Point cloud map precision measurement method and device and computer equipment

Abstract

The invention relates to a point cloud map precision measurement method and device and computer equipment. The method comprises the following steps: acquiring truth value equipment positioning information sent by the positioning truth value equipment and real-time point cloud information sent by the laser radar; performing affine transformation processing on the real-time point cloud information according to the truth value equipment positioning information to obtain first point cloud information; calculating a first error value of a to-be-measured point cloud map according to a preset initial transformation parameter, the first point cloud information and a three-dimensional voxel grid map corresponding to the to-be-measured point cloud map; calculating a second error value of the to-be-measured point cloud map according to the first error value, the first point cloud information and the to-be-measured point cloud map; and determining a precision error result of the to-be-measured point cloud map according to a calibration error value of the positioning truth value device, the first error value and the second error value. The method can accurately evaluate the precision of the point cloud map.

Description

Technical Field

[0001] The present application relates to the field of precision measurement technology, and in particular to a method, device and computer equipment for measuring the precision of a point cloud map. Background Art

[0002] Point cloud map is an environmental map constructed by splicing laser radar point clouds. It is widely used in the field of unmanned driving. It is usually used as the base map of high-precision maps and used in laser SLAM (Simultaneous Localization and Mapping) algorithms. As a high-precision base map and the cornerstone of positioning, the accuracy of point cloud map directly affects the accuracy of subsequent high-precision maps and the accuracy of positioning. If the accuracy of point cloud map cannot meet the requirements of unmanned driving system, then the entire system will not be able to operate normally.

[0003] In the related art, the accuracy index is generally given by directly measuring the length of the road edge line in the map or the relative distance between features using a rangefinder, a ruler, etc., and the accuracy of the obtained accuracy index is relatively low. Summary of the invention

[0004] Based on this, it is necessary to provide a point cloud map accuracy measurement method, device, and computer equipment that can accurately evaluate the accuracy of point cloud maps in response to the above technical problems.

[0005] In a first aspect, the present application provides a point cloud map accuracy measurement method, which is applied to an accuracy measurement system, wherein the accuracy measurement system includes at least one positioning truth value device and at least one laser radar; the method includes:

[0006] Acquire the true value device positioning information sent by the positioning true value device and the real-time point cloud information sent by the laser radar;

[0007] Performing affine transformation processing on the real-time point cloud information according to the true value device positioning information to obtain first point cloud information;

[0008] Calculating a first error value of the point cloud map to be measured according to a preset initial transformation parameter, the first point cloud information, and a three-dimensional voxel grid map corresponding to the point cloud map to be measured;

[0009] Calculating a second error value of the point cloud map to be measured according to the first error value, the first point cloud information and the point cloud map to be measured;

[0010] The accuracy error result of the point cloud map to be measured is determined according to the calibration error value of the positioning true value device, the first error value and the second error value.

[0011] In one embodiment, the true value device positioning information includes three-axis position information and three-axis rotation information; performing affine transformation processing on the real-time point cloud information according to the true value device positioning information to obtain the first point cloud information includes:

[0012] Performing coordinate system conversion processing on the three-axis position information and the three-axis rotation information so that the coordinate systems of the converted three-axis position information, the converted three-axis rotation information and the real-time point cloud information are consistent;

[0013] Constructing an affine matrix according to the converted three-axis position information and three-axis rotation information;

[0014] First point cloud information is acquired according to the affine matrix and the real-time point cloud information.

[0015] In one embodiment, calculating the first error value of the point cloud map to be measured according to the preset initial transformation parameters, the first point cloud information and the three-dimensional voxel grid map corresponding to the point cloud map to be measured includes:

[0016] Acquire second point cloud information according to the first point cloud information and the initial transformation parameters;

[0017] Calculating normal distribution parameters and covariance matrix of each voxel grid in the three-dimensional voxel grid map according to the second point cloud information;

[0018] Calculating a registration score of the three-dimensional voxel grid map according to normal distribution parameters and covariance matrices of each voxel grid;

[0019] Iteratively calculating the registration score according to a preset iterative algorithm until a target registration score not less than a preset threshold is obtained;

[0020] A first error value is determined according to the target registration score and the second point cloud information.

[0021] In one embodiment, calculating a second error value of the point cloud map to be measured according to the first error value, the first point cloud information and the point cloud map to be measured includes:

[0022] Acquire third point cloud information according to the first error value and the first point cloud information;

[0023] A second error value is calculated based on the position information of each point in the third point cloud information and the position information of each point in the point cloud map to be measured.

[0024] In one embodiment, the calculating the second error value according to the position information of each point in the third point cloud information and the position information of each point in the point cloud map to be measured includes:

[0025] According to the position information of each point in the third point cloud information and the position information of each point in the point cloud map to be measured, obtaining a first distance value between each point in the third point cloud information and each point in the point cloud map to be measured, to form a plurality of first distance sequences, wherein the number of the first distance sequences is the same as the number of points in the third point cloud information;

[0026] According to the position information of each point in the third point cloud information and the position information of each point in the point cloud map to be tested, obtaining a second distance value between each point in the point cloud map to be tested and each point in the third point cloud information to form a plurality of second distance sequences, wherein the number of the second distance sequences is the same as the number of points in the point cloud map to be tested;

[0027] Averaging the minimum values ​​in each of the first distance sequences to obtain a first average error value;

[0028] Averaging the minimum values ​​in each of the second distance sequences to obtain a second average error value;

[0029] A minimum value between the first average error value and the second average error value is determined as a second error value.

[0030] In one embodiment, the process of acquiring the three-dimensional voxel grid map corresponding to the point cloud map to be measured includes:

[0031] The point cloud map to be measured is segmented according to the converted three-axis position information and the preset grid size to obtain a three-dimensional voxel grid map.

[0032] In a second aspect, the present application provides a point cloud map accuracy measurement device, which is applied to an accuracy measurement system, wherein the accuracy measurement system includes at least one positioning true value device and at least one laser radar; the device includes:

[0033] A first acquisition module is used to acquire the true value device positioning information sent by the positioning true value device and the real-time point cloud information sent by the laser radar;

[0034] A second acquisition module is used to perform affine transformation processing on the real-time point cloud information according to the true value device positioning information to obtain first point cloud information;

[0035] A first calculation module, used for calculating a first error value of the point cloud map to be measured according to a preset initial transformation parameter, the first point cloud information and a three-dimensional voxel grid map corresponding to the point cloud map to be measured;

[0036] A second calculation module, used for calculating a second error value of the point cloud map to be measured according to the first error value, the first point cloud information and the point cloud map to be measured;

[0037] The first determination module is used to determine the accuracy error result of the point cloud map to be measured according to the calibration error value of the positioning true value device, the first error value and the second error value.

[0038] In a third aspect, the present application provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the steps of the point cloud map accuracy measurement method described in any of the above embodiments are implemented.

[0039] In a fourth aspect, the present application provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the steps of the point cloud map accuracy measurement method described in any of the above embodiments are implemented.

[0040] In a fifth aspect, the present application provides a computer program product, including a computer program, which, when executed by a processor, implements the steps of the point cloud map accuracy measurement method described in any of the above embodiments.

[0041] The above-mentioned point cloud map accuracy measurement method, device and computer equipment include obtaining the true value device positioning information sent by the positioning true value device and the real-time point cloud information sent by the laser radar, and performing affine transformation processing on the real-time point cloud information according to the true value device positioning information to obtain the first point cloud information. Since theoretically, the closer the accuracy of the point cloud map to be measured is to the accuracy of the true value positioning device, the closer the pose transformation matrix calculated after aligning the first point cloud information with the point cloud map to be measured will be to the unit matrix, so in this application, according to the preset initial transformation parameters, the first point cloud information and the three-dimensional voxel grid map corresponding to the point cloud map to be measured are aligned to obtain the first error value of the point cloud map to be measured, and then, the first point cloud information is calibrated according to the first error value, and the gap between the calibrated first point cloud information and the point cloud map to be measured is calculated to obtain the point cloud map to be measured. Finally, according to the calibration error value of the positioning true value device, the first error value and the second error value, the accuracy error result of the point cloud map to be measured is comprehensively determined. The point cloud map accuracy of the present application is measured by positioning a true value device and a point cloud matching method. Compared with the related art of directly measuring the length of the road edge line in the map using a ruler, the operation difficulty is low and does not require a large amount of manpower. The measurement accuracy of the positioning true value device and the lidar is high and the reliability is high. The accuracy of the point cloud map to be measured is evaluated based on the true value device positioning information and the real-time point cloud information, and a high-accuracy accuracy error result can be obtained. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] In order to more clearly illustrate the technical solutions in the embodiments of the present application or related technologies, the drawings required for use in the embodiments of the present application or related technical descriptions will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0043] FIG1 is a schematic diagram of a process of measuring point cloud map accuracy in one embodiment;

[0044] FIG2 is a schematic diagram of a flow chart of step S102 in one embodiment;

[0045] FIG3 is a schematic diagram of a flow chart of step S103 in one embodiment;

[0046] FIG4 is a schematic diagram of a flow chart of step S104 in one embodiment;

[0047] FIG5 is a schematic diagram of a flow chart of step S402 in one embodiment;

[0048] FIG6 is a structural block diagram of a point cloud map accuracy measurement device in one embodiment;

[0049] FIG. 7 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0050] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0051] As described in the background technology section, a point cloud map is an environmental map constructed by splicing laser radar point clouds. It has a wide range of applications in the field of unmanned driving, usually as a base map for high-precision maps and used in laser SLAM algorithms. The accuracy of the point cloud map will directly affect the accuracy of subsequent high-precision maps and the accuracy of positioning. If the accuracy of the point cloud map cannot meet the requirements of the unmanned driving system, the entire unmanned driving system will not be able to operate normally. Therefore, measuring the accuracy of the point cloud map and giving specific indicators is a prerequisite for the design of an unmanned driving system. However, due to the complexity and abstractness of the point cloud map, it is difficult to directly measure the point cloud map. The current mainstream measurement method is to use a rangefinder, a ruler, etc. to directly measure the length of the road edge line in the map, or the relative distance between features to give an accuracy index. This method is difficult to operate, errors are prone to occur during feature extraction, and the point cloud map may be distorted during the construction process. Therefore, the direct measurement method is difficult to give the overall accuracy index of the map, and is insufficient for reference to system design.

[0052] Based on the above technical problems, in an exemplary embodiment, please refer to Figure 1, the present application provides a point cloud map accuracy measurement method, which is applied to an accuracy measurement system, the accuracy measurement system includes at least one positioning true value device and at least one laser radar; the point cloud map accuracy measurement method includes steps S101 to S105.

[0053] S101: Acquire the true value device positioning information sent by the positioning true value device and the real-time point cloud information sent by the laser radar.

[0054] In the application, a true value device and a laser radar are required to perform precision measurement on the point cloud map to be measured. A true value device and a laser radar are installed on a bracket platform, and then the bracket platform is placed at the target position in the actual scene of the point cloud map to be measured. The true value device positioning information sent by the positioning true value device and the real-time point cloud information sent by the laser radar can be obtained.

[0055] S102: Perform affine transformation processing on the real-time point cloud information according to the true value device positioning information to obtain first point cloud information.

[0056] In this embodiment, after obtaining the true value device positioning information and the real-time point cloud information, since there is still a certain distance between the positioning true value device and the laser radar, in order to ensure the consistency of the positions of the positioning true value device and the laser radar, the real-time point cloud information can be translated according to the relative position relationship between the positioning true value device and the laser radar. In addition, since the coordinate systems used by the positioning true value device and the laser radar may be different, it is also necessary to perform coordinate system conversion processing on the true value device positioning information to make the coordinate systems of the true value device positioning information and the real-time point cloud information consistent. After the above preprocessing, the real-time point cloud information can be affine transformed according to the true value device positioning information to obtain the first point cloud information.

[0057] S103: Calculate a first error value of the point cloud map to be measured according to the preset initial transformation parameters, the first point cloud information, and the three-dimensional voxel grid map corresponding to the point cloud map to be measured.

[0058] It can be understood that the first point cloud information obtained after the conversion of the true value device positioning information is very close to the coordinates of the point cloud map to be tested, so the midpoint of the first point cloud information can be distributed to each voxel grid of the three-dimensional voxel grid map according to the nearest principle, and the probability density of each point is calculated according to the normal distribution parameters in each voxel grid. The probability density values ​​of all points are added up to be the registration score of the point cloud map to be tested. In theory, if the accuracy of the point cloud map to be tested is closer to the accuracy of the positioning true value device, the registration score should be larger, that is, the larger the probability density, the pose transformation matrix calculated after the first point cloud information is aligned with the point cloud map to be tested should be infinitely close to the unit matrix, so the pose transformation matrix is ​​calculated to obtain the first error value of the point cloud map to be tested.

[0059] S104: Calculate a second error value of the point cloud map to be measured according to the first error value, the first point cloud information and the point cloud map to be measured.

[0060] In this embodiment, the first point cloud information can continue to be converted according to the first error value. The difference between the converted first point cloud information and the point cloud map to be tested is already very small. At this time, the distance between each point in the converted first point cloud information and each point in the point cloud map to be tested can be directly calculated to obtain the second error value of the point cloud map to be tested.

[0061] S105: Determine the accuracy error result of the point cloud map to be measured according to the calibration error value of the positioning true value device, the first error value and the second error value.

[0062] It can be understood that multiple target positions can be set in the actual scene of the point cloud map to be tested, and a positioning true value device and a laser radar are placed at each target position. The first error value and the second error value at each target position are obtained according to the above method, and finally the accuracy error result of the point cloud map to be tested is comprehensively determined.

[0063] The above-mentioned point cloud map accuracy measurement method includes obtaining the true value device positioning information sent by the positioning true value device and the real-time point cloud information sent by the laser radar, performing affine transformation processing on the real-time point cloud information according to the true value device positioning information, and obtaining the first point cloud information. Since theoretically, the closer the accuracy of the point cloud map to be measured is to the accuracy of the true value positioning device, the closer the pose transformation matrix calculated after aligning the first point cloud information with the point cloud map to be measured will be to the unit matrix, so in this application, according to the preset initial transformation parameters, the first point cloud information and the three-dimensional voxel grid map corresponding to the point cloud map to be measured are aligned to obtain the first error value of the point cloud map to be measured, and then, the first point cloud information is calibrated according to the first error value, and the gap between the calibrated first point cloud information and the point cloud map to be measured is calculated to obtain the point cloud map to be measured. Finally, according to the calibration error value of the positioning true value device, the first error value and the second error value, the accuracy error result of the point cloud map to be measured is comprehensively determined. The point cloud map accuracy of the present application is measured by positioning a true value device and a point cloud matching method. Compared with the related art of directly measuring the length of the road edge line in the map using a ruler, the operation difficulty is low and does not require a large amount of manpower. The measurement accuracy of the positioning true value device and the lidar is high and the reliability is high. The accuracy of the point cloud map to be measured is evaluated based on the true value device positioning information and the real-time point cloud information, and a high-accuracy accuracy error result can be obtained.

[0064] In an exemplary embodiment, please refer to FIG. 2 , step S102 , performing affine transformation processing on the real-time point cloud information according to the true value device positioning information to obtain first point cloud information, including steps S201 to S203 .

[0065] S201: performing coordinate system conversion processing on the three-axis position information and the three-axis rotation information, so that the coordinate systems of the converted three-axis position information, the converted three-axis rotation information and the real-time point cloud information are consistent.

[0066] It can be understood that the true value device positioning information includes three-axis position information and three-axis rotation information, and the three-axis rotation information is the posture information. The true value device positioning information returned by the true value positioning device is usually based on the WGS-84 coordinate system, while the point cloud map to be measured and the real-time point cloud information returned by the lidar may be the local northeast sky coordinate system. Therefore, after obtaining the true value device positioning information returned by the true value positioning device, the coordinate conversion of the true value device positioning information is required according to the actual situation to maintain the consistency of the coordinate system between the true value device positioning information, the real-time point cloud information and the point cloud map to be measured.

[0067] In the application, the laser radar is installed on the true value positioning device. There is a spatial distance between the laser radar and the true value positioning device, so the spatial distance needs to be calibrated. In one example, the direct measurement method can be used for calibration, that is, the positional relationship between the laser radar and the true value positioning device is directly measured using a tool such as a ruler. In another example, the laser radar can be calibrated by a preset calibration algorithm, and the real-time point cloud information of the laser radar and the true value device positioning information of the true value positioning device are matched by timestamps. The real-time point cloud information output by the laser radar at each moment is spliced ​​through the true value device positioning information, and the sum of the distances between each point and its nearest neighbor is calculated. Through continuous iteration, a position transformation matrix is ​​found to minimize the value of this distance sum. At this time, the position transformation matrix is ​​the relative position relationship between the laser radar and the true value positioning device. After finding the position transformation matrix, the real-time point cloud information output by the laser radar needs to be multiplied by the position transformation matrix to obtain the real-time point cloud information with the relative position error eliminated, and the position consistency between the laser radar and the true value device is maintained.

[0068] S202: construct an affine matrix according to the converted three-axis position information and three-axis rotation information.

[0069] In this embodiment, in order to ensure the accuracy of measurement, the real-time point cloud information can also be pre-processed. First, remove the noise according to the distance of the returned laser point, set the minimum distance and the maximum distance, and only retain the laser points between these two distances. Then, perform voxel downsampling on the real-time point cloud information to reduce the density of the real-time point cloud information. Then, calculate the angle offset of each point in the real-time point cloud information relative to the starting point, calculate the timestamp of the point, and compare the timestamp of the true value positioning device. Obtain the posture information corresponding to each laser point by linear interpolation. All points in the real-time point cloud information are transformed to the starting position of the frame through posture transformation based on their own posture information, eliminate the motion error of each frame of point cloud, and complete motion compensation. After that, according to the three-axis position information and three-axis rotation information output by the true value positioning device at the corresponding time, use the three-axis position information and three-axis rotation information to construct an affine matrix.

[0070] S203: Acquire first point cloud information according to the affine matrix and the real-time point cloud information.

[0071] The first point cloud information can be obtained by multiplying the real-time point cloud information with the relative position error eliminated by the affine matrix for affine transformation.

[0072] In an exemplary embodiment, please refer to Figure 3, step S103, according to the preset initial transformation parameters, the first point cloud information and the three-dimensional voxel grid map corresponding to the point cloud map to be tested, the first error value of the point cloud map to be tested is calculated, including steps S301 to S305.

[0073] S301: Acquire second point cloud information according to the first point cloud information and initial transformation parameters.

[0074] In this embodiment, the initial transformation parameters may include an initial rotation transformation matrix and an initial translation transformation matrix, and both the initial rotation transformation matrix and the initial translation transformation matrix are unit matrices. The first point cloud information is multiplied by the initial transformation parameters to obtain the second point cloud information. Since both the initial rotation transformation matrix and the initial translation transformation matrix are unit matrices, the coordinates of each point in the second point cloud information are consistent with those in the first point cloud information.

[0075] S302: Calculate the normal distribution parameters and covariance matrix of each voxel grid in the three-dimensional voxel grid map according to the second point cloud information.

[0076] The process of obtaining the three-dimensional voxel grid map corresponding to the point cloud map to be tested includes the steps of segmenting the point cloud map to be tested according to the converted three-axis position information and the preset grid size to obtain the three-dimensional voxel grid map. Specifically, the point cloud map to be tested is segmented according to the three-axis position information, and the point cloud map near the target position where the true value positioning device is located is taken out as the local point cloud map, and the space occupied by the local point cloud map is divided into a three-dimensional voxel grid of a specified size, which is generally set to 0.1m. The specific value can be adjusted according to the characteristics of the laser radar and the actual scene size.

[0077] After that, the points in the second point cloud information can be distributed to the voxel grid according to the nearest principle, and the multidimensional normal distribution parameters in each voxel grid can be calculated. , ,in, is the multidimensional normal distribution parameter of the j-th voxel grid, n is the number of points in the j-th voxel grid, xi is the coordinate of the i-th point in the j-th voxel grid, and since it is a three-dimensional point cloud, xi represents the three-dimensional coordinate. The covariance matrix of the j-th voxel grid is , express The transposed matrix of .

[0078] S303: Calculate the registration score of the three-dimensional voxel grid map according to the normal distribution parameters and covariance matrix of each voxel grid.

[0079] Calculate the probability density of each point based on the normal distribution parameters of each voxel grid , and then the probability density of all points in each voxel grid is added to obtain the registration score of the 3D voxel grid map.

[0080] S304: Iteratively calculate the registration score according to a preset iterative algorithm until a target registration score not less than a preset threshold is obtained.

[0081] If the second point cloud information matches the point cloud map more closely, the registration score will be higher. Therefore, the registration score can be iteratively calculated using a preset iterative algorithm until the registration score is greater than or equal to a preset threshold, the iterative calculation is stopped, and the registration score at this time is marked as the target registration score. For example, the preset iterative algorithm can be one of the Newton algorithm and the LM algorithm.

[0082] S305: Determine a first error value according to the target registration score and the second point cloud information.

[0083] When the registration score is greater than or equal to the preset threshold, the transformation parameters at this time are the target transformation parameters. The target rotation transformation matrix and the target translation transformation matrix included in the target transformation parameters are the first error values. Therefore, the target transformation parameters can be inferred based on the target registration score and the second point cloud information to obtain the first error value.

[0084] In an exemplary embodiment, please refer to FIG. 4 , step S104 , calculating a second error value of the point cloud map to be measured according to the first error value, the first point cloud information and the point cloud map to be measured, including step S401 and step S402 .

[0085] S401: Acquire third point cloud information according to the first error value and the first point cloud information.

[0086] The third point cloud information can be obtained by multiplying the first point cloud information with the target transformation parameter. The difference between the third point cloud information and the point cloud map to be tested is already very small. The second error value can be obtained by calculating the difference between the point cloud frame and the point cloud map at this time.

[0087] S402: Calculate a second error value according to the position information of each point in the third point cloud information and the position information of each point in the point cloud map to be measured.

[0088] In the application, the relative distance between each point in the third point cloud information and each point in the point cloud map to be tested can be calculated based on the position information of each point in the third point cloud information and the position information of each point in the point cloud map to be tested, and the minimum value of the relative distance is marked as the second error value. Since the distance only represents the translation relationship, the second error value only includes the distance error and no rotation error.

[0089] In an exemplary embodiment, please refer to FIG. 5 , step S402 , calculating a second error value according to the position information of each point in the third point cloud information and the position information of each point in the point cloud map to be measured, including steps S501 to S505 .

[0090] S501: According to the position information of each point in the third point cloud information and the position information of each point in the point cloud map to be measured, obtain the first distance value between each point in the third point cloud information and each point in the point cloud map to be measured, and form a plurality of first distance sequences.

[0091] Specifically, all points in the third point cloud information are taken out, and the distance from each point to each point in the point cloud map to be measured is calculated respectively to form multiple first distance sequences, wherein the number of first distance sequences is the same as the number of points in the third point cloud information.

[0092] S502: According to the position information of each point in the third point cloud information and the position information of each point in the point cloud map to be measured, obtain the second distance value between each point in the point cloud map to be measured and each point in the third point cloud information to form a plurality of second distance sequences.

[0093] Similarly, all points in the point cloud map to be tested are taken out, and the distance from each point to each point in the third point cloud information is calculated respectively to form multiple second distance sequences, wherein the number of second distance sequences is the same as the number of points in the point cloud map to be tested.

[0094] S503: Perform an average processing on the minimum values ​​in each first distance sequence to obtain a first average error value.

[0095] The minimum values ​​in each first distance sequence are taken out and averaged to obtain a first average error value.

[0096] S504: Perform an average processing on the minimum values ​​in each second distance sequence to obtain a second average error value.

[0097] The minimum value in each second distance sequence is taken out, and averaged to obtain a second average error value.

[0098] S505: Determine the minimum value between the first average error value and the second average error value as the second error value.

[0099] The second error value is the minimum value of the first average error value and the second average error value.

[0100] In a detailed embodiment, when it is necessary to perform precision measurement on a point cloud map, multiple true value positioning devices and multiple laser radars can be prepared, and multiple target positions are determined in the actual scene of the point cloud map to be measured, and a true value positioning device and a laser radar are respectively set at each target position. Then, for the true value positioning device and laser radar at each target position, the true value device positioning information and real-time point cloud information fed back by them are received. When the coordinate systems of the true value device positioning information and the real-time point cloud information are different, the true value device positioning information is subjected to coordinate system conversion processing to maintain the consistency of the coordinate systems between the true value device positioning information, the real-time point cloud information and the point cloud map to be measured. In addition, the real-time point cloud information is updated according to the position conversion matrix to eliminate the relative error between the real-time point cloud information and the true value device positioning information. Afterwards, an affine matrix is ​​constructed according to the three-axis position information and the three-axis rotation information output by the true value positioning device at the corresponding time, and the real-time point cloud information with the relative position error eliminated is multiplied by the affine matrix for affine transformation to obtain the first point cloud information.

[0101] Secondly, the first point cloud information is multiplied by the initial rotation transformation matrix and the initial translation transformation matrix to obtain the second point cloud information, and the local point cloud map at the target position is taken out from the point cloud map to be tested according to the three-axis position information, and the local point cloud map is divided according to the preset grid size to obtain a three-dimensional voxel grid map, and the points in the second point cloud information are distributed to each voxel grid according to the nearest principle, and the multidimensional normal distribution parameters and covariance matrix of each voxel grid are calculated, and then the probability density of each point is calculated to obtain the registration score of the three-dimensional voxel grid map. The registration score is iteratively calculated according to the preset iterative algorithm until the registration score is greater than the preset threshold, and the registration score at this time is marked as the target registration score. At this time, the transformation parameter corresponding to the target registration score is the first error value w1, and the first error value w1 includes the distance error and the rotation error.

[0102] Then, the first point cloud information is multiplied by the target transformation parameter to obtain the third point cloud information. All points in the third point cloud information are taken, and the distance from each point to each point in the point cloud map to be tested is calculated. The minimum value is taken, and all the minimum distances are averaged, which is set as the first average error value. Conversely, all points in the point cloud map to be tested are taken, and the distance from each point to all points in the third point cloud information is calculated. The minimum value is taken, and all the minimum distances are averaged, which is recorded as the second average error value. The second error value is the minimum value of the first average error value and the second average error value. The second error value only includes the distance error and no rotation error.

[0103] Again, the first error value, the second error value and the calibration error value of the positioning true value device are added together to obtain the accuracy measurement value of the point cloud map to be measured at the target position. Among them, the calibration error value of the positioning true value device includes distance error and rotation error. Finally, the accuracy measurement values ​​of the point cloud map to be measured at each target position are averaged, or the maximum value, or the minimum value, which is the accuracy measurement result of the entire point cloud map to be measured.

[0104] It should be understood that, although the various steps in the flowcharts involved in the above-mentioned embodiments are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence according to the order indicated by the arrows. Unless there is a clear explanation in this article, the execution of these steps does not have a strict order restriction, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above-mentioned embodiments can include multiple steps or multiple stages, and these steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a part of the steps or stages in other steps.

[0105] Based on the same inventive concept, the embodiment of the present application also provides a point cloud map accuracy measurement device for implementing the point cloud map accuracy measurement method involved above. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above method, so the specific limitations in one or more point cloud map accuracy measurement device embodiments provided below can refer to the limitations of the point cloud map accuracy measurement method above, and will not be repeated here.

[0106] In an exemplary embodiment, as shown in Figure 6, a point cloud map accuracy measurement device is provided, which is applied to an accuracy measurement system, the accuracy measurement system includes at least one positioning truth device and at least one laser radar; the point cloud map accuracy measurement device includes a first acquisition module 601, a second acquisition module 602, a first calculation module 603, a second calculation module 604 and a first determination module 605.

[0107] The first acquisition module 601 is used to acquire the true value device positioning information sent by the positioning true value device and the real-time point cloud information sent by the laser radar.

[0108] The second acquisition module 602 is used to perform affine transformation processing on the real-time point cloud information according to the true value device positioning information to obtain the first point cloud information.

[0109] The first calculation module 603 is used to calculate a first error value of the point cloud map to be measured according to the preset initial transformation parameters, the first point cloud information and the three-dimensional voxel grid map corresponding to the point cloud map to be measured.

[0110] The second calculation module 604 is used to calculate a second error value of the point cloud map to be measured according to the first error value, the first point cloud information and the point cloud map to be measured.

[0111] The first determination module 605 is used to determine the accuracy error result of the point cloud map to be measured according to the calibration error value of the positioning true value device, the first error value and the second error value.

[0112] In an exemplary embodiment, the true value device positioning information includes three-axis position information and three-axis rotation information; the second acquisition module includes a first conversion submodule, a first construction submodule and a first acquisition submodule.

[0113] The first conversion submodule is used to perform coordinate system conversion processing on the three-axis position information and the three-axis rotation information so that the coordinate systems of the converted three-axis position information, the converted three-axis rotation information and the real-time point cloud information are consistent.

[0114] The first construction submodule is used to construct an affine matrix according to the converted three-axis position information and three-axis rotation information.

[0115] The first acquisition submodule is used to acquire first point cloud information according to the affine matrix and the real-time point cloud information.

[0116] In an exemplary embodiment, the first calculation module includes a second acquisition submodule, a first calculation submodule, a second calculation submodule, a third calculation submodule and a first determination submodule.

[0117] The second acquisition submodule is used to acquire second point cloud information according to the first point cloud information and the initial transformation parameters.

[0118] The first calculation submodule is used to calculate the normal distribution parameters and covariance matrix of each voxel grid in the three-dimensional voxel grid map according to the second point cloud information.

[0119] The second calculation submodule is used to calculate the registration score of the three-dimensional voxel grid map according to the normal distribution parameters and covariance matrix of each voxel grid.

[0120] The third calculation submodule is used to iteratively calculate the registration score according to a preset iterative algorithm until a target registration score not less than a preset threshold is obtained.

[0121] The first determination submodule is used to determine a first error value according to the target registration score and the second point cloud information.

[0122] In an exemplary embodiment, the second calculation module includes a third acquisition submodule and a fourth calculation submodule.

[0123] The third acquisition submodule is used to acquire third point cloud information according to the first error value and the first point cloud information.

[0124] The fourth calculation submodule is used to calculate the second error value according to the position information of each point in the third point cloud information and the position information of each point in the point cloud map to be measured.

[0125] In an exemplary embodiment, the fourth calculation submodule includes a first acquisition unit, a second acquisition unit, a first calculation unit, a second calculation unit, and a third calculation unit.

[0126] The first acquisition unit is used to obtain the first distance value between each point in the third point cloud information and each point in the point cloud map to be tested according to the position information of each point in the third point cloud information and the position information of each point in the point cloud map to be tested, so as to form multiple first distance sequences, wherein the number of first distance sequences is the same as the number of points in the third point cloud information.

[0127] The second acquisition unit is used to obtain the second distance value between each point in the point cloud map to be tested and each point in the third point cloud information according to the position information of each point in the third point cloud information and the position information of each point in the point cloud map to be tested, so as to form multiple second distance sequences, wherein the number of the second distance sequences is the same as the number of points in the point cloud map to be tested.

[0128] The first calculation unit is used to average the minimum values ​​in each first distance sequence to obtain a first average error value.

[0129] The second calculation unit is used to average the minimum values ​​in each second distance sequence to obtain a second average error value.

[0130] The third calculation unit is used to determine the minimum value of the first average error value and the second average error value as the second error value.

[0131] Each module in the above-mentioned point cloud map accuracy measurement device can be implemented in whole or in part by software, hardware and their combination. Each of the above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to each of the above modules.

[0132] In an exemplary embodiment, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the steps of the point cloud map accuracy measurement method described in any of the above embodiments are implemented.

[0133] The computer device may be a terminal, and its internal structure diagram may be shown in FIG7. The computer device includes a processor, a memory, an input / output interface, a communication interface, a display unit, and an input device. The processor, the memory, and the input / output interface are connected via a system bus, and the communication interface, the display unit, and the input device are connected to the system bus via the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be implemented through WIFI, a mobile cellular network, near field communication (NFC) or other technologies. When the computer program is executed by the processor, a point cloud map accuracy measurement method is implemented. The display unit of the computer device is used to form a visually visible picture, which may be a display screen, a projection device, or a virtual reality imaging device. The display screen can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device can be a touch layer covering the display screen, or a button, trackball or touchpad set on the computer device shell, or an external keyboard, touchpad or mouse.

[0134] Those skilled in the art will understand that the structure shown in FIG. 7 is merely a block diagram of a partial structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have a different arrangement of components.

[0135] In an exemplary embodiment, the present application provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the steps of the point cloud map accuracy measurement method described in any of the above embodiments are implemented.

[0136] In an exemplary embodiment, the present application provides a computer program product, including a computer program, which, when executed by a processor, implements the steps of the point cloud map accuracy measurement method described in any of the above embodiments.

[0137] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant regulations.

[0138] A person of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment method can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. Non-relational databases may include distributed databases based on blockchains, etc., but are not limited to this. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, an artificial intelligence (AI) processor, etc., but are not limited to this.

[0139] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0140] The above-described embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.

Claims

1. A method for measuring the accuracy of a point cloud map, characterized in that: Applied to a precision measurement system, the precision measurement system includes at least one positioning true value device and at least one laser radar; the method includes: Acquire the true value device positioning information sent by the positioning true value device and the real-time point cloud information sent by the laser radar; Performing affine transformation processing on the real-time point cloud information according to the true value device positioning information to obtain first point cloud information; Calculating a first error value of the point cloud map to be measured according to a preset initial transformation parameter, the first point cloud information, and a three-dimensional voxel grid map corresponding to the point cloud map to be measured; Calculating a second error value of the point cloud map to be measured according to the first error value, the first point cloud information and the point cloud map to be measured; The accuracy error result of the point cloud map to be measured is determined according to the calibration error value of the positioning true value device, the first error value and the second error value.

2. The method according to claim 1, characterized in that The true value device positioning information includes three-axis position information and three-axis rotation information; performing affine transformation processing on the real-time point cloud information according to the true value device positioning information to obtain first point cloud information includes: Performing coordinate system conversion processing on the three-axis position information and the three-axis rotation information so that the coordinate systems of the converted three-axis position information, the converted three-axis rotation information and the real-time point cloud information are consistent; Constructing an affine matrix according to the converted three-axis position information and three-axis rotation information; First point cloud information is acquired according to the affine matrix and the real-time point cloud information.

3. The method according to claim 1, characterized in that The step of calculating a first error value of the point cloud map to be measured according to a preset initial transformation parameter, the first point cloud information, and a three-dimensional voxel grid map corresponding to the point cloud map to be measured comprises: Acquire second point cloud information according to the first point cloud information and the initial transformation parameters; Calculating normal distribution parameters and covariance matrix of each voxel grid in the three-dimensional voxel grid map according to the second point cloud information; Calculating a registration score of the three-dimensional voxel grid map according to normal distribution parameters and covariance matrices of each voxel grid; Iteratively calculating the registration score according to a preset iterative algorithm until a target registration score not less than a preset threshold is obtained; A first error value is determined according to the target registration score and the second point cloud information.

4. The method according to claim 1, characterized in that The calculating a second error value of the point cloud map to be measured according to the first error value, the first point cloud information and the point cloud map to be measured includes: Acquire third point cloud information according to the first error value and the first point cloud information; A second error value is calculated based on the position information of each point in the third point cloud information and the position information of each point in the point cloud map to be measured.

5. The method according to claim 4, characterized in that The calculating the second error value according to the position information of each point in the third point cloud information and the position information of each point in the point cloud map to be measured includes: According to the position information of each point in the third point cloud information and the position information of each point in the point cloud map to be measured, obtaining a first distance value between each point in the third point cloud information and each point in the point cloud map to be measured, to form a plurality of first distance sequences, wherein the number of the first distance sequences is the same as the number of points in the third point cloud information; According to the position information of each point in the third point cloud information and the position information of each point in the point cloud map to be tested, obtaining a second distance value between each point in the point cloud map to be tested and each point in the third point cloud information to form a plurality of second distance sequences, wherein the number of the second distance sequences is the same as the number of points in the point cloud map to be tested; Averaging the minimum values ​​in each of the first distance sequences to obtain a first average error value; Averaging the minimum values ​​in each of the second distance sequences to obtain a second average error value; A minimum value between the first average error value and the second average error value is determined as a second error value.

6. The method according to claim 2, characterized in that The process of obtaining the three-dimensional voxel grid map corresponding to the point cloud map to be measured includes: The point cloud map to be measured is segmented according to the converted three-axis position information and the preset grid size to obtain a three-dimensional voxel grid map.

7. A point cloud map accuracy measurement device, characterized in that: Applied to an accuracy measurement system, the accuracy measurement system includes at least one positioning true value device and at least one laser radar; the device includes: A first acquisition module is used to acquire the true value device positioning information sent by the positioning true value device and the real-time point cloud information sent by the laser radar; A second acquisition module is used to perform affine transformation processing on the real-time point cloud information according to the true value device positioning information to obtain first point cloud information; A first calculation module, used for calculating a first error value of the point cloud map to be measured according to a preset initial transformation parameter, the first point cloud information and a three-dimensional voxel grid map corresponding to the point cloud map to be measured; A second calculation module, used for calculating a second error value of the point cloud map to be measured according to the first error value, the first point cloud information and the point cloud map to be measured; The first determination module is used to determine the accuracy error result of the point cloud map to be measured according to the calibration error value of the positioning true value device, the first error value and the second error value.

8. 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 according to any one of claims 1 to 6 are implemented.

9. 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 according to any one of claims 1 to 6 are implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Images