Map construction method and device of target area, computer equipment and storage medium

Abstract

The invention relates to the technical field of automatic driving, and discloses a map construction method and device for a target area, computer equipment and a storage medium, and the method comprises the steps: obtaining vehicle driving information and road pixel data when a vehicle drives in the target area for many times, and carrying out the clustering according to the road pixel data, the method comprises the steps of obtaining vectorized graphic data of different types representing a target area according to the vehicle driving information, then performing data cleaning on the vectorized graphic data according to the types of the vectorized graphic data and the vehicle driving information, filtering out unnecessary data information, and finally performing clustering and fitting based on the cleaned vectorized graphic data to obtain a map of the target area. Therefore, the vehicle acquires the road information and the driving information of the same area for multiple times, the data acquired for multiple times are analyzed and processed, a high-precision map is finally formed for driving navigation, manual field survey and test are not needed, the manufacturing cost is low, more road information can be provided for vehicle driving, and the precision is higher.

Description

Technical Field

[0001] The present invention relates to the technical field of autonomous driving, and particularly relates to a method, an apparatus, a computer device and a storage medium for constructing a map of a target area. Background Art

[0002] A high-precision map, also known as an "autonomous driving map" or a "basic map for intelligent vehicles", is called HDMap (High Definition Map) in English. A high-precision map refers to a high-resolution and high-abundance navigation map with both absolute accuracy and relative accuracy in the centimeter level.

[0003] Currently, when making traditional high-precision maps, a large amount of on-site survey work is often required, which requires a lot of manpower and material resources and has a high production cost; moreover, for some sections and areas with few vehicles or not open to the public, it is very unnecessary to spend manpower to make high-precision maps of these sections and areas, which results in low map accuracy in these specific areas and little driving reference significance.

[0004] Therefore, for some sections and areas with few vehicles, how to construct high-precision maps of these areas and sections on the premise of low production cost has become a key problem to be solved urgently. Summary of the Invention

[0005] In view of this, the present invention provides a method, an apparatus, a computer device and a storage medium for constructing a map of a target area to solve the problems of high production cost and low accuracy of maps in specific areas.

[0006] In a first aspect, the present invention provides a method for constructing a map of a target area, the method comprising:

[0007] Obtaining vehicle driving information and road pixel data when the vehicle travels in the target area multiple times;

[0008] Performing clustering on the road pixel data to obtain vectorized graphic data representing different types of the target area;

[0009] Performing data cleaning on the vectorized graphic data according to the type of the vectorized graphic data and the vehicle driving information;

[0010] Performing clustering and fitting based on the cleaned vectorized graphic data to obtain a map of the target area.

[0011] Thus, by obtaining the vehicle driving information and road pixel data when the vehicle travels in the target area multiple times, then clustering based on the road pixel data to obtain vectorized graphic data of different types representing the target area, and then performing data cleaning on the vectorized graphic data according to the type of the vectorized graphic data and the vehicle driving information to filter out unnecessary data information, and finally performing clustering and fitting based on the cleaned vectorized graphic data to obtain the map of the target area. Thus, by collecting the road information and driving information of the same area by the vehicle itself multiple times, analyzing and processing the data collected multiple times, and finally forming a high-precision map for driving navigation, there is no need for on-site survey and test by humans, the production cost is low, and more road information can be provided for vehicle driving with higher precision.

[0012] In an alternative embodiment, performing data cleaning on the vectorized graphic data according to the type of the vectorized graphic data and the vehicle driving information includes:

[0013] Determining the cleaning method for the vectorized graphic data according to the type of the vectorized graphic data;

[0014] Performing cleaning on the vectorized graphic data according to the cleaning method and the vehicle driving information.

[0015] Thus, by combining the type of the vectorized graphic data, cleaning different types of vectorized graphic data according to different cleaning methods, filtering out unreasonable data, ensuring data rationality, and improving the accuracy and precision of map construction.

[0016] In an alternative embodiment, different types of vectorized graphic data include lane line vectorized graphic data; performing data cleaning on the vectorized graphic data according to the cleaning method and the vehicle driving information includes:

[0017] For the lane line vectorized graphic data, calculating the lane line azimuth angle according to the coordinate positions of the lane line shape points in the lane line vectorized graphic data;

[0018] Obtaining the trajectory driving direction angle of the vehicle according to the vehicle driving information;

[0019] Calculating the angle difference between the lane line azimuth angle and the trajectory driving direction angle, and determining whether the angle difference is less than a preset angle difference threshold;

[0020] Judging whether the longitudinal relationship of the lane line exists according to the lane line shape points in the lane line vectorized graphic data; wherein, the lane line longitudinal relationship is used to represent the spatial connection relationship of the lane line;

[0021] If the angle difference is less than the preset angle difference threshold and the longitudinal relationship of the lane line exists, then retaining the corresponding lane line vectorized graphic data; otherwise, deleting the corresponding lane line vectorized graphic data.

[0022] Thus, by deleting the vectorized graphic data of the lane lines that do not conform to the vehicle trajectory driving direction and have no longitudinal relationship, the vectorized graphic data is filtered to ensure data rationality and improve the accuracy and precision of map construction.

[0023] In an alternative embodiment, different types of vectorized graphic data include road boundary vectorized graphic data; according to the cleaning method and vehicle driving information, the vectorized graphic data is cleaned, including:

[0024] For the road boundary vectorized graphic data, a road boundary reference line is obtained according to the starting point and ending point of the vectorized graphic in the road boundary vectorized graphic data;

[0025] The shape point farthest from the road boundary reference line in the road boundary vectorized graphic data is selected, and the angle between the farthest shape point and the road boundary reference line is calculated;

[0026] It is determined whether the angle is greater than a preset angle threshold;

[0027] If the angle is greater than the preset angle threshold, the corresponding road boundary vectorized graphic data is retained; otherwise, the corresponding road boundary vectorized graphic data is deleted.

[0028] Thus, by combining the angle of the road boundary vectorized graphic to delete the road boundary vectorized graphic data, the vectorized graphic data is filtered to ensure data rationality and improve the accuracy and precision of map construction.

[0029] In an alternative embodiment, different types of vectorized graphic data include arrow vectorized graphic data; according to the cleaning method and vehicle driving information, the vectorized graphic data is cleaned, including:

[0030] For the arrow vectorized graphic data, the graphic area of the vectorized graphic in the arrow vectorized graphic data is calculated;

[0031] It is determined whether the graphic area conforms to a preset area threshold range;

[0032] If the graphic area conforms to the preset area threshold range, the corresponding arrow vectorized graphic data is retained; otherwise, the corresponding arrow vectorized graphic data is deleted.

[0033] Thus, by judging whether the arrow area is reasonable, the arrow vectorized graphic data is cleaned, the unreasonable data is filtered out, the data rationality is ensured, and the accuracy and precision of map construction are improved.

[0034] In an alternative embodiment, different types of vectorized graphic data include vectorized graphic data of aerial signs; cleaning the vectorized graphic data according to the cleaning method and vehicle driving information includes:

[0035] For the vectorized graphic data of aerial signs, obtain the orientation of the aerial signs in the vectorized graphic data of aerial signs;

[0036] Based on the vehicle driving information, obtain the vehicle steering at the aerial signs;

[0037] Determine whether the orientation of the aerial signs is consistent with the vehicle steering;

[0038] If the orientation of the aerial signs is consistent with the vehicle steering, retain the corresponding vectorized graphic data of the aerial signs; otherwise, delete the corresponding vectorized graphic data of the aerial signs.

[0039] Thus, by determining whether the orientation of the aerial signs is consistent with the vehicle steering, the vectorized graphic data of the aerial signs is cleaned, unreasonable data is filtered out, the data rationality is ensured, and the accuracy and precision of map construction are improved.

[0040] In an alternative embodiment, clustering is performed on the road pixel data to obtain vectorized graphic data representing the target area, including:

[0041] Filter, denoise, and project the road pixel data to obtain point cloud data;

[0042] Perform density clustering on the point cloud data to obtain vectorized graphic data representing the target area.

[0043] Thus, by filtering, denoising, and projecting the road pixel data, point cloud data is obtained, unnecessary noise data is filtered out, and density clustering is performed on the point cloud data to obtain vectorized graphic data representing the target area, and the regional shape information in the map of the target area is obtained.

[0044] In an alternative embodiment, clustering and fitting are performed based on the cleaned vectorized graphic data to obtain a map of the target area, including:

[0045] Perform density clustering on the cleaned vectorized graphic data to obtain different types of clustering clusters;

[0046] Determine the fitting method according to the type of the clustering clusters, and perform data fitting on the clustering clusters according to the fitting method to obtain a target graphic representing the target area;

[0047] Obtain a map of the target area according to the target graphic.

[0048] Thus, by performing density clustering on the vectorized graphic data after cleaning, different types of clustering clusters are obtained, and data fitting is performed on the clustering clusters according to different fitting methods according to the types of the clustering clusters, so as to obtain a target graphic representing the target area. Then, a map of the target area is obtained based on the target graphic, and a high-precision map containing the driving information of the target area can be obtained without manual on-site surveying, which is convenient, fast, simple and efficient.

[0049] In an optional implementation manner, different types of clustering clusters include at least one of the following: lane line clustering clusters, road boundary clustering clusters, arrow clustering clusters, and aerial sign clustering clusters; the fitting methods include line data fitting and surface data fitting.

[0050] In a second aspect, the present invention provides a map construction device for a target area, and the device includes:

[0051] An acquisition module, configured to acquire vehicle driving information and road pixel data when the vehicle travels in the target area multiple times;

[0052] A first processing module, configured to perform clustering on the road pixel data to obtain different types of vectorized graphic data representing the target area;

[0053] A second processing module, configured to perform data cleaning on the vectorized graphic data according to the type of the vectorized graphic data and the vehicle driving information;

[0054] A third processing module, configured to perform clustering and fitting based on the vectorized graphic data after cleaning to obtain a map of the target area.

[0055] In a third aspect, the present invention provides a computer device, including: a memory and a processor, which are communicatively connected to each other, the memory stores computer instructions, and the processor executes the computer instructions to execute the map construction method for the target area in the first aspect or any corresponding implementation manner thereof.

[0056] In a fourth aspect, the present invention provides a computer-readable storage medium, on which computer instructions are stored, and the computer instructions are used to cause a computer to execute the map construction method for the target area in the first aspect or any corresponding implementation manner thereof.

[0057] Advantages of the present invention:

[0058] By obtaining the vehicle driving information and road pixel data when the vehicle travels in the target area multiple times, then clustering according to the road pixel data to obtain vectorized graphic data of different types representing the target area, and then performing data cleaning on the vectorized graphic data according to the type of the vectorized graphic data and the vehicle driving information to filter out unnecessary data information, and finally performing clustering and fitting based on the cleaned vectorized graphic data to obtain the map of the target area. Thus, by collecting the road information and driving information of the same area by the vehicle itself multiple times, analyzing and processing the data collected multiple times, and finally forming a high-precision map for driving navigation, there is no need for on-site survey and test by humans, the production cost is low, and more road information can be provided for vehicle driving with higher accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following will briefly introduce the drawings required for the description of the specific embodiments or the prior art. Obviously, the following drawings are some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0060] Figure 1 It is a flowchart of a method for constructing a map of a target area according to an embodiment of the present invention;

[0061] Figure 2 It is a flowchart of another method for constructing a map of a target area according to an embodiment of the present invention;

[0062] Figure 3 It is a flowchart of yet another method for constructing a map of a target area according to an embodiment of the present invention;

[0063] Figure 4 It is a schematic diagram of the cleaning method of vectorized graphic data of different types according to an embodiment of the present invention;

[0064] Figure 5A It is a schematic diagram of the result of point cloud data output after optimizing and processing the perceived road pixel data according to an embodiment of the present invention;

[0065] Figure 5B It is a schematic diagram of the result of clustering the output point cloud data to form vectorized graphic data according to an embodiment of the present invention;

[0066] Figure 5C It is a schematic diagram of the result of parsing and preprocessing the vectorized graphic data according to an embodiment of the present invention;

[0067] Figure 5D It is a schematic diagram of the result of map construction according to the fitting result according to an embodiment of the present invention;

[0068] Figure 6 is a structural block diagram of a map construction device for a target area according to an embodiment of the present invention;

[0069] Figure 7 is a schematic diagram of the hardware structure of a computer device according to an embodiment of the present invention. Detailed implementation manners

[0070] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Apparently, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0071] A high-precision map, also known as an "autopilot map" or an "intelligent vehicle basic map", is called HDMap (High Definition Map) in English. A high-precision map refers to a high-resolution and high-richness navigation map with both absolute accuracy and relative accuracy in the centimeter level.

[0072] Currently, when making traditional high-precision maps, a large amount of on-site survey work is often required, which requires a lot of manpower and material resources and has a high production cost; moreover, for some sections and areas with few vehicles or not open to the public, it is very unnecessary to spend manpower to make high-precision maps of these sections and areas, which results in low map accuracy in these specific areas and little driving reference significance.

[0073] Therefore, the embodiments of the present invention provide a map construction solution for a target area. When a vehicle lacks a high-precision map of the driving area, the vehicle itself collects road information and driving information of the same section or area multiple times, analyzes and processes the data collected multiple times, and finally forms a high-precision map for autonomous driving navigation, reducing the map construction cost of the high-precision map and completing the map construction work for common sections through daily driving, which is more convenient and fast.

[0074] According to an embodiment of the present invention, an embodiment of a map construction method for a target area is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although the logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in a different order from that here.

[0075] In this embodiment, a method for constructing a map of a target area is provided, which can be used in a computer device or an electronic device capable of performing map construction, such as an on-vehicle computer, an in-vehicle computer, etc. Figure 1 It is a flowchart of the method for constructing a map of a target area according to an embodiment of the present invention, as Figure 1 shown. The process includes the following steps:

[0076] Step S101, obtain vehicle driving information and road pixel data when the vehicle drives in the target area multiple times.

[0077] Specifically, during the process of the vehicle driving in the target area multiple times, elements information such as road markings and aerial signs on the road can be recognized through a front-view camera installed on the vehicle, and further road pixel data can be formed through a visual perception image algorithm.

[0078] Step S102, perform clustering on the road pixel data to obtain vectorized graphic data of different types representing the target area.

[0079] Specifically, after clustering the road pixel data, vectorized graphic data of different types in the target area can be obtained and stored in the form of a binary file. These vectorized graphic data mainly include lane line vectorized graphic data, road boundary vectorized graphic data, arrow vectorized graphic data, and aerial sign vectorized graphic data, covering the main identification graphics in the target area, such as road boundary lines, lane center lines, ground arrows, and aerial signs.

[0080] Step S103, perform data cleaning on the vectorized graphic data according to the type of the vectorized graphic data and the vehicle driving information.

[0081] Specifically, before performing data cleaning on the vectorized graphic data, the vectorized graphic data is first parsed and preprocessed. The process of parsing and preprocessing is mainly to process abnormal data, such as incorrect data timestamp format (for example, not nanoseconds), duplicate trajectory positioning data timestamps, inconsistent data ID and storage path, and inconsistent storage of map semantic data and trajectory data. These abnormal data are deleted to standardize the data and avoid unknown errors in subsequent steps.

[0082] Specifically, when performing data cleaning on the vectorized graphic data, the vectorized graphic data can be grouped according to the timestamps on the vectorized graphic data first, and each group of data is cleaned by type to reduce the workload of data processing and improve the data cleaning efficiency.

[0083] Step S104, perform clustering and fitting based on the cleaned vectorized graphic data to obtain a map of the target area.

[0084] Specifically, by clustering and fitting the vectorized graphic data after cleaning, multiple graphics can be obtained, such as lane lines, road boundary lines, arrow signs, and aerial signs. Combining these obtained graphics can piece together the map of the target area without the need for on-site survey and measurement by humans, with low production costs and providing more road information for vehicle driving with higher accuracy.

[0085] In this way, in the field of autonomous driving, when the vehicle lacks a high-precision map of the driving area, by collecting the road information and driving information of the vehicle on the same road section or area multiple times, analyzing and processing the data collected multiple times, and finally forming a high-precision map for autonomous driving navigation, the mapping cost of the high-precision map is reduced, and the mapping work of the common road sections can be completed through daily driving, which is more convenient and fast.

[0086] The method for constructing a map of a target area provided in this embodiment obtains the vehicle driving information and road pixel data when the vehicle travels in the target area multiple times, then clusters according to the road pixel data to obtain vectorized graphic data of different types representing the target area, then performs data cleaning on the vectorized graphic data according to the type of the vectorized graphic data and the vehicle driving information to filter out unnecessary data information, and finally performs clustering and fitting based on the vectorized graphic data after cleaning to obtain the map of the target area. Thus, by collecting the road information and driving information of the vehicle on the same area multiple times, analyzing and processing the data collected multiple times, and finally forming a high-precision map for driving navigation, without the need for on-site survey and measurement by humans, with low production costs and providing more road information for vehicle driving with higher accuracy.

[0087] In this embodiment, a method for constructing a map of a target area is provided, which can be used in a computer device or an electronic device for map construction, such as an in-vehicle computer, an in-vehicle computer, etc. Figure 2 It is a flowchart of the method for constructing a map of a target area according to an embodiment of the present invention, as Figure 2 shown, and this process includes the following steps:

[0088] Step S201, obtain the vehicle driving information and road pixel data when the vehicle travels in the target area multiple times. For details, please refer to Figure 1 Step S101 of the embodiment shown, which will not be elaborated here.

[0089] Step S202, cluster according to the road pixel data to obtain vectorized graphic data of different types representing the target area.

[0090] Specifically, first filter, denoise, and project the road pixel data to obtain point cloud data. Then, perform density clustering on the point cloud data to obtain vectorized graphic data representing the target area.

[0091] In some alternative embodiments, the road pixel data is filtered and denoised, and the filtered and denoised road pixel data is projected from the vehicle body coordinate system to the world coordinate system to obtain point cloud data in the world coordinate system. The point cloud data is clustered by a density clustering algorithm, and vectorized graphic data of different types is calculated by a concave hull algorithm. It should be noted that the concave hull algorithm is a graph that finds the area occupied by a set of data points, which can express the area occupied by a set of points as realistically as possible. In addition to the concave hull algorithm, other algorithms for finding the area graph according to the point cloud data can also be used to obtain the vectorized graphic data representing the vehicle driving area, and the present invention is not limited thereto.

[0092] Thus, by filtering, denoising and projecting the road pixel data, point cloud data is obtained, unnecessary noise data is filtered out, and the point cloud data is density-clustered to obtain vectorized graphic data representing the target area, and the area shape information in the map of the target area is obtained.

[0093] Step S203: Clean the vectorized graphic data according to the type of the vectorized graphic data and the vehicle driving information.

[0094] Specifically, the above step S203 includes:

[0095] Step S2031: Determine the cleaning method for the vectorized graphic data according to the type of the vectorized graphic data.

[0096] Specifically, the types of the vectorized graphic data mainly include lane lines, road boundaries, arrows and overhead signs. For different types of vectorized graphic data, their data cleaning methods are also different.

[0097] Step S2032: Clean the vectorized graphic data according to the cleaning method and the vehicle driving information.

[0098] In some alternative embodiments, the cleaning method corresponding to the lane line vectorized graphic data is as follows:

[0099] Step a1: For the lane line vectorized graphic data, calculate the lane line azimuth angle according to the coordinate positions of the lane line shape points in the lane line vectorized graphic data.

[0100] Specifically, the coordinate position (x1, y1) of the lane line starting point and the coordinate position (x2, y2) of the lane line ending point can be obtained, and the lane line azimuth angle is calculated according to tan -1 (x1 - x2, y1 - y2).

[0101] Step a2: Obtain the trajectory driving direction angle of the vehicle according to the vehicle driving information.

[0102] Specifically, according to the timestamp on the lane line vectorized graphic data, the trajectory driving direction angle of the vehicle that is consistent with the timestamp of the lane line vectorized graphic data can be selected from the vehicle driving information.

[0103] Step a3, calculate the angle difference between the lane line azimuth angle and the trajectory driving direction angle, and determine whether the angle difference is less than a preset angle difference threshold.

[0104] Specifically, the preset angle difference threshold can be set according to the actual accuracy requirements of map construction. The higher the accuracy requirement, the smaller the preset angle difference threshold.

[0105] Step a4, judge whether the longitudinal relationship of the lane line exists according to the lane line shape points in the lane line vectorized graphic data.

[0106] Among them, the lane line longitudinal relationship is used to characterize the spatial connection relationship of the lane line. In the actual scenario, the lane line is a discontinuous connected line, and this can be used to judge whether the longitudinal relationship of the lane line exists.

[0107] Step a5, if the angle difference is less than the preset angle difference threshold and the longitudinal relationship of the lane line exists, then retain the corresponding lane line vectorized graphic data; otherwise, delete the corresponding lane line vectorized graphic data.

[0108] Thus, by deleting the lane line vectorized graphic data that does not conform to the vehicle trajectory driving direction and does not have a longitudinal relationship, the vectorized graphic data is filtered to ensure data rationality and improve the accuracy and precision of map construction.

[0109] In some alternative embodiments, the cleaning method corresponding to the road boundary vectorized graphic data is as follows:

[0110] Step b1, for the road boundary vectorized graphic data, obtain the road boundary reference line according to the starting point and the ending point of the vectorized graphic in the road boundary vectorized graphic data.

[0111] Specifically, connect the starting point and the ending point of the vectorized graphic corresponding to the road boundary to obtain the road boundary reference line.

[0112] Step b2, select the shape point farthest from the road boundary reference line in the road boundary vectorized graphic data, and calculate the included angle between the farthest shape point and the road boundary reference line.

[0113] Step b3, judge whether the included angle is greater than a preset included angle threshold.

[0114] Specifically, the preset included angle threshold can be set according to the actual situation and the map scenario, and the present invention is not limited thereto.

[0115] Step b4, if the included angle is greater than the preset included angle threshold, retain the corresponding vectorized graphic data of the road boundary; otherwise, delete the corresponding vectorized graphic data of the road boundary.

[0116] Thus, by combining the included angle of the vectorized graphic of the road boundary to delete the vectorized graphic data of the road boundary, the vectorized graphic data is filtered to ensure data rationality and improve the accuracy and precision of map construction.

[0117] In some alternative embodiments, the cleaning method for the arrow vectorized graphic data is as follows:

[0118] Step c1, for the arrow vectorized graphic data, calculate the graphic area of the vectorized graphic in the arrow vectorized graphic data.

[0119] Specifically, the detailed calculation method of the graphic area of the vectorized graphic can refer to the description of related technologies and will not be elaborated here.

[0120] Step c2, determine whether the graphic area meets the preset area threshold range.

[0121] Specifically, in the actual scenario, the area size of the arrow signs on the road is often relatively fixed. Therefore, the arrow vectorized graphic data can be cleaned by filtering out the arrows that do not meet the preset area range.

[0122] Step c3, if the graphic area meets the preset area threshold range, retain the corresponding arrow vectorized graphic data; otherwise, delete the corresponding arrow vectorized graphic data.

[0123] Thus, by determining whether the arrow area size is reasonable, the arrow vectorized graphic data is cleaned, the unreasonable data is filtered out, the data rationality is ensured, and the accuracy and precision of map construction are improved.

[0124] In some alternative embodiments, the cleaning method for the aerial sign vectorized graphic data is as follows:

[0125] Step d1, for the aerial sign vectorized graphic data, obtain the orientation of the aerial sign in the aerial sign vectorized graphic data.

[0126] Specifically, the obtained aerial sign vectorized graphic data can be compared with the stored aerial sign turning graphics to obtain the orientation of the aerial sign corresponding to the aerial sign vectorized graphic data.

[0127] Step d2, based on the vehicle driving information, obtain the vehicle turning at the aerial sign.

[0128] Specifically, according to the timestamps on the vectorized graphic data of the aerial signs, the vehicle steering that is consistent with the timestamps of the vectorized graphic data of the aerial signs can be selected from the vehicle driving information.

[0129] Step d3: Determine whether the orientation of the aerial sign is consistent with the vehicle steering.

[0130] Step d4: If the orientation of the aerial sign is consistent with the vehicle steering, retain the corresponding vectorized graphic data of the aerial sign; otherwise, delete the corresponding vectorized graphic data of the aerial sign.

[0131] Thus, by determining whether the orientation of the aerial sign is consistent with the vehicle steering, the vectorized graphic data of the aerial sign is cleaned, unreasonable data is filtered out, the data rationality is ensured, and the accuracy and precision of map construction are improved.

[0132] Thus, by combining the types of vectorized graphic data, different types of vectorized graphic data are cleaned according to different cleaning methods, unreasonable data is filtered out, the data rationality is ensured, and the accuracy and precision of map construction are improved.

[0133] Step S204: Based on the cleaned vectorized graphic data, perform clustering and fitting to obtain the map of the target area.

[0134] Specifically, the above step S204 includes:

[0135] Step S2041: Perform density clustering on the cleaned vectorized graphic data to obtain different types of clustering clusters.

[0136] Specifically, according to the relative positional relationship of the cleaned vectorized graphic data, clustering is performed through the DBSCAN clustering algorithm, and based on the data in the same clustering cluster obtained, the coordinates (m, n) of its centroid are calculated through the following formula:

[0137]

[0138] where mq represents the abscissa of the q-th shape point in the vectorized graphic, and nq represents the ordinate of the q-th shape point in the vectorized graphic.

[0139] Then, according to the coordinate position of the centroid, a unique ID value is calculated through the Geohash algorithm, and the corresponding clustering cluster is assigned the ID value. It should be noted that the Geohash algorithm is a geocoding method that encodes longitude and latitude into a string composed of numbers and letters, which can help hide the exact location information and protect privacy.

[0140] Specifically, after density clustering of the cleaned vectorized graphic data, different types of clustering clusters can be obtained. The different types of clustering clusters include at least one of the following: lane line clustering clusters, road boundary clustering clusters, arrow clustering clusters, and aerial sign clustering clusters.

[0141] Step S2042: Determine the fitting method according to the type of the clustering cluster, and perform data fitting on the clustering cluster according to the fitting method to obtain a target graphic representing the target area.

[0142] Specifically, fit the clustering clusters with the same ID value. The fitting method can be determined by combining the specification requirements of the high-precision map and the type of the clustering cluster (mainly including line data fitting and surface data fitting). For example, for lane line clustering clusters and road boundary clustering clusters, line data fitting can be performed, and for arrow clustering clusters and aerial sign clustering clusters, surface data fitting can be performed, so as to obtain corresponding lane line graphics, road boundary graphics, arrow graphics, or aerial sign graphics.

[0143] In some alternative embodiments, the fitting methods mainly include line data fitting and surface data fitting. The method of line data fitting is the binomial fitting algorithm, and the method of surface data fitting is the concave hull algorithm. Additionally, other graphic fitting algorithms can be selected according to specific scenario requirements, and the present invention is not limited thereto.

[0144] Step S2043: Obtain the map of the target area according to the target graphic.

[0145] Specifically, according to the fitted lane line graphics, road boundary graphics, arrow graphics, or aerial sign graphics, these graphics can be combined according to their coordinate positions to obtain the map of the target area, so as to provide a reference for vehicle driving.

[0146] In some alternative embodiments, for the fitted lane line graphics, longitudinally connect the discontinuous lane lines, and at the position where the lane line is incomplete, supplement the lane line by combining the trajectory positioning data, so as to obtain the lane center line of the road in the target area.

[0147] Thus, by performing density clustering on the cleaned vectorized graphic data, different types of clustering clusters are obtained, and according to the type of the clustering cluster, data fitting is performed on the clustering cluster according to different fitting methods to obtain a target graphic representing the target area. Then, according to the target graphic, the map of the target area is obtained. Without manual on-site survey, a high-precision map containing driving information of the target area can be obtained, which is convenient, fast, simple, and efficient.

[0148] The map construction method for the target area provided in this embodiment obtains the vehicle driving information and road pixel data when the vehicle drives in the target area multiple times. By performing filtering, noise reduction, and projection processing on the road pixel data, point cloud data is obtained, unnecessary noise data is filtered out, and density clustering is performed on the point cloud data to obtain vectorized graphic data representing the target area, and the area shape information in the map of the target area is obtained. Then, according to the type of the vectorized graphic data and the vehicle driving information, different methods are used to clean the vectorized graphic data to filter out unnecessary data information. Finally, clustering and fitting are performed based on the cleaned vectorized graphic data to obtain graphics such as lane lines and road boundaries, and the map of the target area is obtained. Thus, by collecting the road information and driving information of the same area by the vehicle itself multiple times, analyzing and processing the data collected multiple times, a high-precision map is finally formed for driving navigation, without the need for manual on-site survey and testing, with low production costs, and can provide more road information for vehicle driving and higher accuracy.

[0149] The following further illustrates the map construction method for the target area of the present invention in combination with a specific application example. As Figure 3 shown, the specific application example includes the following steps:

[0150] Step 1, identify road surface elements. The vehicle-end perception module perceives road information, and uses the front-view camera installed on the vehicle to identify element information such as road markings and overhead signs on the road, and further forms road pixel data through visual perception image algorithms.

[0151] Step 2, optimize the perceived road pixel data and output point cloud data. Filter, denoise, and project the road pixel data to obtain point cloud data.

[0152] Step 3, cluster the output point cloud data to form vectorized graphic data. Cluster the point cloud data through a density clustering algorithm, and calculate different types of vectorized graphic data through a concave hull algorithm.

[0153] Step 4, analyze and preprocess the vectorized graphic data. The process of analysis and preprocessing mainly deals with abnormal data, such as incorrect data timestamp format (for example, not nanoseconds), duplicate trajectory positioning data timestamps, inconsistent data ID and storage path, and inconsistent storage of map semantic data and trajectory data. Delete these abnormal data to standardize the data and avoid unknown errors in subsequent steps.

[0154] Step 5, perform data cleaning according to the type of vectorized graphic data. Figure 4 is the cleaning method for different types of vectorized graphic data.

[0155] AsFigure 4 As shown, for lane line data, obtain the coordinate positions (x1, y1) of the starting point of the lane line and the coordinate positions (x2, y2) of the ending point of the lane line, and calculate the azimuth angle of the lane line according to tan -1 (x1 - x2, y1 - y2). Then, according to the vehicle driving information, obtain the trajectory driving direction angle of the vehicle. Then calculate the angle difference between the azimuth angle of the lane line and the trajectory driving direction angle, and determine whether the angle difference is less than the preset angle difference threshold. According to the lane line shape points in the vectorized graphic data of the lane line, determine whether the longitudinal relationship of the lane line exists. If all are satisfied, retain the data; otherwise, delete this piece of data.

[0156] For road boundary data, determine whether the included angle between the road boundary and the trajectory driving direction of the vehicle meets the threshold. Determine whether the included angle between the farthest shape point and the line connecting the starting and ending points is greater than the threshold. If it is satisfied, retain the data; otherwise, delete this piece of data.

[0157] For arrow data, determine whether the area of the vectorized graphic meets the threshold. If it is satisfied, retain the data; otherwise, delete this piece of data.

[0158] For aerial sign data, determine whether the orientation of the aerial sign is reasonable. If it is not reasonable, delete the data; if it is reasonable, retain the data.

[0159] Step 6: Use the DBSCAN clustering algorithm to cluster the cleaned vectorized graphic data.

[0160] Step 7: Fit the clustering clusters obtained by clustering. The fitting method can be determined in combination with the type of clustering clusters. For example, for lane line clustering clusters and road boundary clustering clusters, line data fitting can be performed, and for arrow clustering clusters and aerial sign clustering clusters, surface data fitting can be performed, so as to obtain the corresponding lane line graphics, road boundary graphics, arrow graphics or aerial sign graphics.

[0161] Step 8: Build a map according to the fitting results. According to the lane line graphics, road boundary graphics, arrow graphics or aerial sign graphics obtained by fitting, these graphics can be combined according to the coordinate positions to obtain the map of the target area, so as to provide a reference for vehicle driving.

[0162] Specifically, in the specific application example provided by the present invention, the point cloud data output after optimizing and processing the perceived road pixel data in step 2 is as Figure 5A shown; the vectorized graphic data formed by clustering the output point cloud data in step 3 is as Figure 5B shown; the result after parsing and preprocessing the vectorized graphic data in step 4 is as Figure 5C shown; the result of building a map according to the fitting results in step 8 is as Figure 5DAs shown in the figure. In this way, in the field of autonomous driving, when a vehicle lacks a high-precision map of the driving area, the vehicle can collect road information and driving information on the same road section or area multiple times, analyze and process the data collected multiple times, and finally form a high-precision map for autonomous driving navigation, reducing the mapping cost of the high-precision map. The mapping of common road sections can be completed through daily driving, which is more convenient and fast.

[0163] In this embodiment, a map construction device for a target area is also provided. This device is used to implement the above-mentioned embodiments and preferred implementation manners, and those that have been described will not be repeated. As used hereinafter, the term "module" can be a combination of software and / or hardware that can achieve a predetermined function. Although the devices described in the following embodiments are preferably implemented in software, implementation in hardware, or a combination of software and hardware is also possible and contemplated.

[0164] This embodiment provides a map construction device for a target area, as Figure 6 shown, including:

[0165] An acquisition module 601, configured to acquire vehicle driving information and road pixel data when the vehicle travels in the target area multiple times;

[0166] A first processing module 602, configured to perform clustering on the road pixel data to obtain vectorized graphic data representing different types of the target area;

[0167] A second processing module 603, configured to perform data cleaning on the vectorized graphic data according to the type of the vectorized graphic data and the vehicle driving information;

[0168] A third processing module 604, configured to perform clustering and fitting on the cleaned vectorized graphic data to obtain a map of the target area.

[0169] In some alternative implementation manners, the first processing module 602 includes:

[0170] A first processing unit, configured to filter, denoise, and project the road pixel data to obtain point cloud data;

[0171] A second processing unit, configured to perform density clustering on the point cloud data to obtain vectorized graphic data representing the target area.

[0172] In some alternative implementation manners, the second processing module 603 includes:

[0173] A third processing unit, configured to determine a cleaning method for the vectorized graphic data according to the type of the vectorized graphic data;

[0174] A fourth processing unit, configured to perform cleaning on the vectorized graphic data according to the cleaning method and the vehicle driving information.

[0175] In some alternative embodiments, different types of vectorized graphic data include lane line vectorized graphic data; the fourth processing unit includes:

[0176] A first processing subunit, configured to, for the lane line vectorized graphic data, calculate the azimuth angle of the lane line according to the coordinate positions of the lane line shape points in the lane line vectorized graphic data; obtain the trajectory driving direction angle of the vehicle according to the vehicle driving information; calculate the angle difference between the azimuth angle of the lane line and the trajectory driving direction angle, and determine whether the angle difference is less than a preset angle difference threshold; determine whether the longitudinal relationship of the lane line exists according to the lane line shape points in the lane line vectorized graphic data; wherein the longitudinal relationship of the lane line is used to represent the spatial connection relationship of the lane line; if the angle difference is less than the preset angle difference threshold and the longitudinal relationship of the lane line exists, retain the corresponding lane line vectorized graphic data; otherwise, delete the corresponding lane line vectorized graphic data.

[0177] In some alternative embodiments, different types of vectorized graphic data include road boundary vectorized graphic data; the fourth processing unit includes:

[0178] A second processing subunit, configured to, for the road boundary vectorized graphic data, obtain the road boundary reference line according to the starting point and the ending point of the vectorized graphic in the road boundary vectorized graphic data; select the shape point farthest from the road boundary reference line in the road boundary vectorized graphic data, and calculate the included angle between the farthest shape point and the road boundary reference line; determine whether the included angle is greater than a preset included angle threshold; if the included angle is greater than the preset included angle threshold, retain the corresponding road boundary vectorized graphic data; otherwise, delete the corresponding road boundary vectorized graphic data.

[0179] In some alternative embodiments, different types of vectorized graphic data include arrow vectorized graphic data; the fourth processing unit includes:

[0180] A third processing subunit, configured to, for the arrow vectorized graphic data, calculate the graphic area of the vectorized graphic in the arrow vectorized graphic data; determine whether the graphic area meets a preset area threshold range; if the graphic area meets the preset area threshold range, retain the corresponding arrow vectorized graphic data; otherwise, delete the corresponding arrow vectorized graphic data.

[0181] In some alternative embodiments, different types of vectorized graphic data include aerial sign vectorized graphic data; the fourth processing unit includes:

[0182] A fourth processing subunit is configured to obtain the orientation of the in-air sign vector graphic data for the in-air sign vector graphic data; obtain the vehicle steering at the in-air sign based on the vehicle driving information; determine whether the in-air sign orientation is consistent with the vehicle steering; if the in-air sign orientation is consistent with the vehicle steering, retain the corresponding in-air sign vector graphic data; otherwise, delete the corresponding in-air sign vector graphic data.

[0183] In some alternative embodiments, the third processing module 604 includes:

[0184] A fifth processing unit is configured to perform density clustering on the cleaned vector graphic data to obtain different types of clustering clusters;

[0185] A sixth processing unit is configured to determine a fitting method according to the type of the clustering cluster, and perform data fitting on the clustering cluster according to the fitting method to obtain a target graphic representing the target area;

[0186] A seventh processing unit is configured to obtain a map of the target area according to the target graphic.

[0187] In some alternative embodiments, the different types of clustering clusters in the fifth processing unit include at least one of the following: lane line clustering clusters, road boundary clustering clusters, arrow clustering clusters, and in-air sign clustering clusters; the fitting methods in the sixth processing unit include line data fitting and surface data fitting.

[0188] The further function descriptions of the above-mentioned various modules and units are the same as those in the corresponding foregoing embodiments, and will not be elaborated herein.

[0189] The map construction device for the target area in this embodiment is presented in the form of functional units. Here, the unit refers to an ASIC (Application Specific Integrated Circuit) circuit, a processor and a memory that execute one or more software or fixed programs, and / or other devices that can provide the above functions.

[0190] The embodiment of the present invention further provides a computer device having the above-mentioned Figure 6 shown map construction device for the target area.

[0191] Please refer to Figure 7 , Figure 7 which is a schematic structural diagram of a computer device provided by an alternative embodiment of the present invention. As shown in Figure 7As shown, the computer device includes: one or more processors 10, a memory 20, and interfaces for connecting the components, including high-speed interfaces and low-speed interfaces. Each component communicates with each other using different buses and can be installed on a common motherboard or installed in other ways as needed. The processor can process instructions executed within the computer device, including instructions stored in the memory or on the memory to display graphical information of the GUI on an external input / output device (such as a display device coupled to the interface). In some alternative embodiments, if necessary, multiple processors and / or multiple buses can be used together with multiple memories and multiple memories. Similarly, multiple computer devices can be connected, and each device provides some necessary operations (such as an array of servers, a set of blade servers, or a multi-processor system). Figure 7 In the example, one processor 10 is taken.

[0192] The processor 10 can be a central processing unit, a network processor, or a combination thereof. Among them, the processor 10 can further include a hardware chip. The above hardware chip can be an application-specific integrated circuit, a programmable logic device, or a combination thereof. The above programmable logic device can be a complex programmable logic device, a field programmable gate array, a general array logic, or any combination thereof.

[0193] Among them, the memory 20 stores instructions executable by at least one processor 10, so that the at least one processor 10 executes the method shown in the above embodiments.

[0194] The memory 20 can include a program storage area and a data storage area. Among them, the program storage area can store an operating system and application programs required for at least one function; the data storage area can store data created according to the use of the computer device. In addition, the memory 20 can include high-speed random access memory, and can also include non-transitory memory, such as at least one disk storage device, a flash memory device, or other non-transitory solid-state storage devices. In some alternative embodiments, the memory 20 can optionally include a memory remotely set relative to the processor 10, and these remote memories can be connected to the computer device through a network. Examples of the above network include but are not limited to the Internet, an enterprise intranet, a local area network, a mobile communication network, and combinations thereof.

[0195] The memory 20 can include volatile memory, such as random access memory; the memory can also include non-volatile memory, such as flash memory, a hard disk, or a solid-state drive; the memory 20 can also include a combination of the above types of memory.

[0196] The computer device further includes a communication interface 30 for the computer device to communicate with other devices or communication networks.

[0197] Embodiments of the present invention also provide a computer-readable storage medium. The method according to the embodiments of the present invention can be implemented in hardware, firmware, or be implemented as computer code that can be recorded on a storage medium, or be implemented by downloading over a network and originally stored in a remote storage medium or a non-transitory machine-readable storage medium and to be stored in a local storage medium, so that the method described herein can be stored as such software processing on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. Among them, the storage medium can be a magnetic disk, an optical disk, a read-only memory, a random access memory, a flash memory, a hard disk, or a solid-state drive, etc.; further, the storage medium can also include a combination of the above-mentioned types of memories. It can be understood that a computer, a processor, a microprocessor controller, or programmable hardware includes a storage component that can store or receive software or computer code, and when the software or computer code is accessed and executed by the computer, the processor, or the hardware, the method shown in the above embodiments is implemented.

[0198] Although embodiments of the present invention have been described in conjunction with the accompanying drawings, those skilled in the art can make various modifications and variations without departing from the spirit and scope of the present invention, and such modifications and variations all fall within the scope defined by the appended claims.

Claims

1. A method for constructing a map of a target area, characterized in that, The method includes: Obtaining vehicle driving information and road pixel data when the vehicle travels in the target area multiple times; Performing clustering on the road pixel data to obtain vectorized graphic data of different types representing the target area; Performing data cleaning on the vectorized graphic data according to the type of the vectorized graphic data and the vehicle driving information; Performing clustering and fitting based on the cleaned vectorized graphic data to obtain the map of the target area.

2. The method according to claim 1, characterized in that, The performing data cleaning on the vectorized graphic data according to the type of the vectorized graphic data and the vehicle driving information includes: Determining a cleaning method for the vectorized graphic data according to the type of the vectorized graphic data; Performing cleaning on the vectorized graphic data according to the cleaning method and the vehicle driving information.

3. The method according to claim 2, characterized in that, The vectorized graphic data of different types includes lane line vectorized graphic data; the performing cleaning on the vectorized graphic data according to the cleaning method and the vehicle driving information includes: For the lane line vectorized graphic data, calculating the azimuth angle of the lane line according to the coordinate positions of the shape points of the lane line in the lane line vectorized graphic data; Obtaining the trajectory driving direction angle of the vehicle according to the vehicle driving information; Calculating the angle difference between the azimuth angle of the lane line and the trajectory driving direction angle, and determining whether the angle difference is less than a preset angle difference threshold; Judging whether the longitudinal relationship of the lane line exists according to the shape points of the lane line in the lane line vectorized graphic data; wherein, the lane line longitudinal relationship is used to represent the spatial connection relationship of the lane line; If the angle difference is less than the preset angle difference threshold and the longitudinal relationship of the lane line exists, retaining the corresponding lane line vectorized graphic data; otherwise, deleting the corresponding lane line vectorized graphic data.

4. The method according to claim 2, characterized in that, The vectorized graphic data of different types includes road boundary vectorized graphic data; the performing cleaning on the vectorized graphic data according to the cleaning method and the vehicle driving information includes: For the road boundary vectorized graphic data, obtaining a road boundary reference line according to the starting point and the ending point of the vectorized graphic in the road boundary vectorized graphic data; Selecting the shape point farthest from the road boundary reference line in the road boundary vectorized graphic data, and calculating the included angle between the farthest shape point and the road boundary reference line; Judging whether the included angle is greater than a preset included angle threshold; If the included angle is greater than the preset included angle threshold, retaining the corresponding road boundary vectorized graphic data; otherwise, deleting the corresponding road boundary vectorized graphic data.

5. The method according to claim 2, characterized in that, The vectorized graphic data of different types includes arrow vectorized graphic data; the performing cleaning on the vectorized graphic data according to the cleaning method and the vehicle driving information includes: For the arrow vectorized graphic data, calculating the graphic area of the vectorized graphic in the arrow vectorized graphic data; Judging whether the graphic area meets a preset area threshold range; If the graphic area meets the preset area threshold range, retaining the corresponding arrow vectorized graphic data; otherwise, deleting the corresponding arrow vectorized graphic data.

6. The method according to claim 2, characterized in that, The different types of vectorized graphic data include vectorized graphic data of aerial signs; the cleaning of the vectorized graphic data according to the cleaning method and the vehicle driving information includes: For the vectorized graphic data of aerial signs, obtain the orientation of the aerial signs of the vectorized graphic data of aerial signs; Based on the vehicle driving information, obtain the vehicle steering at the aerial signs; Determine whether the orientation of the aerial signs is consistent with the vehicle steering; If the orientation of the aerial signs is consistent with the vehicle steering, retain the corresponding vectorized graphic data of the aerial signs; otherwise, delete the corresponding vectorized graphic data of the aerial signs.

7. The method according to claim 1, characterized in that, The clustering of the road pixel data to obtain the vectorized graphic data representing the target area includes: Filter, denoise, and project the road pixel data to obtain point cloud data; Perform density clustering on the point cloud data to obtain the vectorized graphic data representing the target area.

8. The method according to any one of claims 1 to 7, characterized in that, The clustering and fitting based on the cleaned vectorized graphic data to obtain the map of the target area includes: Perform density clustering on the cleaned vectorized graphic data to obtain different types of clustering clusters; Determine the fitting method according to the type of the clustering clusters, and perform data fitting on the clustering clusters according to the fitting method to obtain the target graphic representing the target area; Obtain the map of the target area according to the target graphic.

9. The method according to claim 8, wherein, The different types of clustering clusters include at least one of the following: lane line clustering clusters, road boundary clustering clusters, arrow clustering clusters, and aerial sign clustering clusters; the fitting methods include line data fitting and surface data fitting.

10. A map construction device for a target area, wherein, The device includes: An acquisition module for acquiring vehicle driving information and road pixel data when the vehicle drives in the target area multiple times; A first processing module for clustering the road pixel data to obtain different types of vectorized graphic data representing the target area; A second processing module for cleaning the vectorized graphic data according to the type of the vectorized graphic data and the vehicle driving information; A third processing module for clustering and fitting based on the cleaned vectorized graphic data to obtain the map of the target area.

11. A computer device, wherein, Includes: A memory and a processor, which are communicatively connected to each other. The memory stores computer instructions, and the processor executes the computer instructions to execute the method for constructing the map of the target area according to any one of claims 1 to 9.

12. A computer-readable storage medium, wherein, Computer instructions are stored on the computer-readable storage medium, and the computer instructions are used to cause the computer to execute the method for constructing the map of the target area according to any one of claims 1 to 9.

Images