Calibration evaluation parameter acquisition method and device, storage medium and electronic device

Abstract

The present invention discloses a method and device for obtaining calibration evaluation parameters, a storage medium, and an electronic device. The method comprises: obtaining current scene point cloud data determined for a target vehicle in a current scene, wherein the current scene point cloud data includes positioning results of the target vehicle calibration in at least two calibration coordinate systems; segmenting the current scene point cloud data to obtain multiple grid point cloud data; determining target grid point cloud data from the multiple grid point cloud data based on the geometric features corresponding to each grid point cloud data in the multiple grid point cloud data, wherein the geometric features of the target grid point cloud data indicate that a reference surface is included in the grid area; and performing statistics on the geometric features of the target grid point cloud data to generate calibration evaluation parameters matching the current scene point cloud data. The present invention solves the technical problem of low evaluation accuracy due to subjective evaluation.

Description

Technical Field

[0001] The present invention relates to the field of computers, and in particular to a method and device for obtaining calibration evaluation parameters, a storage medium, and an electronic device. Background Art

[0002] In order to ensure the driver's driving safety, various sensors are often used in vehicle assisted driving systems to measure the vehicle's motion state, so as to provide the driver with safe driving prompt information.

[0003] Because different sensors are installed at different angles or positions within the vehicle, they actually use different coordinate systems. Therefore, to determine the vehicle's current trajectory based on the positioning data collected by different sensors, it is necessary to integrate and calibrate the positioning data collected in different coordinate systems into the vehicle's own coordinate system. However, the calibration algorithm often cannot know the actual values before calibration, resulting in calibration errors during the process, and these errors cannot be corrected in a timely manner.

[0004] To determine this calibration error, a common method currently used in related technologies involves subjective observation by experienced evaluation experts, who then combine their own experience to determine whether the calibration results contain calibration errors. This subjective evaluation method relies solely on the expert's visual observation and personal experience, making it difficult to guarantee the objective accuracy of the evaluation results.

[0005] To address the above-mentioned problems, no effective solutions have been proposed so far. Summary of the Invention

[0006] Embodiments of the present invention provide a calibration evaluation parameter acquisition method and apparatus, a storage medium, and an electronic device to at least solve the technical problem of low accuracy of evaluation results caused by subjective evaluation.

[0007] According to one aspect of an embodiment of the present invention, a calibration evaluation parameter acquisition method is provided, comprising: acquiring current scene point cloud data determined by a target vehicle in a current scene, wherein the current scene point cloud data includes positioning results of the target vehicle calibrated in at least two calibration coordinate systems; segmenting the current scene point cloud data to obtain a plurality of grid point cloud data; determining target grid point cloud data from the plurality of grid point cloud data based on geometric features corresponding to each of the plurality of grid point cloud data, wherein the geometric features of the target grid point cloud data indicate that a reference surface is included in a grid area; and performing statistics on the geometric features of the target grid point cloud data to generate calibration evaluation parameters matching the current scene point cloud data.

[0008] According to another aspect of an embodiment of the present invention, a calibration evaluation parameter acquisition device is also provided, including: an acquisition module for acquiring current scene point cloud data determined by a target vehicle in a current scene, wherein the current scene point cloud data includes positioning results of the target vehicle calibration under at least two calibration coordinate systems; a segmentation module for segmenting the current scene point cloud data to obtain a plurality of grid point cloud data; a determination module for determining target grid point cloud data from the plurality of grid point cloud data based on the geometric features corresponding to each of the plurality of grid point cloud data, wherein the geometric features of the target grid point cloud data indicate that a reference surface is included in the grid area; a generation module for performing statistics on the geometric features of the target grid point cloud data to generate calibration evaluation parameters matching the current scene point cloud data.

[0009] According to another aspect of an embodiment of the present invention, a computer-readable storage medium is provided, in which a computer program is stored. The computer program is configured to execute the above-mentioned calibration evaluation parameter acquisition method when running.

[0010] According to another aspect of an embodiment of the present invention, an electronic device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor is configured to execute the calibration evaluation parameter acquisition method through the computer program.

[0011] In an embodiment of the present invention, the current scene point cloud data formed by the positioning results under at least two calibration coordinate systems is segmented to obtain multiple grid point cloud data, target grid point cloud data containing a reference surface is determined based on the geometric features of the grid point cloud data, and calibration evaluation parameters are statistically generated based on the geometric features of the target grid point cloud data. The target grid point cloud data containing the reference surface is obtained from the current scene point cloud data, and calibration evaluation parameters are generated based on the geometric features of the target grid point cloud data. This objectively analyzes and statistics the geometric features of the scene point cloud data to obtain calibration evaluation parameters for objectively evaluating the calibration measurement results, rather than relying on the subjective experience and judgment of professionals. This allows the vehicle calibration measurement process to be evaluated using objectively measurable calibration evaluation parameters, thereby improving the objective accuracy of the calibration evaluation and overcoming the problem of low calibration evaluation accuracy caused by subjective judgment in related technologies. Furthermore, the solution provided in the embodiment of the present application uses objective pricing indicators to measure the calibration measurement process, eliminating the need for manual evaluation and also helping to improve the efficiency of calibration measurement. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] The drawings described herein are used to provide a further understanding of the present invention and constitute a part of this application. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings:

[0013] Figure 1 is a schematic diagram of an application environment of an optional calibration evaluation parameter acquisition method according to an embodiment of the present invention;

[0014] Figure 2 is a flow chart of an optional method for obtaining calibration evaluation parameters according to an embodiment of the present invention;

[0015] Figure 3 is a flowchart of another optional calibration evaluation parameter acquisition method according to an embodiment of the present invention;

[0016] Figure 4 is a flowchart of another optional calibration evaluation parameter acquisition method according to an embodiment of the present invention;

[0017] Figure 5 is a flowchart of another optional calibration evaluation parameter acquisition method according to an embodiment of the present invention;

[0018] Figure 6 is a flowchart of another optional calibration evaluation parameter acquisition method according to an embodiment of the present invention;

[0019] Figure 7 is a flowchart of another optional calibration evaluation parameter acquisition method according to an embodiment of the present invention;

[0020] Figure 8 is a flowchart of another optional calibration evaluation parameter acquisition method according to an embodiment of the present invention;

[0021] Figure 9 is a flowchart of another optional calibration evaluation parameter acquisition method according to an embodiment of the present invention;

[0022] Figure 10 is a flowchart of another optional calibration evaluation parameter acquisition method according to an embodiment of the present invention;

[0023] Figure 11 is a schematic diagram of current scene point cloud data according to an embodiment of the present invention;

[0024] Figure 12 is a structural diagram of an optional calibration evaluation parameter acquisition device according to an embodiment of the present invention;

[0025] Figure 13 FIG. 4 is a schematic structural diagram of an optional electronic device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0026] In order to enable those skilled in the art to better understand the solutions of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the embodiments described 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 should fall within the scope of protection of the present invention.

[0027] It should be noted that the terms "first", "second", etc. in the description and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that the numbers used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0028] To make the scheme clear, the key terms involved are explained below:

[0029] Lidar: A sensing device used for 3D perception and positioning, widely used in L4 autonomous driving.

[0030] Inertial Measurement Unit (IMU): A device that measures its own angular velocity and linear acceleration in the sensor coordinate system.

[0031] Global Positioning System (GPS): This mainly refers to the GPS receiving device.

[0032] Integrated navigation: Combined use of IMU+GPS to obtain high-frequency and high-precision global positioning results. The output coordinate system is generally the IMU coordinate system.

[0033] According to one aspect of an embodiment of the present invention, a calibration evaluation parameter acquisition method is provided. Optionally, as an optional implementation, the calibration evaluation parameter acquisition method can be applied to, but is not limited to, Figure 1In the calibration evaluation parameter acquisition system in the network environment shown, the calibration evaluation parameter acquisition system includes: a terminal device 102, a network 104 and a server 106. A target application client (such as a calibration measurement application client, etc.) is running in the terminal device 102. The above-mentioned terminal device 102 can be, but is not limited to, located in the target vehicle 100, and includes a human-computer interaction screen 1022, a processor 1024 and a memory 1026. The human-computer interaction screen 1022 is used to present the calibration evaluation parameter acquisition results provided in the above-mentioned target application client; it is also used to provide a human-computer interaction interface to receive human-computer interaction operations performed on the human-computer interaction interface; the processor 1024 is used to respond to the above-mentioned human-computer interaction operations to obtain human-computer interaction instructions, triggering the start of the calibration evaluation parameter acquisition process. The memory 1026 is used to store the above-mentioned calibration evaluation parameter acquisition results.

[0034] In addition, the server 106 includes a database 1062 and a processing engine 1064. The database 1062 is used to store the above-mentioned grid point cloud data and their corresponding geometric features. The processing engine 1064 is used to generate calibration evaluation parameters based on the above-mentioned geometric features.

[0035] The specific process is as follows: Assuming that a calibration measurement application client is running in the terminal device 102, the calibration measurement application client will use the method provided in the embodiment of the present application to obtain objective evaluation parameters of the calibration result, specifically as shown in steps S102-S112:

[0036] The terminal device 102 obtains the current scene point cloud data of the target vehicle 100 in the current scene and transmits the obtained current scene point cloud data to the server 106 via the network 104. The server 106 stores the received current scene point cloud data in the database 1062 and activates the processing engine 1064 to segment the current scene point cloud data to obtain multiple grid point cloud data. Based on the geometric features corresponding to each grid point cloud data in the multiple grid point cloud data, the server 106 determines the grid point cloud data in the grid region where the geometric features indicate the reference surface as the target grid point cloud data. After determining the target point cloud data, the server 106 calculates the geometric features of the target grid point cloud data to generate calibration evaluation parameters for matching the current scene point cloud data. The server 106 then returns the calibration evaluation parameters for matching the current scene point cloud data to the terminal device 102 via the network 104. The terminal device 102 stores the received calibration evaluation parameters in the memory 1026, and the processor 1024 determines whether to adjust the posture of the calibration measurement device (such as the terminal device 102 and its associated devices) in the target vehicle used to collect the above-mentioned scene point cloud data.

[0037] It should be noted that in this embodiment, after the current scene point cloud data is determined in the current scene for target estimation and measurement, the current scene point cloud data is segmented to obtain multiple grid point cloud data. Based on the geometric features corresponding to each grid point cloud data in the multiple grid point cloud data, target grid point cloud data, including the reference surface, is determined from the multiple grid point cloud data. The geometric features of the target grid point cloud data are then statistically analyzed to generate calibration evaluation parameters that match the current scene point cloud data. In other words, by objectively analyzing and statistically analyzing the geometric features of the scene point cloud data, calibration evaluation parameters are obtained for objectively evaluating the calibration measurement results, rather than relying on the subjective experience of professionals. This allows the vehicle calibration measurement process to be evaluated using objectively measurable calibration evaluation parameters, thereby improving the objective accuracy of the calibration evaluation and overcoming the low accuracy of calibration evaluations caused by subjective judgment in related technologies. Furthermore, the solution provided in the embodiments of this application uses objective pricing indicators to measure the calibration measurement process, eliminating the need for manual evaluation and further improving the efficiency of calibration measurement.

[0038] Optionally, in this embodiment, the terminal device 102 may be a device configured with a target client for acquiring the current scene point cloud data, and may include but is not limited to at least one of the following: a mobile phone (such as an Android phone, an IOS phone, etc.), a laptop, a tablet computer, a PDA, a MID (Mobile Internet Devices), a PAD, a desktop computer, a smart TV, etc. The target client may be a calibration measurement application client, or other client that can be used to measure the motion state parameters of a moving vehicle, etc. The network 104 may include but is not limited to: a wired network, a wireless network, wherein the wired network includes: a local area network, a metropolitan area network, and a wide area network, and the wireless network includes: Bluetooth, WIFI, and other networks that implement wireless communication. The server 106 may be a single server, or a server cluster consisting of multiple servers, or a cloud server. The above is only an example, and this embodiment does not impose any limitation on this.

[0039] As an optional implementation, Figure 2 As shown, the above calibration evaluation parameter acquisition method includes:

[0040] S202, obtaining current scene point cloud data determined for the target vehicle in the current scene, wherein the current scene point cloud data includes positioning results of the target vehicle calibrated in at least two calibration coordinate systems;

[0041] S204, segmenting the current scene point cloud data to obtain a plurality of grid point cloud data;

[0042] S206, determining target grid point cloud data from the plurality of grid point cloud data based on geometric features corresponding to each grid point cloud data in the plurality of grid point cloud data, wherein the geometric features of the target grid point cloud data indicate that the grid area includes a reference surface;

[0043] S208 , performing statistics on the geometric features of the target grid point cloud data to generate calibration evaluation parameters for matching the current scene point cloud data.

[0044] Optionally, the calibration evaluation parameter acquisition method described above may be applied to, but is not limited to, a vehicle calibration evaluation process. The vehicle herein may be, but is not limited to, a real vehicle in a real scene, or a virtual vehicle controlled in a virtual scene (e.g., a virtual vehicle in a game application).

[0045] In an embodiment of the present application, in the process of positioning the target vehicle using a laser radar and an inertial measurement unit in the related art, since the coordinate systems used by the two are different, it is necessary to uniformly calibrate the coordinate systems of the two. However, due to the different positions or postures of the calibration measurement equipment (or devices), the calibration results will be quite different. In order to overcome the above problems, it is usually necessary to adjust and correct the positions or postures of the calibration measurement equipment (or devices) in different coordinate systems based on the calibration results. However, the commonly used method at present can only be subjective observation and judgment, and in an embodiment of the present application, a method for obtaining objective and measurable calibration evaluation parameters is provided, by objectively analyzing and counting the geometric features of the scene point cloud data to obtain calibration evaluation parameters for objectively evaluating the calibration measurement results, instead of relying on the subjective experience judgment of professionals, thereby achieving objective evaluation of different calibration processes through the calibration evaluation parameters.

[0046] Optionally, in this embodiment, the current scene point cloud data determined by the target vehicle in the current scene may be, but is not limited to, the current scene point cloud data determined by a vehicle operating on an actual road during driving, or the current scene point cloud data determined by a vehicle operating on a target client during driving. The positioning results of the target vehicle under at least two calibrated coordinate systems included in the current scene point cloud data may be positioning results simultaneously obtained by each calibrated coordinate system during a single driving process of the target vehicle in the current scene, or positioning results obtained by each calibrated coordinate system based on a time axis during multiple driving processes of the target vehicle in the current scene.

[0047] Optionally, segmenting the current scene point cloud data may be performed according to a preset segmentation standard or preset segmentation condition. The preset segmentation standard may include, but is not limited to, a segmentation scale determined based on scene objects in the current scene, a segmentation scale determined based on the scale of the current scene, a segmentation scale determined based on the data distribution of the current scene point cloud data, or a segmentation scale determined based on the data volume of the current scene point cloud data. The preset segmentation condition may include, but is not limited to, a segmentation condition set based on the current scene, a segmentation condition set based on scene objects in the current scene, or a segmentation condition set based on the point cloud data.

[0048] Optionally, the segmentation condition set according to the scene objects in the current scene may be, but is not limited to, a segmentation condition that includes the scene objects or a segmentation condition that does not include the scene objects.

[0049] Optionally, when the segmentation scale is determined based on the scene objects of the current scene, the segmentation scale can be a scale determined based on the ratio of the size of the scene objects to the size of the current scene, a scale determined based on the maximum size of the scene objects, a scale determined based on the minimum size of the scene objects, or a scale determined based on the average size of the scene objects.

[0050] Optionally, when the segmentation scale is determined based on the scale of the current scene, the segmentation scale may be a scale for segmenting the current scene point cloud data into two or more raster point cloud data. The scale for segmenting the current scene point cloud data into two or more raster point cloud data is not limited to a scale for segmenting the current scene point cloud data into two or more parts in each dimension. The segmentation scales in each dimension may be different, but the segmentation scale for the current scene point cloud data remains consistent during the segmentation process.

[0051] Optionally, in this embodiment, the point cloud data of the current scene may be segmented according to, but is not limited to, two or more scales. For example, taking the scale of the current scene as 500 meters on both the X and Y axes and 9 meters on the Z axis, to ensure the amount of grid data and the number of grids, the segmentation scale is set to 20 meters on both the X and Y axes and 3 meters on the Z axis, ensuring that the number of grids in the same dimension is not unique.

[0052] Through the embodiments provided in this application, based on the specific circumstances of the current scene and the different amounts of data contained in the point cloud data of the current scene, different limitations are imposed on the segmentation conditions, thereby making the segmentation basis more reasonable, so that the raster point cloud data formed by the segmentation can have better geometric features, so as to facilitate the determination of target raster point cloud data based on the geometric features.

[0053] As an optional implementation, Figure 3 As shown, the current scene point cloud data is segmented to obtain multiple grid point cloud data including:

[0054] S302, extracting scene objects contained in the current scene based on the current scene point cloud data;

[0055] S304, determining a target scene object from the scene objects included in the current scene, wherein the target scene object is the scene object with the smallest size;

[0056] S306, determining a segmentation scale corresponding to the current scene point cloud data based on the size of the target scene object;

[0057] S308 , performing segmentation according to a segmentation scale to obtain a plurality of grid point cloud data.

[0058] Optionally, the scene object may be an object included in the current scene. In the case where the current scene indicates a scene of a vehicle driving process, the scene object may be all objects included in the current scene.

[0059] Optionally, the scene object with the smallest size may be, but is not limited to, an object with the smallest size in a three-dimensional volume, an object with the smallest size in a set dimension, or an object with the smallest size in a set plane.

[0060] Optionally, the segmentation scale corresponding to the current scene point cloud data determined based on the size of the target scene object may be, but is not limited to, the minimum standard scale that includes the target scene object. The standard scale for the segmentation scale may be, but is not limited to, set according to the segmentation threshold, so that the size of the segmented raster point cloud data is more conducive to the implementation of subsequent methods.

[0061] Optionally, the minimum standard scale that includes the target scene object may be, but is not limited to, determining the minimum standard scale that includes the target scene object in each dimension, and determining the minimum standard scale that includes the target scene object based on the minimum standard scale determined in each dimension. For example, if the segmentation threshold is 1 meter and the dimensions of the target scene object in a three-dimensional coordinate system are [2.3, 1.6, 0.8], then the minimum standard scales that include the target scene object in the coordinate system are 3, 2, and 1, respectively. Therefore, the segmentation scale corresponding to the target scene object size is [3, 2, 1].

[0062] In an embodiment of the present application, the current scene point cloud data is segmented by setting a segmentation scale, and the multiple grid point cloud data obtained by segmentation are calculated one by one, thereby reducing the amount of computational data required to obtain geometric features each time. The smaller amount of computational data ensures the accuracy of each calculation result, thereby accurately obtaining the geometric features of each grid point cloud data.

[0063] Optionally, the geometric features may be, but are not limited to, utilizing data features in the raster point cloud data. The data features may be, but are not limited to, being obtained through data analysis methods. The data analysis methods used may be, but are not limited to, Locally Linear Embedding (LLE), Linear Discriminant Analysis (LDA), Principal Components Analysis (PCA), and Laplacian Eigenmaps.

[0064] Locally Linear Embedding (LLE) is a nonlinear dimensionality reduction algorithm that maps three-dimensional data to two-dimensional data while maintaining the original data flow structure. It finds k adjacent data points near each sample data point, calculates the local reconstruction weight matrix of each sample data point from its adjacent data points, and obtains the output data point from the local reconstruction weight matrix of the sample data point and its adjacent data points.

[0065] Linear Discriminant Analysis (LDA) is a supervised linear dimensionality reduction algorithm that calculates the mapping vector of the original data so that the mapped data has data points of the same type as close as possible and data points of different types as separate as possible, thereby mapping the original data into two data sets with different characteristics.

[0066] Principal Components Analysis (PCA) is a linear dimensionality reduction algorithm. It removes the mean value of the original data, calculates the covariance matrix and the matrix eigenvalues and eigenvectors, sorts the eigenvalues in descending order, and retains the eigenvectors corresponding to the first N eigenvalues. The original data is converted into a new space constructed by N eigenvectors to map the original data into an N-dimensional data set, remove the interfering dimensions in the original data, and achieve data dimensionality reduction.

[0067] Laplacian Eigenmaps are a graph-based dimensionality reduction algorithm that constructs data relationships from a local perspective. By constructing a graph from all sample data points, connecting each point to nearby points and determining the weights between them, the eigenvectors and eigenvalues of the Laplacian matrix are calculated. The eigenvectors corresponding to the smallest m non-zero eigenvalues are used as the output to achieve data dimensionality reduction.

[0068] As an optional implementation, Figure 4As shown, according to the geometric features corresponding to each grid point cloud data in the plurality of grid point cloud data, determining the target grid point cloud data from the plurality of grid point cloud data includes:

[0069] S402, performing principal component analysis on each of the plurality of grid point cloud data in sequence to extract a point cloud feature set that matches each of the grid point cloud data, wherein the point cloud feature set is used to determine geometric features corresponding to the grid point cloud data;

[0070] S404 : determining target grid point cloud data from the plurality of grid point cloud data according to the point cloud feature set matched to each grid point cloud data.

[0071] Optionally, the principal component analysis may be performed using, but is not limited to, principal component analysis (PCA) technology.

[0072] Optionally, the geometric features corresponding to the grid point cloud data determined by the point cloud feature set may be used, but not limited to, to indicate whether the grid point cloud data includes a reference surface and to indicate the normal distribution of the grid point cloud data. The reference surface may be, but not limited to, a plane, a curved surface, or a sphere.

[0073] In an embodiment of the present application, principal component analysis is used to extract a point cloud feature set indicating geometric features from raster point cloud data with a larger data dimension, and a point cloud feature set with a smaller data dimension is used to analyze the geometric features of the raster point cloud data, so that the target raster point cloud features containing the reference surface can be quickly determined from multiple raster point cloud data. By quickly determining the geometric features, the efficiency of determining the target raster point cloud data is improved.

[0074] As an optional implementation, principal component analysis is performed on each of the plurality of grid point cloud data in turn to extract a point cloud feature set that matches each grid point cloud data, including:

[0075] Each grid point cloud data in the multiple grid point cloud data is taken as the current grid point cloud data in turn, and the following is executed: Figure 5 The following operations are shown:

[0076] S502, constructing a point cloud feature matrix based on the coordinates of each point in the current grid point cloud data;

[0077] S504, calculating the point cloud covariance matrix corresponding to the point cloud feature matrix;

[0078] S506 , performing eigenvalue decomposition on the point cloud covariance matrix to obtain a current point cloud feature set that matches the current grid point cloud data.

[0079] Optionally, the point cloud feature matrix constructed based on the coordinates of each point in the current grid point cloud data can be, but is not limited to, a point cloud feature matrix composed of each point coordinate minus the average value of the point coordinates in each dimension. The point cloud feature matrix can be expressed as:

[0080]

[0081] Where n represents the number of coordinate points contained in the grid point cloud data, 3 represents the coordinate matrix based on 3D coordinate data, μ is the average coordinate of n points, and μ is expressed as:

[0082] μ=[μ x μ y μ z ] (2)

[0083] Among them, μ1, μ2, and μ3 are the average values calculated for n points in each dimension in the 3-dimensional coordinate system.

[0084] Optionally, the point cloud covariance matrix corresponding to the point cloud feature matrix can be calculated by, but is not limited to, the point cloud covariance matrix and the transposed matrix of the point cloud feature matrix. The calculation method of the point cloud covariance matrix can be expressed as:

[0085]

[0086] in, For X n×3 The transposed matrix of .

[0087] Optionally, the point cloud feature set may be obtained by, but is not limited to, the following calculation methods:

[0088]

[0089]

[0090] Among them, M is the matrix representation of the point cloud feature set.

[0091] Optionally, the point cloud feature set is a data set containing [λ1, λ2, λ3].

[0092] In an embodiment of the present application, a point cloud feature matrix is constructed and a covariance matrix is calculated to obtain a point cloud feature set indicating geometric features through matrix decomposition. The matrix calculation is simple and accurate, so that the point cloud feature set can be accurately and quickly extracted, thereby improving the accuracy of determining the target grid point cloud data and improving the accuracy of the calibration parameters.

[0093] As an optional implementation, determining target grid point cloud data from a plurality of grid point cloud data according to a point cloud feature set matched to each grid point cloud data includes:

[0094] When the difference between the target point cloud feature value included in the current point cloud feature set and zero is less than the target threshold, and the difference between the point cloud feature values other than the target point cloud feature value in the current point cloud feature set and zero is greater than the target threshold, the geometric features determined based on the current point cloud feature set will indicate that the current grid area where the current raster point cloud data is located includes a plane or a curved surface, and the current raster point cloud data is determined to be the target raster point cloud data.

[0095] Optionally, the target point cloud feature value may be, but is not limited to, the maximum value data determined based on all data included in the point cloud feature set. The target point cloud feature value may be, but is not limited to, the data with the smallest value in the point cloud feature set.

[0096] Optionally, the target threshold may be, but is not limited to, a threshold determined based on the maximum value of all data values contained in the point cloud feature set, a threshold determined based on the average value of all data values, or a threshold determined based on sorting of all data values according to data size.

[0097] Optionally, when the threshold is determined based on the data value sorting of all data values from small to large, the target threshold may be, but is not limited to, any value between the data value with the first rank and the data value with the second rank.

[0098] Optionally, the feature data with the smallest data value among all the data values included in the point cloud feature set is used as the target point cloud feature value. If the value of the target point cloud feature value is close to 0, and the values of the other values except the target point cloud feature value are not close to 0, the current grid point cloud data is determined as the target grid point cloud data.

[0099] In an embodiment of the present application, a point cloud feature matrix is obtained through raster point cloud data, a point cloud covariance matrix is calculated, and the eigenvalues of the point cloud covariance matrix are decomposed to obtain a point cloud feature set including the eigenvalues, thereby obtaining a feature data set for representing the raster point cloud data, so as to determine the target point cloud data from multiple raster point cloud data based on the feature data values of the raster point cloud data.

[0100] As an optional implementation, Figure 6 As shown in the figure, the geometric features of the target grid point cloud data are statistically analyzed to generate calibration evaluation parameters for matching the current scene point cloud data, including:

[0101] S602, determining a reference thickness of a reference surface included in a grid area corresponding to each of the target grid point cloud data according to geometric features of the target grid point cloud data, wherein the reference surface includes a plane or a curved surface;

[0102] S604: Count the reference thicknesses corresponding to all target grid point cloud data to generate calibration evaluation parameters.

[0103] Optionally, the geometric features of the target grid point cloud data may be the geometric features indicated by the point cloud feature set of the target grid point cloud, the geometric features of the reference surface determined by the target grid point cloud data, and the geometric features of the target grid point cloud data compared with the reference surface.

[0104] In an embodiment of the present application, the reference thickness of the reference surface is determined by the geometric features of the target grid point cloud data, and calibration evaluation parameters are generated based on the statistically analyzed reference thickness, so as to achieve objective evaluation data indicating geometric features through the geometric features of the target data that are representative of the data, and use the objective data to evaluate the calibration results of the point cloud data to improve the accuracy of the calibration parameters.

[0105] As an optional implementation manner, determining the thickness of the reference surface in the grid area corresponding to each target grid point cloud data according to the geometric characteristics of the target grid point cloud data includes:

[0106] like Figure 7 As shown, the following operations are performed on each target grid point cloud data:

[0107] S702, determining a fitting surface in a grid area corresponding to the target grid point cloud data;

[0108] S704, obtaining the distance between each point in the target grid point cloud data and the fitting surface;

[0109] S706, determining the thickness of the fitting surface according to the average value of the distances;

[0110] S708 : Using the thickness of the fitting surface as the thickness of the reference surface in the grid area corresponding to the target grid point cloud data.

[0111] Optionally, the fitting surface determined in the grid area corresponding to the target grid point cloud data may be, but is not limited to, a fitting surface obtained by plane fitting using the least squares method. The reference surface may be a fitting surface obtained using the least squares method.

[0112] In an embodiment of the present application, the fitting surface corresponding to the target grid point cloud data is determined based on the geometric characteristics of the target grid point cloud data, thereby obtaining the distance between each point in the target grid point cloud data and the fitting surface, and the average distance value of all point cloud data is used as the thickness of the reference surface of the grid area corresponding to the target grid point cloud data. The fitting surface is determined by the target grid point cloud data, and the reference surface thickness is calculated to obtain the thickness corresponding to the grid area, and further obtain calibration evaluation parameters for evaluating the scene point cloud data, and the calibration evaluation parameters are used to perform calibration evaluation to improve the accuracy of the calibration evaluation.

[0113] As an optional implementation, Figure 8As shown, before determining the fitting surface in the grid area corresponding to the target grid point cloud data, the following steps are also included:

[0114] S802, determining a candidate fitting surface based on the target grid point cloud data;

[0115] S804: Remove discrete points whose distances from the candidate fitting surface are greater than a target distance threshold to update the target grid point cloud data.

[0116] Optionally, determining the candidate fitting surface based on the target grid point cloud data may include, but is not limited to, using data screening to determine the candidate fitting surface based on the screened data, and using an algorithm to determine the data set corresponding to the candidate fitting surface to determine the candidate fitting surface based on the data set.

[0117] Optionally, the data set corresponding to the candidate fitting surface is determined by using an algorithm, which may include, but is not limited to, using a Random Sample Consensus (RANSAC) algorithm to determine the data set included in the candidate fitting surface.

[0118] The Random Sample Consensus (RANSAC) algorithm is a non-deterministic algorithm that iteratively estimates a mathematical model from a set of observed data that contains outliers. By assuming that outliers are unsuitable for modeling, the remaining data is fitted into the same data model, resulting in a reasonable output with a certain probability.

[0119] In an embodiment of the present application, candidate fitting surfaces are calculated for the target grid point cloud data before determining the fitting surface and data points far away from the candidate fitting surface are removed to reduce the influence of edge data on the fitting surface, thereby improving the accuracy of fitting surface determination and improving the accuracy of calibration parameters.

[0120] Optionally, the grid area corresponding to the target grid point cloud data may be a grid area formed by dividing the target grid point cloud data with respect to the current scene point cloud data. Each grid area may contain one target grid point cloud data.

[0121] Optionally, the grid area division can be, but is not limited to, balanced division based on the amount of grid point cloud data contained, so that the amount of grid point cloud data contained in multiple grid areas is balanced while ensuring that each grid area only contains one target grid point cloud data.

[0122] Optionally, when the target grid point cloud data are adjacent grid point cloud data, the area where each target grid point cloud data is located is divided into a grid area. When the target grid point cloud data are not adjacent grid point cloud data, the grid area corresponding to each target grid point cloud data is determined.

[0123] Optionally, the calibration evaluation parameter for matching the current scene point cloud data may be an average thickness calculated based on the average thickness of the reference surfaces corresponding to all grid areas.

[0124] In an embodiment of the present application, grid area attribution is set for multiple grid point cloud data divided into the current scene point cloud data based on the target grid data, and the fitting surface thickness of the target grid point cloud data is used as the reference surface thickness of the grid area. The average thickness of multiple grid areas is used as the calibration evaluation parameter of the current scene point cloud data to evaluate the marking effect of the current scene point cloud data, which makes the evaluation data of the current scene point cloud data more representative and improves the accuracy of the calibration.

[0125] As an optional implementation, Figure 9 As shown, obtaining the current scene point cloud data determined by the target vehicle in the current scene includes:

[0126] S902, obtaining a point cloud positioning result and an integrated navigation positioning result collected by the target vehicle in the current scene, wherein the point cloud positioning result is a positioning result obtained based on point cloud data collected by a first sensor device in the target vehicle, and the integrated navigation positioning result is a positioning result obtained by combining multiple motion data of the target vehicle collected by a second sensor device in the target vehicle;

[0127] S904: splicing the point cloud positioning result and the combined navigation positioning result to obtain the current scene point cloud data corresponding to the target vehicle.

[0128] Optionally, the point cloud positioning result may be, but is not limited to, a distribution result of point cloud data in a coordinate system established based on the first sensing device. The first sensing device may be, but is not limited to, include multiple sensing sub-devices using the same coordinate system.

[0129] Optionally, the first sensing device may be, but is not limited to, a laser radar device, a sensing device based on stereo perception and positioning. The coordinate system corresponding to the first sensing device may be a coordinate system used by a laser radar.

[0130] Optionally, the combined navigation positioning result may be, but is not limited to, a distribution result of positioning data in a coordinate system established based on the second sensing device. The second sensing device may be, but is not limited to, include multiple sensing sub-devices using the same coordinate system.

[0131] Optionally, the second sensing device may be, but is not limited to, a positioning sensing device based on an inertial measurement unit and a global positioning system. The coordinate system corresponding to the second sensing device may be a coordinate system used by the inertial measurement unit.

[0132] Optionally, the point cloud positioning results and the combined navigation positioning results can be spliced by determining the relative relationship between the coordinate system corresponding to the first sensing device and the coordinate system corresponding to the second sensing device, and displaying the point cloud positioning results and the combined navigation positioning results in the same coordinate system.

[0133] In an embodiment of the present application, by splicing the point cloud positioning result determined by the first sensing device and the combined navigation positioning result determined by the second sensing device, the current scene point cloud data corresponding to the current scene is obtained, and the positioning results displayed by different sensing devices in different coordinate systems are spliced to generate scene point cloud data displayed in the same coordinate system, thereby obtaining a more accurate positioning result.

[0134] Optionally, the method for calibrating the evaluation parameters can be as follows: Figure 10 As shown. In the case that radar navigation and combined navigation exist in the target vehicle, execute S1002 to obtain the point cloud positioning result, and execute S1004 to obtain the combined navigation positioning result. In the case that the point cloud positioning result and the combined navigation positioning result are obtained, execute S1006 to splice the point cloud positioning result and the combined navigation positioning result to generate the current scene point cloud data. The current scene point cloud data can be as follows Figure 11 As shown, Figure 11 The point cloud data in the image is densely distributed and connected to form a plane in some areas. Such areas are areas where reference planes may exist 1102. Current technology uses the naked eye to evaluate the calibration results of scene point cloud data based on areas 1102 in the current scene point cloud data where reference screens may exist, resulting in low evaluation accuracy.

[0135] When the current scene point cloud data is acquired, S1008 is executed to segment the current scene point cloud data. After the current scene point cloud data is segmented according to the standard size corresponding to the smallest scene object contained therein, S1010 is executed to acquire the multiple raster point cloud data formed by the segmentation. After the scene point cloud data is rasterized, S1012 is executed to determine whether the geometric features of the raster point cloud data indicate that it is the target raster point cloud data.

[0136] To determine the geometric features of the raster point cloud data, a point cloud feature matrix is constructed based on the raster point cloud data, and the point cloud covariance matrix is calculated. The point cloud covariance matrix is then subjected to eigendecomposition to obtain a point cloud feature set. If the point cloud feature set contains only one target eigenvalue close to 0 and the remaining eigenvalues are not close to 0, the set features of the raster point cloud data are determined to indicate that it is not the target raster point cloud data. If the features of the point cloud feature set do not meet the above characteristics, the raster point cloud data is determined not to be the target raster point cloud data.

[0137] If the result of the judgment in S1012 is yes, that is, the geometric characteristics of the grid point cloud data indicate that it is the target grid point cloud data, S1014 is executed to determine that the grid point cloud data is the target grid point cloud data participating in the evaluation. If the result of the judgment in S1012 is no, that is, the geometric characteristics of the grid point cloud data indicate that it is not the target grid point cloud data, S1016 is executed to determine that the grid point cloud data is grid point cloud data not participating in the evaluation.

[0138] When the target grid point cloud data to be evaluated is determined, S1018 is executed to calculate the reference surface thickness of the target grid point cloud data. The reference surface thickness of the target grid point cloud data is calculated by determining the fitting surface corresponding to the target grid point cloud data, calculating the distance from each point in the grid point cloud data to the fitting surface, and taking the average distance from all points in the target grid point cloud data to the fitting surface as the reference surface thickness of the target grid point cloud data.

[0139] When the reference surface thickness of the target grid point cloud data is obtained, S1020 is executed to generate a calibration evaluation parameter based on the reference surface thickness of the target grid point cloud data. The reference surface thickness of all target grid point cloud data is counted to obtain the reference surface thickness of the current scene point cloud data, and the reference surface thickness of the current scene point cloud data is used as the evaluation calibration parameter.

[0140] It should be noted that for the aforementioned method embodiments, for simplicity of description, they are all expressed as a series of action combinations. However, those skilled in the art should be aware that the present invention is not limited by the order of the actions described, because according to the present invention, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in this specification are all preferred embodiments, and the actions and modules involved are not necessarily required by the present invention.

[0141] According to another aspect of the embodiments of the present invention, there is also provided a calibration evaluation parameter acquisition device for implementing the above calibration evaluation parameter acquisition method. Figure 12 As shown, the device includes:

[0142] An acquisition module 1202 is configured to acquire current scene point cloud data of a target vehicle determined in the current scene, wherein the current scene point cloud data includes positioning results of the target vehicle calibrated in at least two calibration coordinate systems;

[0143] The segmentation module 1204 is used to segment the current scene point cloud data to obtain a plurality of grid point cloud data;

[0144] a determination module 1206 for determining target grid point cloud data from the plurality of grid point cloud data based on geometric features corresponding to each grid point cloud data in the plurality of grid point cloud data, wherein the geometric features of the target grid point cloud data indicate that the grid area includes the reference surface;

[0145] The generation module 1208 is used to perform statistics on the geometric features of the target grid point cloud data to generate calibration evaluation parameters for matching the current scene point cloud data.

[0146] In an embodiment of the present application, in the process of positioning the target vehicle using a laser radar and an inertial measurement unit in the related art, since the coordinate systems used by the two are different, it is necessary to uniformly calibrate the coordinate systems of the two. However, due to the different positions or postures of the calibration measurement equipment (or devices), the calibration results will be quite different. In order to overcome the above problems, it is usually necessary to adjust and correct the positions or postures of the calibration measurement equipment (or devices) in different coordinate systems based on the calibration results. However, the commonly used method at present can only be subjective observation and judgment, and in an embodiment of the present application, a method for obtaining objective and measurable calibration evaluation parameters is provided, by objectively analyzing and counting the geometric features of the scene point cloud data to obtain calibration evaluation parameters for objectively evaluating the calibration measurement results, instead of relying on the subjective experience judgment of professionals, thereby achieving objective evaluation of different calibration processes through the calibration evaluation parameters.

[0147] Optionally, the current scene point cloud data determined by the target vehicle in the current scene may be, but is not limited to, the current scene point cloud data determined by a vehicle operating on an actual road while driving, or the current scene point cloud data determined by a vehicle operating on a target client while driving. The target client may be a client providing virtual vehicle functionality, including but not limited to a driving simulation client, or a game client providing virtual vehicle driving functionality.

[0148] As an optional implementation, the determining module 1206 includes:

[0149] an analysis unit, configured to sequentially perform principal component analysis on each of the plurality of grid point cloud data to extract a point cloud feature set that matches each of the grid point cloud data, wherein the point cloud feature set is used to determine geometric features corresponding to the grid point cloud data;

[0150] The target unit determines the target grid point cloud data from the multiple grid point cloud data according to the point cloud feature set matched respectively to each grid point cloud data.

[0151] As an optional implementation, the analysis unit includes:

[0152] Use each of the multiple grid point cloud data as the current grid point cloud data in turn and perform the following operations:

[0153] A construction unit, used to construct a point cloud feature matrix based on the coordinates of each point in the current grid point cloud data;

[0154] A calculation unit, used to calculate the point cloud covariance matrix corresponding to the point cloud feature matrix;

[0155] The decomposition unit is used to perform eigenvalue decomposition on the point cloud covariance matrix to obtain a current point cloud feature set that matches the current grid point cloud data.

[0156] As an optional implementation, the target unit includes:

[0157] The target grid determination unit is used to, when the difference between the target point cloud feature value included in the current point cloud feature set and zero is less than the target threshold, and the difference between the point cloud feature values other than the target point cloud feature value in the current point cloud feature set and zero is greater than the target threshold, the geometric features determined based on the current point cloud feature set will indicate that the current grid area where the current grid point cloud data is located includes a plane or a curved surface, and determine that the current grid point cloud data is the target grid point cloud data.

[0158] As an optional implementation, the generating module 1208 includes:

[0159] A reference unit is used to determine a reference thickness of a reference surface included in a grid area corresponding to each target grid point cloud data according to geometric features of the target grid point cloud data, wherein the reference surface includes a plane or a curved surface;

[0160] The statistical unit is used to perform statistics on the reference thickness corresponding to all target grid point cloud data to generate calibration evaluation parameters.

[0161] As an optional implementation, the reference unit includes:

[0162] Perform the following operations on each target raster point cloud data:

[0163] A fitting unit, used to determine a fitting surface in a grid area corresponding to the target grid point cloud data;

[0164] The distance unit is used to obtain the distance between each point in the target grid point cloud data and the fitting surface; the thickness unit is used to determine the thickness of the fitting surface based on the average value of the distance.

[0165] Determine a unit for using the thickness of the fitted surface as the thickness of the reference surface in the grid area corresponding to the target grid point cloud data.

[0166] As an optional implementation, the segmentation module 1204 includes:

[0167] An extraction unit, configured to extract scene objects contained in the current scene based on the current scene point cloud data;

[0168] An object unit, configured to determine a target scene object from scene objects included in the current scene, wherein the target scene object is the scene object with the smallest size;

[0169] A scale unit, used to determine the segmentation scale corresponding to the current scene point cloud data based on the size of the target scene object;

[0170] The segmentation unit is used to perform segmentation according to the segmentation scale to obtain multiple grid point cloud data.

[0171] As an optional implementation, the acquisition module 1202 may also be used to:

[0172] Obtaining a point cloud positioning result and a combined navigation positioning result collected by the target vehicle in the current scene, wherein the point cloud positioning result is a positioning result obtained based on point cloud data collected by a first sensor device in the target vehicle, and the combined navigation positioning result is a positioning result obtained by combining multiple motion data of the target vehicle collected by a second sensor device in the target vehicle;

[0173] The point cloud positioning results and the combined navigation positioning results are spliced to obtain the current scene point cloud data corresponding to the target vehicle.

[0174] According to another aspect of the embodiment of the present invention, there is also provided an electronic device for implementing the above calibration evaluation parameter acquisition method, which may be Figure 1 The terminal device or server shown in FIG. This embodiment is described by taking the electronic device as a server as an example. Figure 13 As shown, the electronic device includes a memory 1302 and a processor 1304. The memory 1302 stores a computer program, and the processor 1304 is configured to execute the steps in any of the above method embodiments through the computer program.

[0175] Optionally, in this embodiment, the electronic device may be located in at least one network device among a plurality of network devices of a computer network.

[0176] Optionally, in this embodiment, the processor may be configured to execute the following steps through a computer program:

[0177] S1, obtaining current scene point cloud data determined for a target vehicle in a current scene, wherein the current scene point cloud data includes positioning results of the target vehicle calibrated in at least two calibration coordinate systems;

[0178] S2, segmenting the current scene point cloud data to obtain multiple grid point cloud data;

[0179] S3, determining target grid point cloud data from the plurality of grid point cloud data based on geometric features corresponding to each grid point cloud data in the plurality of grid point cloud data, wherein the geometric features of the target grid point cloud data indicate that the grid area includes the reference surface;

[0180] S4, performing statistics on the geometric features of the target grid point cloud data to generate calibration evaluation parameters for matching the current scene point cloud data.

[0181] Alternatively, those skilled in the art will appreciate that Figure 13 The structure shown is for illustration only, and the electronic device may also be a smart phone (such as an Android phone, an iOS phone, etc.), a tablet computer, a PDA, a mobile Internet device (MID), a PAD, or other terminal devices. Figure 13 It does not limit the structure of the electronic device. For example, the electronic device may also include Figure 13 More or fewer components (such as network interfaces, etc.) as shown in, or with Figure 13 Different configurations shown.

[0182] Among them, the memory 1302 can be used to store software programs and modules, such as the program instructions / modules corresponding to the calibration evaluation parameter acquisition method and device in the embodiment of the present invention. The processor 1304 executes various functional applications and data processing by running the software programs and modules stored in the memory 1302, that is, realizes the above-mentioned calibration evaluation parameter acquisition method. The memory 1302 may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory 1302 may further include a memory remotely located relative to the processor 1304, and these remote memories may be connected to the terminal via a network. Examples of the above-mentioned networks include but are not limited to the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof. Among them, the memory 1302 can be used specifically but not limited to store information such as scene point cloud data and evaluation calibration parameters. As an example, such as Figure 13As shown, the memory 1302 may include, but is not limited to, the acquisition module 1202, segmentation module 1204, determination module 1206, and generation module 1208 in the calibration evaluation parameter acquisition device. Furthermore, the memory 1302 may also include, but is not limited to, other module units in the calibration evaluation parameter acquisition device, which will not be described in detail in this example.

[0183] Optionally, the transmission device 1306 is configured to receive or send data via a network. Specific examples of the network may include a wired network and a wireless network. In one embodiment, the transmission device 1306 includes a network interface controller (NIC), which can be connected to other network devices and a router via a network cable to communicate with the Internet or a local area network. In one embodiment, the transmission device 1306 is a radio frequency (RF) module, which is configured to communicate with the Internet wirelessly.

[0184] In addition, the electronic device further includes: a display 1308 for displaying the current scene point cloud data; and a connection bus 1310 for connecting various module components in the electronic device.

[0185] In other embodiments, the terminal device or server may be a node in a distributed system, wherein the distributed system may be a blockchain system, and the blockchain system may be a distributed system formed by connecting multiple nodes through network communication. The nodes may form a peer-to-peer (P2P) network, and any computing device, such as a server, terminal, or other electronic device, may become a node in the blockchain system by joining the peer-to-peer network.

[0186] According to one aspect of the present application, a computer program product or computer program is provided, comprising computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform the calibration evaluation parameter acquisition method provided in the various optional implementations of the calibration evaluation parameter acquisition method described above. The computer program is configured to execute the steps of any of the above-described method embodiments when executed.

[0187] Optionally, in this embodiment, the computer-readable storage medium may be configured to store a computer program for performing the following steps:

[0188] S1, obtaining current scene point cloud data determined for a target vehicle in a current scene, wherein the current scene point cloud data includes positioning results of the target vehicle calibrated in at least two calibration coordinate systems;

[0189] S2, segmenting the current scene point cloud data to obtain multiple grid point cloud data;

[0190] S3, determining target grid point cloud data from the plurality of grid point cloud data based on geometric features corresponding to each grid point cloud data in the plurality of grid point cloud data, wherein the geometric features of the target grid point cloud data indicate that the grid area includes the reference surface;

[0191] S4, performing statistics on the geometric features of the target grid point cloud data to generate calibration evaluation parameters for matching the current scene point cloud data.

[0192] Optionally, in this embodiment, a person of ordinary skill in the art may understand that all or part of the steps in the various methods of the above embodiments may be completed by instructing the hardware related to the terminal device through a program, and the program may be stored in a computer-readable storage medium, which may include: a flash drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, etc.

[0193] The serial numbers of the above embodiments of the present invention are for description only and do not represent the advantages or disadvantages of the embodiments.

[0194] If the integrated units in the above embodiments are implemented in the form of software functional units and sold or used as independent products, they can be stored in the above-mentioned computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the existing technology, or all or part of the technical solution can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes a number of instructions for causing one or more computer devices (such as personal computers, servers, or network devices) to execute all or part of the steps of the methods described in various embodiments of the present invention.

[0195] In the above embodiments of the present invention, the description of each embodiment has its own focus. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0196] In the several embodiments provided in this application, it should be understood that the disclosed client can be implemented in other ways. Among them, the device embodiments described above are merely illustrative. For example, the division of the units is merely a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of units or modules, and can be electrical or other forms.

[0197] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0198] In addition, the functional units in the various embodiments of the present invention may be integrated into a single processing unit, each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or software functional units.

[0199] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as within the scope of protection of the present invention.

Claims

1. A method for obtaining calibration evaluation parameters, characterized in that: include: Acquire current scene point cloud data determined for the target vehicle in the current scene, wherein the current scene point cloud data includes a positioning result of the target vehicle calibrated according to calibration coordinate systems corresponding to at least two devices; Segmenting the current scene point cloud data to obtain a plurality of grid point cloud data; Determining target grid point cloud data from the plurality of grid point cloud data according to geometric features corresponding to each of the plurality of grid point cloud data, wherein the geometric features of the target grid point cloud data indicate that a grid area includes a reference surface; The geometric features of the target grid point cloud data are statistically analyzed to generate calibration evaluation parameters for matching the current scene point cloud data.

2. The method according to claim 1, characterized in that Determining target grid point cloud data from the plurality of grid point cloud data according to the geometric features corresponding to each of the grid point cloud data includes: Performing principal component analysis on each of the plurality of grid point cloud data in sequence to extract a point cloud feature set that matches each of the grid point cloud data, wherein the point cloud feature set is used to determine geometric features corresponding to the grid point cloud data; Target grid point cloud data are determined from the plurality of grid point cloud data according to the point cloud feature set matched to each grid point cloud data.

3. The method according to claim 2, characterized in that The step of sequentially performing principal component analysis on each of the plurality of grid point cloud data to extract a point cloud feature set that matches each of the grid point cloud data comprises: Each of the plurality of grid point cloud data is sequentially used as the current grid point cloud data, and the following operations are performed: Constructing a point cloud feature matrix based on the coordinates of each point in the current grid point cloud data; Calculating the point cloud covariance matrix corresponding to the point cloud feature matrix; Performing eigenvalue decomposition on the point cloud covariance matrix to obtain a current point cloud feature set that matches the current grid point cloud data.

4. The method according to claim 3, characterized in that Determining target grid point cloud data from the plurality of grid point cloud data according to the point cloud feature set matched to each grid point cloud data includes: When the difference between the target point cloud feature value included in the current point cloud feature set and zero is less than the target threshold, and the difference between the point cloud feature values other than the target point cloud feature value in the current point cloud feature set and zero is greater than the target threshold, the geometric features determined based on the current point cloud feature set will indicate that the reference surface is included in the current grid area where the current grid point cloud data is located, and the current grid point cloud data is determined to be the target grid point cloud data.

5. The method according to claim 1, wherein The statistical analysis of the geometric features of the target grid point cloud data to generate calibration evaluation parameters for matching the current scene point cloud data includes: Determining, based on geometric features of the target grid point cloud data, a reference thickness of the reference surface included in the grid area corresponding to each of the target grid point cloud data, wherein the reference surface includes a plane or a curved surface; The reference thicknesses corresponding to all the target grid point cloud data are counted to generate the calibration evaluation parameters.

6. The method according to claim 5, characterized in that The determining, based on the geometric features of the target grid point cloud data, the thickness of the reference surface in the grid area corresponding to each of the target grid point cloud data comprises: The following operations are performed on each target grid point cloud data: Determining a fitting surface in a grid area corresponding to the target grid point cloud data; Obtaining the distance between each point in the target grid point cloud data and the fitting surface; Determine the thickness of the fitting surface based on the average value of the distance The thickness of the fitting surface is used as the thickness of the reference surface in the grid area corresponding to the target grid point cloud data.

7. The method according to claim 6, characterized in that Before determining the fitting surface in the grid area corresponding to the target grid point cloud data, the method further includes: Determining a candidate fitting surface based on the target grid point cloud data; The discrete points whose distance from the candidate fitting surface is greater than a target distance threshold are removed to update the target grid point cloud data.

8. The method according to claim 1, characterized in that The segmenting of the current scene point cloud data to obtain a plurality of grid point cloud data includes: Extracting scene objects contained in the current scene according to the current scene point cloud data; Determining a target scene object from the scene objects included in the current scene, wherein the target scene object is the scene object with the smallest size; Determining a segmentation scale corresponding to the current scene point cloud data based on the size of the target scene object; Segmentation is performed according to the segmentation scale to obtain the plurality of grid point cloud data.

9. The method according to any one of claims 1 to 8, characterized in that The step of obtaining the current scene point cloud data determined by the target vehicle in the current scene includes: Obtaining a point cloud positioning result and a combined navigation positioning result collected by the target vehicle in the current scene, wherein the point cloud positioning result is a positioning result obtained based on point cloud data collected by a first sensing device in the target vehicle, and the combined navigation positioning result is a positioning result obtained by combining multiple motion data of the target vehicle collected by a second sensing device in the target vehicle; The point cloud positioning result and the combined navigation positioning result are spliced to obtain the current scene point cloud data corresponding to the target vehicle.

10. A calibration evaluation parameter acquisition device, characterized in that: include: an acquisition module, configured to acquire current scene point cloud data determined by a target vehicle in a current scene, wherein the current scene point cloud data includes a positioning result of the target vehicle calibrated according to calibration coordinate systems corresponding to at least two devices; A segmentation module, configured to segment the current scene point cloud data to obtain a plurality of grid point cloud data; a determination module, configured to determine target grid point cloud data from the plurality of grid point cloud data based on geometric features corresponding to each of the plurality of grid point cloud data, wherein the geometric features of the target grid point cloud data indicate that a grid area includes a reference surface; The generation module is used to perform statistics on the geometric features of the target grid point cloud data to generate calibration evaluation parameters for matching the current scene point cloud data.

11. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored program, wherein the method according to any one of claims 1 to 9 is executed when the program is executed.

12. An electronic device comprising a memory and a processor, characterized in that: A computer program is stored in the memory, and the processor is configured to execute the method according to any one of claims 1 to 9 through the computer program.

Images