A method and system for generating credibility for high-precision map production

Abstract

This invention discloses a credibility generation method for high-precision map production. The credibility generation method includes: constructing a production process topology chain for high-precision maps; the production process topology chain sequentially includes a field surveying process, a field quality inspection process, a cartographic production process, and an office quality inspection process; calculating initial credibility values ​​for each of the field surveying process, field quality inspection process, cartographic production process, and office quality inspection process; establishing credibility transmission equations for the field quality inspection process, cartographic production process, and office quality inspection process based on the weighted result of the credibility of the preceding process and the historical credibility of the current process; the credibility of the preceding process refers to the credibility of the previous process, and the historical credibility of the current process refers to the credibility of the previous iteration process; performing credibility iterative optimization on the credibility transmission equations of each process until the iteration termination condition is reached. Based on this, this invention achieves accurate tracing of production quality problems and greatly improves quality inspection efficiency.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of electronic map, in particular to a credibility generation method and system for high-precision map production. BACKGROUND

[0002] In recent years, with the development of automatic driving technology to L4 level, high-precision map as the "digital base" of vehicle environment perception, its centimeter-level spatial accuracy and millisecond-level update requirement has become a key technical bottleneck restricting the development of the industry. The strict industrial-level precision standard (horizontal positioning accuracy ≤10 cm, vertical accuracy ≤20 cm) and the composite quality requirement (element attribute accuracy rate ≥99%, topological logic correctness rate 100%, dynamic update delay <100 ms) are reshaping the industrial technology paradigm. At present, a single city-level high-precision map in the industry needs to pass more than 100 geometric precision tests and at least 80 semantic consistency verifications, and the manual review efficiency of the traditional quality inspection method is less than 1.2 kilometers per person per day when facing daily PB-level data throughput. SUMMARY

[0003] In order to overcome the three technical bottlenecks of fragmented quality evaluation, missing quantitative standards and out-of-control error propagation in high-precision map production, the present application provides a credibility generation method and system for high-precision map production by constructing a full-process credibility evaluation system and a credibility transmission model, which greatly improves the quality inspection accuracy and efficiency of high-precision map.

[0004] According to one aspect of the present application, a credibility generation method for high-precision map production is provided, and a production link topology of high-precision map is constructed. An outside surveying and mapping link, an outside quality inspection link, a mapping production link and an inside quality inspection link are sequentially arranged on the production link topology. The outside surveying and mapping link is used for collecting road environment spatial data. The outside quality inspection link is used for real-time checking the road environment spatial data. The mapping production link is used for converting the checked road environment spatial data into a high-precision map. The inside quality inspection link is used for quality verification of the high-precision map. The credibility initial values of the outside surveying and mapping link, the outside quality inspection link, the mapping production link and the inside quality inspection link are calculated respectively. The credibility transmission equations of the outside quality inspection link, the mapping production link and the inside quality inspection link are established based on the weighted results of the credibility of the previous link and the historical credibility of the current link. The credibility of the previous link refers to the credibility of the previous link of the current link, and the historical credibility of the current link refers to the credibility of the previous iteration process of the current link. The credibility iteration optimization of the credibility transmission equation of each link is performed until the iteration termination condition is reached.

[0005] Furthermore, initial credibility values ​​were calculated for the field surveying, field quality inspection, map production, and office quality inspection stages, including: initial credibility values ​​for the field surveying stage based on camera calibration error, LiDAR point cloud density, and positioning drift; initial credibility values ​​for the field quality inspection stage based on data integrity rate, anomaly detection rate, and spatiotemporal consistency indicators; initial credibility values ​​for the map production stage based on SLAM loop closure error, semantic annotation accuracy, and manual review pass rate; and initial credibility values ​​for the office quality inspection stage based on topological connectivity verification results, lane curvature continuity, and traffic element compliance rate.

[0006] Furthermore, based on camera calibration error, lidar point cloud density, and positioning drift, the initial reliability value for the field surveying process is calculated using the following formula: ,in, For camera calibration error; This refers to the point cloud density of the lidar. This is the location drift amount; This serves as the initial reliability value for the field surveying process. For camera calibration error LiDAR point cloud density and positioning drift amount The initial value equation for the reliability of the field surveying process is constructed.

[0007] Furthermore, based on data integrity rate, outlier detection rate, and spatiotemporal consistency indicators, an initial reliability value for the field quality inspection process is calculated using the following formula: ,in, For data integrity rate; The outlier detection rate; As a spatiotemporal consistency indicator; , , For adjustment coefficients, This serves as the initial reliability value for the field quality inspection process. Based on data integrity rate Anomaly detection rate Spatiotemporal consistency index The initial reliability equation for the field quality inspection process is constructed.

[0008] Furthermore, the mapping production process is calculated based on SLAM loop closure error, semantic annotation accuracy, and manual review pass rate. The initial confidence level is calculated using the following formula: ,in, This refers to the SLAM loop closure error. For semantic annotation accuracy, To improve the pass rate of manual review, a confidence initial value of a mapping production link, a loop closure error based on SLAM , a semantic labeling accuracy and a manual review pass rate to construct a confidence initial value equation of the mapping production link.

[0009] Further, a confidence initial value of an office inspection link is calculated based on a topological connectivity verification result, lane line curvature continuity and traffic element compliance rate, and the formula is: , wherein is the topological connectivity verification result; is the lane line curvature continuity; is the traffic element compliance rate; is a piecewise function: 1 is taken when ; ; ; ; ; is the confidence initial value of the office inspection link, is a confidence initial value equation of the office inspection link constructed based on a topological connectivity verification result , lane line curvature continuity and traffic element compliance rate .

[0010] Further, a confidence transmission equation of the field inspection link, the mapping production link and the office inspection link is respectively established based on the weighted results of the confidence of the previous link and the historical confidence of the current link, including: a confidence transmission equation of the field inspection link is established by using the weighted results of the confidence of the field survey link and the historical confidence of the field inspection link; a confidence transmission equation of the mapping production link is established by using the weighted results of the confidence of the field inspection link and the historical confidence of the mapping production link; and a confidence transmission equation of the office inspection link is established by using the weighted results of the confidence of the mapping production link and the historical confidence of the office inspection link.

[0011] Further, the confidence transmission equation of each link is executed for confidence iteration optimization until an iteration termination condition is reached, including: a confidence loss function is established based on the predicted confidence, the target confidence and the weight matrix calculated by the confidence transmission equation; a gradient descent algorithm is used to optimize the weight matrix, and the confidence transmission equation is updated by using the optimized weight matrix; an iteration loss change rate of the confidence loss function is calculated; and when the iteration loss change rate is less than a preset threshold and / or the number of iterations reaches a maximum number of iterations, the confidence iteration optimization is stopped.

[0012] According to an aspect of the present application, the present application provides a credibility generation system for high-precision map production, comprising: a production link topology chain construction module for constructing a production link topology chain of a high-precision map; the production link topology chain is sequentially provided with an outdoor surveying and mapping link, an outdoor quality inspection link, a cartographic production link and an indoor quality inspection link; the outdoor surveying and mapping link is used for collecting road environment spatial data; the outdoor quality inspection link is used for real-time checking the road environment spatial data; the cartographic production link is used for converting the checked road environment spatial data into a high-precision map; the indoor quality inspection link is used for quality verification of the high-precision map; a credibility initial value calculation module is used for calculating credibility initial values of the outdoor surveying and mapping link, the outdoor quality inspection link, the cartographic production link and the indoor quality inspection link respectively; a cross-link conduction module is used for establishing credibility conduction equations of the outdoor quality inspection link, the cartographic production link and the indoor quality inspection link based on a weighted result of a previous link credibility and a current link historical credibility; the previous link credibility refers to the credibility of the previous link of the current link, and the current link historical credibility refers to the credibility of a previous iteration process of the current link; an iteration optimization module is used for performing credibility iteration optimization on the credibility conduction equations of each link until an iteration termination condition is reached.

[0013] According to an aspect of the present application, the present application provides a non-transitory computer readable storage medium, the non-transitory computer readable storage medium stores computer instructions, the computer instructions make the computer execute the credibility generation method for high-precision map production.

[0014] The above technical solution constructs a whole-process credibility evaluation system covering the outdoor surveying and mapping link, the outdoor quality inspection link, the cartographic production link and the indoor quality inspection link, and realizes accurate error tracing of production quality problems through the construction of a credibility conduction model, which greatly improves the quality inspection efficiency and supports dynamic allocation of production resources compared with the traditional independent quality inspection mode.

[0015] Compared with the prior art, the present application has the following advantages:

[0016] (1) Whole-process connection: The credibility conduction chain solves the problem of link fragmentation, and the actual measurement shows that the error tracing efficiency of outdoor positioning is improved.

[0017] (2) Dynamic quantitative evaluation: The multi-modal data fusion model improves the device abnormality detection rate and reduces the cartographic misjudgment rate to 1.8%.

[0018] (3) Intelligent error suppression: The gradient optimization mechanism realizes error propagation attenuation rate ≥65%, and the map error after conduction of outdoor error is controlled within 0.22m.

[0019] (4) Strong engineering applicability: supports edge-cloud distributed deployment, greatly improves map production efficiency, and reduces labor costs. BRIEF DESCRIPTION OF DRAWINGS

[0020] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings used in the embodiments or prior art description will be briefly introduced. Obviously, the drawings in the following description are some embodiments of the present application, and for those skilled in the art, other drawings can be obtained without creative labor based on these drawings.

[0021] Figure 1 A flowchart of a credibility generation method for high-precision map production provided by the embodiment of the present application.

[0022] Figure 2 A first sub-flowchart of a credibility generation method for high-precision map production provided by the embodiment of the present application.

[0023] Figure 3 A second sub-flowchart of a credibility generation method for high-precision map production provided by the embodiment of the present application.

[0024] Figure 4 A third sub-flowchart of a credibility generation method for high-precision map production provided by the embodiment of the present application.

[0025] Figure 5 A schematic diagram of a credibility generation system for high-precision map production provided by the embodiment of the present application.

[0026] In the figure, 1 is a production link topology chain construction module; 2 is a credibility initial value calculation module; 3 is a cross-link conduction module; 4 is an iterative optimization module; 100 is a credibility generation system. DETAILED DESCRIPTION

[0027] The terms "include" and "have" and any variations thereof in the specification and claims of the present application and the above drawings are intended to cover non-exclusive inclusion, for example, a process, method, system, product or device including a series of steps or units does not have to be limited to clearly listed steps or units, but can include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0028] The block diagrams shown in the drawings are merely functional entities, and do not necessarily correspond to physically independent entities. That is, the functional entities can be implemented in the form of software, or in one or more hardware modules or integrated circuits, or in different networks and / or processor devices and / or microcontroller devices. The flowcharts shown in the drawings are merely exemplary illustrations, and do not necessarily include all contents and operations / steps, nor are they necessarily executed in the order described. For example, some operations / steps can be further decomposed, and some operations / steps can be combined or partially combined, so the actual execution order can be changed according to actual conditions.

[0029] In order to make the purposes, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative labor fall within the scope of protection of the present application. In addition, the technical features in each embodiment or in a single embodiment provided by the present application can be combined with each other at will to form new technical solutions, and such combination is not restricted by the order of steps and / or structure composition mode, but must be based on the fact that those of ordinary skill in the art can implement it. When the combination of technical solutions contradicts each other or cannot be implemented, it should be considered that such combination of technical solutions does not exist, nor is it within the scope of protection required by the present application.

[0030] Please refer to the accompanying drawings Figure 1 The present application provides a credibility generation method for high-precision map production, proposes a quantitative credibility evaluation system for a high-precision map production pipeline, and realizes accurate quality defect traceability and evaluation automation by establishing a full-process credibility model of field surveying and mapping links, field inspection links, mapping production links, and in-house inspection links. The credibility generation method includes steps S101-S107.

[0031] Step S101, constructing a production link topology of a high-precision map.

[0032] In step S101, four links of field surveying link (A), field inspection link (B), mapping production link (C), and office inspection link (D) and their data flow relationship are defined. Specifically, the field surveying link, the field inspection link, the mapping production link, and the office inspection link are sequentially arranged on the production link topology chain (i.e., the data flow relationship of the four links is: field surveying link (A) → field inspection link (B) → mapping production link (C) → office inspection link (D)). The field surveying link is used to collect road environment spatial data through a vehicle-mounted mobile measurement system (sensors such as satellite positioning, inertial measurement unit, laser radar, and panoramic camera). The field inspection link is used to perform real-time verification on the collected road environment spatial data, including but not limited to device state (such as inertial measurement unit offset angle), data completeness (such as time synchronization error), and environmental adaptability (light intensity). The mapping production link is used to convert the verified road environment spatial data (i.e., original data) into a high-precision map. The office inspection link is used to perform quality verification on the high-precision map product.

[0033] In step S103, the initial credibility values of the field surveying link, the field inspection link, the mapping production link, and the office inspection link are calculated respectively.

[0034] In step S103, the present application maps the multi-source heterogeneous data into standardized initial credibility. It can be understood that multi-source heterogeneous data fusion improves the device anomaly detection rate, thereby reducing the mapping misjudgment rate. At the same time, the standardized initial credibility process realizes the quantification of each index, such as quantifying the spatial accuracy of camera calibration error, quantifying the coverage integrity of laser radar point cloud density, and quantifying the algorithm stability of SLAM loop error, thereby solving the problem of missing quantization standards in high-precision map production.

[0035] For reference, please see Figure 2 The initial credibility calculation process of each link (including steps S1031-S1035) will be further introduced below.

[0036] In step S1031, the initial credibility value of the field surveying link is calculated based on the camera calibration error, the laser radar point cloud density, and the positioning drift amount. Specifically, the calculation formula of the initial credibility value of the field surveying link is: wherein, is the camera calibration error, which refers to the error generated by the projection when converting from the image coordinate system of the camera to the world coordinate system; is the laser radar point cloud density, which refers to the number of effective point clouds collected by the laser radar per unit ground projection area; is the positioning drift amount, which refers to the plane deviation between the actual position and the satellite positioning; is the initial credibility value of the field surveying link, to calibrate the camera , the laser radar point cloud density , and the positioning drift amount The reliability initial value equation of the field survey link is constructed. In this embodiment, in pixels, in points / m², in meters.

[0037] Step S1033, the reliability initial value of the field quality inspection link is calculated based on the data integrity rate, the abnormal point detection rate, and the space-time consistency index. Specifically, the calculation formula of the reliability initial value of the field quality inspection link is: wherein, is the data integrity rate, the data integrity rate refers to the ratio of the effective collection mileage to the total planned mileage; is the abnormal point detection rate, the abnormal point detection rate refers to the proportion of the identified defective data relative to the high-precision map; is the space-time consistency index, the space-time consistency index refers to the error generated in the time synchronization process of multiple sensors; , , is the adjustment coefficient, is the reliability initial value of the field quality inspection link, is the reliability initial value of the field quality inspection link constructed based on the data integrity rate , the abnormal point detection rate , and the space-time consistency index . In this embodiment, , , .

[0038] Step S1035, the reliability initial value of the mapping production link is calculated based on the SLAM loop closure error, the semantic labeling accuracy, and the artificial review pass rate. Specifically, the calculation formula of the reliability initial value of the mapping production link is: wherein, is the SLAM loop closure error, the SLAM loop closure error refers to the inconsistency degree of the position and attitude when the mobile device moves along the trajectory and finally returns to the original point trajectory closure during mapping; is the semantic labeling accuracy, the semantic labeling accuracy refers to the correctness rate of the machine automatic labeling of high-precision map elements; is the artificial review pass rate, the artificial review pass rate refers to the qualified rate of the human operation acceptance of high-precision map elements; is the reliability initial value of the mapping production link, is the reliability initial value of the mapping production link constructed based on the SLAM loop closure error , the semantic labeling accuracy and manual review pass rate The reliability initial value equation of the mapping production link is constructed. In step S1037, the reliability initial value of the office inspection link is calculated based on the topological connectivity verification result, the lane line curvature continuity and the traffic element compliance rate. Specifically, the calculation formula of the reliability initial value of the office inspection link is as follows: , wherein, is the topological connectivity verification result, the topological connectivity verification refers to the verification of the reachability (i.e. effective path) between the nodes of the map road network, 1 indicates pass, and 0 indicates fail; is the lane line curvature continuity, the lane line curvature continuity refers to the smoothness of the curvature change of the lane center line; is the traffic element compliance rate, the traffic element compliance rate refers to the ratio of the map elements meeting the actual traffic rules; is a piecewise function: 1 is taken when is taken when ; is taken when is taken when ; is the reliability initial value of the office inspection link, is the reliability initial value equation of the office inspection link constructed based on the topological connectivity verification result , the lane line curvature continuity and the traffic element compliance rate .

[0039] In step S105, the reliability transmission equations of the field inspection link, the mapping production link and the office inspection link are respectively established based on the weighted results of the preceding link reliability and the historical reliability of the current link.

[0040] In step S105, the preceding link reliability refers to the reliability of the previous link of the current link, the historical reliability of the current link refers to the reliability of the previous iteration process of the current link, and the reliability transmission equation is obtained by substituting the weighted results of the preceding link reliability and the historical reliability of the current link into the clip function. is the clip function: It can be understood that the reliability transmission equation of the present application sets the weight matrix Λ in series with the link coupling, and uses the clip truncation function to prevent the reliability from overflowing, thereby solving the problem of fragmentation of quality evaluation in high-precision map production.

[0041] Please refer to the accompanying drawings Figure 3 The step S105 (including steps S1051-S1055) will be further introduced below.

[0042] Step S1051, the reliability of the field inspection link is established by using the weighted result of the reliability of the field survey link and the historical reliability of the field inspection link. Specifically, the reliability transmission equation of the field inspection link is, . represents the reliability weight transmitted from the field survey link (A) to the field inspection link (B). represents the weight of the self-reliability of the field inspection link (B) in the initialization process or the last iteration process. t represents the number (time) of iterations, represents the reliability of the field inspection link (B) in the tthiteration, represents the reliability of the field survey link (A) in the tthiteration (current iteration process) (i.e., the reliability of the previous link of the field inspection link). represents the reliability of the field inspection link (B) in the t-1thiteration (last iteration process) (i.e., the historical reliability of the field inspection link). It can be understood that the reliability of the field inspection link (B) in the second iteration process is: . It should be noted that the reliability of the field survey link (A) does not need to be iterated, and is obtained according to the error estimation of the equipment / equipment of the field inspection department at each field collection. It can be understood that the reliability of the field survey link (A) is equal to the initial value of the reliability of the field survey link .

[0043] Step S1053, the reliability of the mapping production link is established by using the weighted result of the reliability of the field inspection link and the historical reliability of the mapping production link. Specifically, the reliability transmission equation of the mapping production link is: . Wherein, represents the weight transmitted from the field inspection link (B) to the mapping production link (C); represents the weight of the self-reliability of the mapping production link (C) in the initialization process or the last iteration process. represents the reliability of the mapping production link (C) in the tthiteration, represents the reliability of the field inspection link (B) in the tthiteration (current iteration process) (i.e., the reliability of the previous link of the mapping production link); represents the reliability of the mapping production link (C) in the t-1thiteration (last iteration process) (i.e., the historical reliability of the mapping production link). It can be understood that the reliability of the mapping production link (C) in the second iteration process is: .

[0044] Step S1055, the confidence of the internal office inspection link is established by using the weighted results of the confidence of the cartographic production link and the historical confidence of the internal office inspection link. The confidence of the internal office inspection link is . Wherein, represents the weight of the confidence of the cartographic production link (C) conducted to the internal office inspection link (D); represents the weight of the confidence of the internal office inspection link (D) in the initialization process or the last iteration process. represents the confidence of the internal office inspection link (D) in the tthiteration, represents the confidence of the cartographic production link (C) in the tthiteration (the current iteration process) (i.e. the confidence of the previous link of the internal office inspection link); represents the confidence of the internal office inspection link (D) in the t-1thiteration (the last iteration process) (i.e. the historical confidence of the internal office inspection link). It can be understood that the confidence of the internal office inspection link (D) in the second iteration process is: .

[0045] Step S107, the confidence iteration optimization of the confidence transmission equation of each link is performed until the iteration termination condition is reached.

[0046] In step S107, the weight matrix is initialized: , the transmission weight satisfies the constraint condition: ; ; . The weight matrix is optimized by the gradient descent algorithm , and the confidence transmission equation is updated by using the optimized weight matrix , and the above process is repeated to realize the confidence iteration optimization of the confidence transmission equation. It can be understood that the weight matrix at the iteration termination condition is brought into the confidence transmission equation, and the confidence value calculated is the confidence value of the high-precision map.

[0047] Please refer to the attached Figure 4 , the step S107 (including steps S1071-S1075) will be further introduced below.

[0048] Step S1071, the confidence loss function is established based on the predicted confidence calculated by the confidence transmission equation, the target confidence and the weight matrix. Specifically, the confidence loss function is: , wherein, represents the confidence loss function, represents the total number of training samples, represents the regularization coefficient, ​The target confidence level represents the number of training samples i (which can be obtained from historical data or expert annotation). The prediction confidence represents the number of training samples i (obtained through the confidence propagation equation of the current initialization or the previous step). Understandably, during the confidence iteration process of the confidence propagation equation at each stage, the prediction confidence... and weight matrix It is constantly updated, and the credibility of new predictions... It is also composed of a new weight matrix. The weight matrix is ​​calculated, and the present invention also utilizes the gradient descent algorithm to optimize the weight matrix. Therefore, this invention achieves error attenuation while employing a regularization coefficient. Suppressing overfitting and preventing uncontrolled error propagation solves the problem of uncontrolled error propagation in high-precision map production.

[0049] Step S1073: The gradient descent algorithm is used to optimize the weight matrix, and the confidence propagation equation is updated using the optimized weight matrix. Specifically, the gradient descent algorithm employs a momentum optimization strategy: .in, The weight matrix represents the weight matrix for the (t+1)th iteration; The weight matrix represents the weight matrix for the t-th iteration; This represents the weight matrix for the (t-1)th iteration; The rate of change of the iterative loss is obtained by taking the partial derivative of the loss function with respect to each weight coefficient λ at time t. Represents the learning rate; This represents the momentum factor. Momentum optimization strategies utilize the weight matrix at different old time points (t, t-1). To calculate the weight matrix at the new time (t+1) The method. In this embodiment, the learning rate Momentum factor , =[0.6,0.4,0.7,0.3,0.8,0.2]. Understandably, the learning rate... Momentum factor Adjustments can be made according to actual needs; no restrictions are imposed here.

[0050] Step S1075: Calculate the rate of change of iterative loss for the credibility loss function. When the rate of change of iterative loss is less than a preset threshold (i.e., ... In the formula, The confidence iteration optimization stops when the preset threshold (representing a threshold value) and / or the maximum number of iterations is reached. In this embodiment, the iteration termination condition is the rate of change of loss after 10 consecutive iterations. It can be understood that the preset threshold and the maximum number of iterations can be set according to actual needs, and are not limited herein.

[0051] Please refer to Figure 5 Based on the same inventive concept as the foregoing embodiments, the present application also provides a credibility generation system 100 for high-precision map production, which comprises a production link topology chain construction module 1, a credibility initial value calculation module 2, a cross-link conduction module 3, and an iterative optimization module 4. The production link topology chain construction module 1 is used to construct a production link topology chain of a high-precision map. The credibility initial value calculation module 2 is used to calculate credibility initial values of an outdoor surveying and mapping link, an outdoor quality inspection link, a cartographic production link, and an indoor quality inspection link, respectively. The cross-link conduction module 3 is used to establish credibility conduction equations of each link based on a weighted result of a previous link credibility and a historical credibility of the current link. The iterative optimization module 4 is used to perform credibility iterative optimization on the credibility conduction equations of each link until an iterative termination condition is reached.

[0052] Based on the same inventive concept as the foregoing embodiments, the present application also provides a non-transitory computer-readable storage medium storing computer instructions, which cause the computer to execute the credibility generation method for high-precision map production. These computer instructions can be loaded onto a computer or other programmable data processing device, so that a series of operation steps are executed on the computer or other programmable device to produce a computer-implemented process, so that the instructions executed on the computer or other programmable device provide steps for implementing the functions specified in the flow Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0053] In summary of the foregoing embodiments, the present application provides a credibility generation method for high-precision map production, which comprises: constructing a whole-process credibility evaluation system covering an outdoor surveying and mapping link, an outdoor quality inspection link, a cartographic production link, and an indoor quality inspection link, and mapping sensor errors, quality inspection indicators, and cartographic parameters of each link into initial credibility through a multi-source data fusion algorithm; establishing a dynamic credibility conduction model, and realizing cross-link credibility iterative optimization based on an adaptive weight matrix, wherein the credibility calculation of each link comprehensively considers the conduction value of a previous link and the initial value of the current link, and the system converges to an optimal solution through a gradient descent algorithm. The present application first proposes a quantitative credibility evaluation system for a high-precision map production pipeline, realizes accurate traceability of production quality problems through the construction of a credibility conduction model, and greatly improves the quality inspection efficiency and supports dynamic allocation of production resources compared with a traditional independent quality inspection method.

[0054] It should be noted that the above examples are only used to illustrate the technical solutions of the present application, and are not intended to limit the present application; although the present application has been described in detail with reference to the foregoing examples, those skilled in the art should understand that the technical solutions recorded in the foregoing examples can still be modified, or some or all of the technical features can be replaced by equivalents; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the technical solutions of the embodiments of the present application.

Claims

1. A reliability generation method for high-precision map production, characterized by, The credibility generation method comprises: A production link topology chain of the high-precision map is constructed; the production link topology chain is sequentially provided with an outdoor surveying and mapping link, an outdoor quality inspection link, a mapping production link, and an indoor quality inspection link; the outdoor surveying and mapping link is used for collecting road environment spatial data; the outdoor quality inspection link is used for performing real-time checking on the road environment spatial data; the mapping production link is used for converting the checked road environment spatial data into a high-precision map; and the indoor quality inspection link is used for performing quality verification on the high-precision map. The initial values of the reliabilities of the field surveying and mapping link, the field quality inspection link, the cartographic production link and the internal quality inspection link are calculated respectively; the reliability transmission equations of the field quality inspection link, the cartographic production link and the internal quality inspection link are respectively established based on the weighted results of the preceding link reliability and the historical reliability of the current link; the preceding link reliability refers to the reliability of the previous link of the current link, and the historical reliability of the current link refers to the reliability of the previous iteration process of the current link; wherein the reliability transmission equations of the field quality inspection link, the cartographic production link and the internal quality inspection link are respectively established based on the weighted results of the preceding link reliability and the historical reliability of the current link, including: the weighted results of the reliability of the field surveying and mapping link and the historical reliability of the field quality inspection link are substituted into a clip function to obtain the reliability transmission equation of the field quality inspection link; the weighted results of the reliability of the field quality inspection link and the historical reliability of the cartographic production link are substituted into the clip function to obtain the reliability transmission equation of the cartographic production link; the weighted results of the reliability of the cartographic production link and the historical reliability of the internal quality inspection link are substituted into the clip function to obtain the reliability transmission equation of the internal quality inspection link; the clip function is a clipping function. ​ A credibility iterative optimization is performed on a credibility transmission equation of each link until an iterative termination condition is reached.

2. The reliability generation method for high-precision map production according to claim 1, characterized by, Credibility initial values of the outdoor surveying and mapping link, the outdoor quality inspection link, the mapping production link, and the indoor quality inspection link are respectively calculated, comprising: The credibility initial value of the outdoor surveying and mapping link is calculated based on camera calibration errors, laser radar point cloud density, and positioning drift amount; The credibility initial value of the outdoor quality inspection link is calculated based on data integrity rate, abnormal point detection rate, and space-time consistency index; The credibility initial value of the mapping production link is calculated based on SLAM loop closure error, semantic labeling accuracy, and artificial review pass rate; and the credibility initial value of the indoor quality inspection link is calculated based on topological connectivity verification result, lane line curvature continuity, and traffic element compliance rate.

3. The method of claim 2, wherein the method is used for high-precision map production. The reliability initial value of the field survey link is calculated based on camera calibration error, laser radar point cloud density and positioning drift amount, and the formula is: wherein, is the camera calibration error; is the laser radar point cloud density; is the positioning drift amount; is the reliability initial value of the field survey link, is the reliability initial value of the field survey link constructed based on the camera calibration error , the laser radar point cloud density and the positioning drift amount .

4. The reliability generation method for high-precision map production according to claim 2, characterized by, The initial reliability value of the field quality inspection process is calculated based on data integrity rate, outlier detection rate, and spatiotemporal consistency indicators. The formula is as follows: ,in, For data integrity rate; The outlier detection rate; As a spatiotemporal consistency indicator; , , For adjustment coefficients, This serves as the initial reliability value for the field quality inspection process. Based on data integrity rate Anomaly detection rate Spatiotemporal consistency index The initial reliability equation for the field quality inspection process is constructed.

5. The reliability generation method for high-precision map production according to claim 2, characterized by, The SLAM loop closure error, the semantic labeling accuracy, and the manual review pass rate are used to calculate the initial value of the reliability of the mapping production link, and the formula is: wherein, is the SLAM loop closure error, is the semantic labeling accuracy, is the manual review pass rate, is the initial value of the reliability of the mapping production link, is the initial value of the reliability of the mapping production link based on the SLAM loop closure error, is the initial value of the reliability of the mapping production link based on the semantic labeling accuracy, is the initial value of the reliability of the mapping production link based on the manual review pass rate, and is the initial value of the reliability of the mapping production link constructed based on the SLAM loop closure error, the semantic labeling accuracy, and the manual review pass rate.

6. The method of claim 2, wherein the method is used for high-precision map production. The reliability initial value of the in-office inspection link is calculated based on the topological connectivity verification result, the lane line curvature continuity and the traffic element compliance rate, and the formula is: wherein, is the topological connectivity verification result; is the lane line curvature continuity; is the traffic element compliance rate; is a piecewise function: 1 when t is 1; ; ; is the reliability initial value of the in-office inspection link, is the reliability initial value of the in-office inspection link constructed based on the topological connectivity verification result , the lane line curvature continuity and the traffic element compliance rate .​​​ ​ 7. The method of claim 1, wherein the method is used for high-precision map production. A credibility iterative optimization is performed on a credibility transmission equation of each link until an iterative termination condition is reached, comprising: A credibility loss function is established based on a predicted credibility obtained by the credibility transmission equation, a target credibility, and a weight matrix; A gradient descent algorithm is used to optimize the weight matrix, and the credibility transmission equation is updated by using the optimized weight matrix; An iterative loss change rate of the credibility loss function is calculated; when the iterative loss change rate is less than a preset threshold and / or the number of iterations reaches a maximum number of iterations, the credibility iterative optimization is stopped.

8. A reliability generation system for high-precision map production, characterized by, Comprise: A production link topology chain construction module is configured to construct a production link topology chain of the high-precision map; the production link topology chain is sequentially provided with an outdoor surveying and mapping link, an outdoor quality inspection link, a mapping production link, and an indoor quality inspection link; the outdoor surveying and mapping link is used for collecting road environment spatial data; the outdoor quality inspection link is used for performing real-time checking on the road environment spatial data; the mapping production link is used for converting the checked road environment spatial data into a high-precision map; and the indoor quality inspection link is used for performing quality verification on the high-precision map; A credibility initial value calculation module is configured to calculate credibility initial values of the outdoor surveying and mapping link, the outdoor quality inspection link, the mapping production link, and the indoor quality inspection link respectively; The cross-link conduction module is configured to establish a reliability conduction equation of the field inspection link, the mapping production link and the office inspection link based on a weighted result of the previous link reliability and the historical reliability of the current link; the previous link reliability refers to the reliability of the previous link of the current link, and the historical reliability of the current link refers to the reliability of the previous iteration process of the current link; wherein the reliability conduction equation of the field inspection link, the mapping production link and the office inspection link based on the weighted result of the previous link reliability and the historical reliability of the current link comprises: substituting the weighted result of the reliability of the field mapping link and the historical reliability of the field inspection link into a clip function to obtain the reliability conduction equation of the field inspection link; substituting the weighted result of the reliability of the field inspection link and the historical reliability of the mapping production link into the clip function to obtain the reliability conduction equation of the mapping production link; and substituting the weighted result of the reliability of the mapping production link and the historical reliability of the office inspection link into the clip function to obtain the reliability conduction equation of the office inspection link. The clip function is a truncation function. An iterative optimization module is configured to perform a credibility iterative optimization on a credibility transmission equation of each link until an iterative termination condition is reached.

9. A non-transitory computer-readable storage medium, comprising: The non-transitory computer readable storage medium stores computer instructions, and the computer instructions cause the computer to perform the credibility generation method for high-precision map production according to any one of claims 1 to 7.

Images