate points with the same name selected in the S2 point cloud registration process include: brim corner points, front and rear corner points of side panels, edge points of stone retaining panels and / or hanging points; The cropping in S3 point cloud denoising is to crop the overlapping empty and loaded carriages at the same time. After cropping, the edges of the empty and loaded carriages are consistent, and a spatially distributed denoising algorithm is used to remove drift points, isolated points, and mixed point noise data. The S4 point cloud voxelization operation specifically includes: setting the voxel resolution to 0.2 meters, voxelizing the point cloud data using a distance field-based algorithm, and adaptively adjusting the voxel resolution using an octree algorithm driven by the elevation gradient |G|; For voxels with |G| ≥ 2, a 3-level octree algorithm is used to adjust the voxel resolution. For voxels with 1.2 ≤ |G| < 3, a 2-level octree algorithm is used to adjust the voxel resolution. For voxels with 0.6 ≤ |G| < 1.2, a 1-level octree algorithm is used to adjust the voxel resolution.

4. The method for testing the loading volume of a mining dump truck compartment according to claim 3, wherein: The calculation of S2 point cloud registration processing is as follows: Given the full and empty carriage structures, select three pairs of coordinate points with the same name P = {p i } and Q={q j }, use the following formula to solve the orthogonal rotation matrix R and translation matrix t: Q=R·P+t Where R is the orthogonal rotation matrix, t is the translation matrix, P is the coordinate point of the fully loaded carriage, and Q is the coordinate point of the empty carriage; In order to solve the optimal rotation matrix R and translation matrix t, the objective function is used to continuously iterate until the registration error is minimized to less than 0.1, as shown in the formula: Substitute the optimal rotation matrix R and translation matrix t and the full-loaded carriage point cloud coordinates P into Q = R·P+t to calculate the registered full-loaded carriage point cloud data.

5. The method for testing the loading volume of a mining dump truck compartment according to claim 1 or 4, characterized in that: The calculation of S6 grid projection processing is as follows: Assume that a vertex of the fully loaded triangulated network is P(x, y, z), and its projection onto the unloaded triangulated network is P'(x, y, z'). P' is located within the triangular surface ABC of the unloaded triangulated network, and the plane equation can be constructed: ax+by+cz+d=0 Substitute the coordinates of the three vertices of the triangle to solve the parameters a, b, c, and d of the plane equation, and substitute the x and y coordinates of the projection point P' into the plane equation to solve the elevation value of the point:

6. The method for testing the loading volume of a mining dump truck compartment according to claim 5, characterized in that: The calculation of S7 volume integral is as follows: The three vertex coordinates of the fully loaded triangulated network are projected onto the surface of the empty carriage to form a triangle. The area A of the triangle on the empty carriage surface is calculated using vector cross product: Where, and are the two adjacent side vectors projected onto the surface of the empty carriage; The plane equation interpolation method is used to calculate the elevation values ​​of the three points projected onto the empty 3D surface grid, thereby obtaining the elevation difference between the loaded and empty carriages: Use the triangular prism volume formula to calculate the volume V of a single prism p : Where Δz i is the elevation difference between the fully loaded and empty carriages; The total volume of the material is obtained by summing up the volumes of all triangular prisms:

7. A mining dump truck compartment loading volume testing system, used in the mining dump truck compartment loading volume testing method according to any one of claims 1 to 6, characterized in that: Includes: multi-rotor UAV system, laser payload integrated lidar, mapping camera, three-axis gimbal and inertial navigation system; Multi-rotor UAV system, the multi-rotor UAV system has RTK positioning function; Laser payload integrated laser radar, used to transmit laser beams and receive echoes to obtain three-dimensional information of mining dump trucks; A mapping camera is used to render the three-dimensional information of the mining dump truck and restore its true color; Three-axis gimbal, used to adjust the attitude angle of the lidar; and Inertial navigation system, used to obtain the real-time attitude angle of the lidar; Among them, the three-axis gimbal performs dynamic compensation based on the real-time posture to achieve the expected posture angle of the lidar.

8. The method for testing the loading volume of a mining dump truck compartment according to claim 7, wherein: The angular jitter of the three-axis gimbal is ≤0.02°.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method for testing the loading volume of a mining dump truck compartment according to any one of claims 1 to 6 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method for testing the loading volume of a mining dump truck compartment according to any one of claims 1 to 6 are implemented.

Images