Map generation method and device, computer device and storage medium

Abstract

The application relates to a map generation method and device, computer equipment, a storage medium and a computer program product. The method comprises the following steps: acquiring image data of a target road surface collected by a target camera; adopting an image segmentation algorithm to identify a lane element category to which each data point contained in the image data belongs; based on parameter information of the target camera and first position information of each data point in the image data, determining second position information of the data points corresponding to each lane element category in a world coordinate system; based on the second position information of the data points corresponding to each lane element category, clustering the data points corresponding to each lane element category, and obtaining target position information of a lane element based on the second position information of the clustered data points; and based on the target position information of each lane element and the lane element category, generating a map corresponding to the target road surface. The method can reduce the production cost of a high-precision map, and can also improve the map production efficiency and update efficiency.

Description

Technical Field

[0001] This application relates to the field of autonomous driving technology, and in particular to a map generation method, apparatus, computer device, storage medium, and computer program product. Background Technology

[0002] With the development of autonomous driving, high-definition maps, acting as the brain of autonomous driving, play a crucial role. High-definition maps provide an environmental model of the autonomous vehicle's environment, including lane markings, directional arrows, and other lane element information, as well as other road conditions and surrounding environment information. High-definition map vectorization technology, a key technique in the high-definition map production process, transforms road semantic information into vectorized information usable by autonomous vehicles, playing a vital role in the field of autonomous driving.

[0003] In related technologies, the method of creating high-precision maps can utilize LiDAR and high-precision inertial navigation technology to build maps. This involves stitching together multiple frames of point clouds into a point cloud base map, and then vectorizing and extracting the semantic elements of the point cloud base map to create the map.

[0004] However, the aforementioned method of creating high-precision maps using lidar and inertial navigation technologies involves high-cost equipment, resulting in high production costs for high-precision maps. Summary of the Invention

[0005] Therefore, it is necessary to provide a map generation method, apparatus, computer equipment, computer-readable storage medium, and computer program product that can reduce the cost of producing high-precision maps, in response to the aforementioned technical problems.

[0006] Firstly, this application provides a map generation method. The method includes:

[0007] Acquire image data of the target road surface captured by the target camera;

[0008] An image segmentation algorithm is used to identify the lane feature category to which each data point in the image data belongs; each data point corresponds to a pixel in the image data;

[0009] Based on the parameter information of the target camera and the first position information of the data points corresponding to each lane element category in the image data, the second position information of the data points corresponding to each lane element category in the world coordinate system is determined.

[0010] For each lane element category, based on the second location information of the data points corresponding to the lane element category, the data points corresponding to the lane element category are clustered, and the target location information of the lane element of the lane element category is obtained based on the second location information of the clustered data points.

[0011] Based on the target location information of each lane element and the lane element category of each lane element, a map corresponding to the target road surface is generated.

[0012] In one embodiment, determining the second position information of the data points corresponding to each lane element category in the world coordinate system based on the parameter information of the target camera and the first position information of the data points corresponding to each lane element category in the image data includes:

[0013] Obtain the parameter information of the target camera, including the installation angle information of the target camera relative to the target road surface, and the installation angle information and position information of the target camera in the world coordinate system;

[0014] Based on the parameter information of the target camera, the first position information of the data points corresponding to each lane element category in the image data is transformed by coordinate transformation to obtain the second position information of the data points corresponding to each lane element category in the world coordinate system.

[0015] In one embodiment, obtaining the target location information of lane features of the lane feature category based on the second location information of the clustered data points includes:

[0016] Based on the second location information of the clustered data points, the data points corresponding to each lane element of the lane element category are determined;

[0017] For each lane element, the target location information of the lane element is determined based on the second location information of the data point corresponding to the lane element and the location information determination strategy corresponding to the lane element category.

[0018] In one embodiment, the determination of the target location information of the lane element based on the second location information of the data points corresponding to the lane element and the location information determination strategy corresponding to the lane element category includes:

[0019] Based on the second location information of the data points corresponding to the lane features, determine the minimum bounding rectangle corresponding to the lane features;

[0020] The position information of the minimum bounding rectangle is determined as the position information of the lane element.

[0021] In one embodiment, the determination of the target location information of the lane element based on the second location information of the data points corresponding to the lane element and the location information determination strategy corresponding to the lane element category includes:

[0022] Based on the second location information of the data points corresponding to the lane elements, the data points corresponding to the lane elements are fitted to obtain the straight line region corresponding to the lane elements.

[0023] The location information of the straight area is determined as the target location information of the lane element.

[0024] In one embodiment, after generating the map corresponding to the target road surface based on the target location information of each lane element and the lane element category of each lane element, the method further includes:

[0025] Obtain a historical map of the target road surface;

[0026] The degree of difference between the historical map and the currently generated map is determined based on the historical location information of each lane element contained in the historical map and the target location information of each lane element contained in the currently generated map.

[0027] Based on the degree of difference, the target map to be used is determined from the historical map and the currently generated map.

[0028] Secondly, this application also provides a map generation apparatus. The apparatus includes:

[0029] The first acquisition module is used to acquire image data of the target road surface collected by the target camera;

[0030] The recognition module is used to identify the lane feature category to which each data point in the image data belongs by using an image segmentation algorithm; each data point corresponds to a pixel in the image data;

[0031] The first determining module is used to determine the second position information of the data points corresponding to each lane element category in the world coordinate system based on the parameter information of the target camera and the first position information of the data points corresponding to each lane element category in the image data.

[0032] The clustering module is used to cluster the data points corresponding to the lane element category for each lane element category based on the second location information of the data points corresponding to the lane element category, and obtain the target location information of the lane element of the lane element category based on the second location information of the clustered data points.

[0033] The generation module is used to generate a map corresponding to the target road surface based on the target location information of each lane element and the lane element category of each lane element.

[0034] In one embodiment, the first determining module is specifically used for:

[0035] Obtain the parameter information of the target camera, including the installation angle information of the target camera relative to the target road surface, and the installation angle information and position information of the target camera in the world coordinate system; based on the parameter information of the target camera, perform coordinate transformation on the first position information of the data points corresponding to each lane element category in the image data to obtain the second position information of the data points corresponding to each lane element category in the world coordinate system.

[0036] In one embodiment, the clustering module is specifically used for:

[0037] Based on the second location information of the clustered data points, the data points corresponding to each lane element of the lane element category are determined; for each lane element, the target location information of the lane element is determined based on the second location information of the data points corresponding to the lane element and the location information determination strategy corresponding to the lane element category.

[0038] In one embodiment, the clustering module is specifically used for:

[0039] Based on the second location information of the data points corresponding to the lane feature, the minimum bounding rectangle corresponding to the lane feature is determined; the location information of the minimum bounding rectangle is determined as the location information of the lane feature.

[0040] In one embodiment, the clustering module is specifically used for:

[0041] Based on the second location information of the data points corresponding to the lane element, the data points corresponding to the lane element are fitted to obtain the straight line region corresponding to the lane element; the location information of the straight line region is determined as the target location information of the lane element.

[0042] In one embodiment, the device further includes:

[0043] The second acquisition module is used to acquire a historical map of the target road surface;

[0044] The second determining module is used to determine the degree of difference between the historical map and the currently generated map based on the historical location information of each lane element contained in the historical map and the target location information of each lane element contained in the currently generated map.

[0045] The third determining module is used to determine the target map to be used from the historical map and the currently generated map based on the degree of difference.

[0046] Thirdly, this application also provides a computer device. The computer device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the steps of the method described in the first aspect.

[0047] Fourthly, this application also provides a computer-readable storage medium. The computer-readable storage medium stores a computer program thereon, which, when executed by a processor, implements the steps of the method described in the first aspect.

[0048] Fifthly, this application also provides a computer program product. The computer program product includes a computer program that, when executed by a processor, implements the steps of the method described in the first aspect.

[0049] The aforementioned map generation method, apparatus, computer equipment, storage medium, and computer program product acquire road surface image data using a camera, identify the lane feature category to which each data point (corresponding pixel) in the road surface image data belongs, and then transform the position information (such as pixel coordinates) of each data point in the image data based on the camera's parameter information to obtain the position information of each data point corresponding to the lane feature category in the world coordinate system. Next, the position information of the data points in the world coordinate system is clustered, and the target position information of lane features in each lane feature category is determined based on the position information of the clustered data points. Thus, based on the target position information of each lane feature and the lane feature category, a map corresponding to the road surface is generated. In this method, low-cost equipment (camera) can be used to acquire image data of the target road surface to generate a high-precision map of the target road surface, thereby reducing the production cost of high-precision maps. Furthermore, this method can also acquire image data and generate maps in real time, improving map production and update efficiency. Attached Figure Description

[0050] Figure 1 This is a flowchart illustrating a map generation method in one embodiment;

[0051] Figure 2 This is a flowchart illustrating the process of determining the second location information of a data point in one embodiment;

[0052] Figure 3 This is a schematic diagram of the process for obtaining target location information of lane elements in one embodiment;

[0053] Figure 4 This is a structural block diagram of a map generation device in one embodiment;

[0054] Figure 5 This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation

[0055] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0056] First, before introducing the technical solutions of the embodiments of this application, we will first introduce the technical background or technical evolution on which the embodiments of this application are based. With the development of autonomous driving, high-precision maps, as the brain of autonomous driving, play an important role. High-precision maps provide an environmental model of the autonomous vehicle's location, including lane markings, directional arrows, and other lane element information, as well as other road conditions and surrounding environment information. Vehicles can obtain vehicle location, road conditions, and surrounding environment information based on high-precision maps in order to plan routes and control vehicle driving strategies. Among them, high-precision map vectorization technology, as one of the most important technologies in the high-precision map production process, transforms road semantic information into vectorized information that autonomous vehicles can use, and plays an important role in the field of autonomous driving.

[0057] In related technologies, high-precision map production methods utilize LiDAR and high-precision inertial navigation technology. This involves stitching together multiple frames of point cloud data collected by LiDAR into a single point cloud base map, then vectorizing and extracting semantic elements from the base map to create the map. However, the high cost of LiDAR equipment leads to high production costs for high-precision maps. Furthermore, due to the high cost of LiDAR equipment, it is difficult to deploy LiDAR equipment on a large scale in road environments for point cloud data collection. Instead, data collection vehicles equipped with LiDAR are typically used to collect point cloud data from various road environments. These vehicles require significant time to travel to the target environment to collect data; therefore, LiDAR-based high-precision map production methods are inefficient and have long update cycles.

[0058] Against this backdrop, through long-term research and development and experimental verification, the applicant proposes the map generation method of this application. This method can use low-cost equipment (camera) to collect information on the target environment (including the target road surface) to generate a high-precision map of the target environment, thereby reducing the production cost of high-precision maps. Furthermore, it can collect data and generate maps in real time, improving map production and update efficiency. It should also be noted that the applicant has devoted considerable creative effort to discovering the technical problem of this application and to developing the technical solutions described in the following embodiments.

[0059] In one embodiment, such as Figure 1As shown, a map generation method is provided, which can be applied to computer devices. The computer device can be, but is not limited to, various personal computers, laptops, smartphones, tablets, IoT devices, vehicle-to-everything (V2X) devices, and servers. In this embodiment, the method includes the following steps:

[0060] Step 101: Acquire image data of the target road surface captured by the target camera.

[0061] In implementation, cameras or other electronic devices with camera functionality can be pre-installed in the target environment (such as the roadside or above the road surface) to collect image data of the target environment (including the target road surface) in real time or at regular intervals. The cameras can upload the collected image data to a cloud server, so that computer equipment can obtain the image data of the target road surface from the cloud server.

[0062] Step 102: Use an image segmentation algorithm to identify the lane feature category to which each data point in the image data belongs.

[0063] In implementation, the target road surface may contain various types of lane elements, such as directional arrows (including those indicating straight ahead, left turn, U-turn, right turn, straight ahead or right turn, and others), lane markings (including single dashed / solid lines, double dashed / solid lines, zebra crossings, and other lane markings), yield signs, and no-stopping signs. Computer equipment can use image segmentation algorithms to identify the lane element category corresponding to each pixel in the target road surface image data. One pixel in the image data can correspond to one data point. Accordingly, the lane element category corresponding to that pixel is the same as the lane element category of the corresponding data point.

[0064] For example, roadside video streams can be collected in autonomous driving demonstration zones, sample images can be obtained from the video streams, and lane feature category labels such as guide arrows and lane lines can be added to the sample images to obtain multiple training samples. Then, a computer device can use the training samples to train a constructed image segmentation model (such as a semantic segmentation model like DDRNet) to obtain a qualified image segmentation model (e.g., a 99% quasi-recall rate after 300 training rounds is considered qualified). Afterwards, the computer device can input the image data of the target road surface into the qualified image segmentation model to obtain the lane feature category corresponding to each pixel in the image data.

[0065] Step 103: Based on the parameter information of the target camera and the first position information of the data points corresponding to each lane element category in the image data, determine the second position information of the data points corresponding to each lane element category in the world coordinate system.

[0066] In implementation, the first positional information refers to the positional information in the image coordinate system (such as pixel coordinates). The image data may contain the first positional information of each pixel (as the first positional information of the data point). The second positional information refers to the positional information in the world coordinate system, that is, the actual spatial positional information of that point on the target road surface. The computer equipment can perform coordinate transformation on the first positional information of each data point identified as a lane feature category based on the parameter information of the target camera, to obtain the second positional information of each data point (i.e., the actual spatial positional information of each point on the target road surface). The parameter information of the target camera may include distortion coefficients, focal length, installation angle, and positional information in the world coordinate system.

[0067] Step 104: For each lane feature category, cluster the data points corresponding to the lane feature category based on the second location information of the data points corresponding to the lane feature category, and obtain the target location information of the lane feature of the lane feature category based on the second location information of the clustered data points.

[0068] In implementation, after the computer equipment identifies the lane feature category to which each data point belongs, it can cluster the data points corresponding to each lane feature category. Since the target road surface may contain multiple lane features belonging to the same lane feature category (such as multiple solid lane lines), and since the accuracy of image segmentation algorithms is difficult to achieve 100%, there may be data points misidentified as belonging to that lane feature category. Therefore, clustering can be performed on data points identified as belonging to the same lane feature category to group them together and remove misidentified isolated points. Clustering can employ algorithms such as K-Means and K-Medoids. For dashed line lane features, dashed lines whose center points are within a preset range (e.g., within 0.3m) are identified as lane features of the same dashed line category. Then, the computer device can obtain the target position information of each lane element based on the second position information of the clustered data points (i.e., the position information in the world coordinate system obtained in step 103). For example, the position information of the contour line composed of data points clustered into the same group can be used as the target position information of the lane element corresponding to that group.

[0069] Step 105: Generate a map corresponding to the target road surface based on the target location information of each lane element and the lane element category of each lane element.

[0070] In implementation, after the computer equipment determines the target location information of each lane element on the target road surface, it can use the target location information of the lane elements and the corresponding lane element categories as high-precision map data for the target road surface to generate a high-precision map. In the visualized high-precision map, the element identifier corresponding to the lane element category can be displayed at the location corresponding to the target position of the lane element. Understandably, when displaying element identifiers in the visualized high-precision map, other road information can also be combined. For example, for lane elements related to vehicle flow (such as directional arrows), element identifiers consistent with that flow direction can be displayed based on the vehicle flow information in the road information. Furthermore, the high-precision map data can also include other information about the environment to which the target road surface belongs. The camera can collect image data of objects other than the road surface in the target environment (such as traffic lights, signs, roadside buildings, etc.). Based on all the collected image data of the target environment, a comprehensive map of the target environment can be constructed as needed; the specific process will not be elaborated here.

[0071] The aforementioned map generation method acquires road surface image data using a camera, identifies the lane feature category to which each data point (corresponding pixel) in the road surface image data belongs, and then transforms the position information (such as pixel coordinates) of each data point in the image data based on the camera's parameter information to obtain the position information of each data point in the world coordinate system corresponding to the lane feature category. Next, the position information of the data points in the world coordinate system is clustered, and the target position information of lane features in each lane feature category is determined based on the position information of the clustered data points. Therefore, a map corresponding to the road surface is generated based on the target position information and lane feature category of each lane feature. This method can use low-cost equipment (camera) to acquire image data of the target road surface to generate a high-precision map of the target road surface, thereby reducing the production cost of high-precision maps. Furthermore, this method can also acquire image data and generate maps in real time, improving map production efficiency.

[0072] In one embodiment, such as Figure 2 As shown, the process of determining the second location information of the data points in step 103 specifically includes the following steps:

[0073] Step 201: Obtain the parameter information of the target camera.

[0074] The parameter information includes the installation angle information of the target camera relative to the target road surface, as well as the installation angle information and position information of the target camera in the world coordinate system.

[0075] In implementation, the parameter information of the target camera installed in the target road environment can be pre-calibrated and stored, allowing computer equipment to acquire this parameter information. For example, a radar-equipped data acquisition vehicle can be used to collect point cloud data of the target road surface (such as at an intersection) and the camera installed in that environment, obtaining point cloud information for both the target road surface and the target camera. Then, based on the point cloud information of the target camera, the outer contour can be determined, and a bounding box can be fitted to obtain the hexahedral information of the target camera's bounding box. Subsequently, based on the hexahedral information and the camera's optical axis, the normal vector of the camera's optical axis and the world coordinates of the camera's optical center can be calculated, thus obtaining the camera's installation angle and position information in the world coordinate system. Finally, the road surface normal vector can be calculated based on the road surface point cloud information, and based on the road surface normal vector and the camera's optical axis normal vector, the installation angle and position information of the camera relative to the target road surface can be calculated.

[0076] Step 202: Based on the parameter information of the target camera, perform coordinate transformation on the first position information of the data points corresponding to each lane element category in the image data to obtain the second position information of the data points corresponding to each lane element category in the world coordinate system.

[0077] In implementation, computer equipment can perform coordinate transformation on the first position information (such as pixel coordinates) of data points identified as lane element categories in the image data based on the parameter information of the target camera, to obtain the second position information of each data point in the world coordinate system. In one example, the position information of each data point in the camera coordinate system can be calculated first based on the first position information of each data point (which can be denoted as [u,v,1], since the position information in the world coordinate system is generally in three-dimensional coordinates, the pixel coordinates are padded with dimension for ease of calculation) and the installation angle information of the camera relative to the center of the target road surface (which can be denoted as [Rc,tc]). v ,y v The calculation formula can be shown below:

[0078]

[0079] Then, the computer device can determine the position information of each data point in the camera coordinate system [x] v ,y v [1], and the camera's installation angle information in the world coordinate system (which can be denoted as R(qi), representing the rotation matrix of the camera's installation angle in the world coordinate system) and position information (which can be denoted as pi), and calculate the position information of each data point in the world coordinate system (which can be denoted as [x]). w ,y w ,z w The calculation formula can be as follows:

[0080]

[0081] In this embodiment, the actual spatial location information of each data point is obtained by using image data collected by the camera and the camera's parameter information. Based on the actual spatial location information of each data point, the spatial location information of the lane element to which each data point belongs can be determined, which can be used as high-precision map data to generate a high-precision map.

[0082] In one embodiment, such as Figure 3 As shown, step 104, the process of obtaining the target location information of lane features based on the clustered data points, specifically includes the following steps:

[0083] Step 301: Based on the second location information of the clustered data points, determine the data points corresponding to each lane element of the lane element category.

[0084] In practice, after clustering data points identified as belonging to the same lane feature category, data points clustered into the same group (such as those that are close in distance) can be identified as data points corresponding to a lane feature of that lane feature category.

[0085] Step 302: For each lane element, determine the target location information of the lane element based on the second location information of the data point corresponding to the lane element and the location information corresponding to the lane element category.

[0086] In implementation, since the characteristics of lane elements of different categories may vary significantly (e.g., directional arrows and solid lane lines differ greatly in length and width), corresponding position information determination strategies can be set for different lane element categories. Then, for each lane element in the target road surface, the computer device can use a strategy corresponding to the category of that lane element, based on the second position information of the data points corresponding to that lane element, to determine the target position information of that lane element. For example, for curved lane lines, a local center point can be determined at intervals within the length range. Connecting these local center points and smoothing the lines results in a smooth line corresponding to the lane line, and the position information of this smooth line can be used as the position information of the lane line. For directional arrows, the area enclosed by the circumscribed rectangle surrounding each data point can be considered the area where the directional arrow is located; therefore, the position information of this circumscribed rectangle can be used as the position information of the directional arrow.

[0087] In this embodiment, after clustering data points identified as belonging to the same category, data points in the same group are treated as data points corresponding to the same lane feature. Then, based on the location information of the data points within that group, a strategy corresponding to the lane feature category is adopted to determine the location information of that lane feature. Thus, the location information of each lane feature contained in the target road surface can be accurately determined and used as map data for map generation.

[0088] In one embodiment, the process of determining the target location information of the lane feature in step 302 specifically includes the following steps: determining the minimum bounding rectangle corresponding to the lane feature based on the second location information of the data points corresponding to the lane feature; and determining the location information of the minimum bounding rectangle as the location information of the lane feature.

[0089] In implementation, lane features can be categorized into two main types based on their shape characteristics: linear lane features (such as solid lines and dashed lines) and non-linear lane features (such as directional arrows and yield signs). For non-linear lane features, the computer equipment can use methods such as the rotating caliper algorithm to calculate the minimum bounding rectangle for each lane feature based on the second position information of the data points corresponding to that type of lane feature. The minimum bounding rectangle will contain all the clustered data points corresponding to that lane feature. The position information of this minimum bounding rectangle reflects the location of the lane feature on the target road surface, and the lane feature's identifier can be contained within this minimum bounding rectangle. For example, the computer equipment can calculate the endpoints of four polygons, denoted as xminP, xmaxP, yminP, and ymaxP, based on the second position information of the data points belonging to the same lane feature obtained from clustering. Then, by constructing four tangents to P through these four points, two sets of "calipers" are determined. If one (or two) lines coincide with an edge, calculate the area of ​​the rectangle defined by the four lines and save it as the current minimum value; otherwise, define the current minimum value as infinity. Then, rotate the lines clockwise until one of them coincides with an edge of the polygon, calculate the area of ​​the new rectangle, and compare it with the current minimum value. If it is less than the current minimum value, update it and save the rectangle information for the determined minimum value. Repeat the steps of rotating the lines and determining the area of ​​the new rectangle until the angle through which the lines have rotated is greater than 90 degrees. Output the minimum area of ​​the bounding rectangle, which is the minimum bounding rectangle corresponding to this lane feature.

[0090] Then, the computer device can determine the position information of the minimum bounding rectangle (such as the position information of the endpoints, and / or the position information of the outline, and / or the position information of the center point, etc.) as the position information of the lane feature (position information in the world coordinate system).

[0091] In this embodiment, for lane feature identifiers that are relatively complex and short and wide, the smallest bounding rectangle containing the data points corresponding to the lane feature can be determined. The lane feature is located within this rectangle, so the position information of the smallest bounding rectangle can be determined as the position information of the lane feature, so as to balance positioning accuracy and positioning efficiency, thereby improving the accuracy and efficiency of map production.

[0092] In another embodiment, the process of determining the target location information of the lane element in step 302 specifically includes the following steps: based on the second location information of the data points corresponding to the lane element, the data points corresponding to the lane element are fitted to obtain the straight line area corresponding to the lane element; the location information of the straight line area is determined as the target location information of the lane element.

[0093] In implementation, for linear lane features, the computer can perform straight-line fitting on the data points corresponding to each lane feature within that category. For example, the computer can first process the data points using the RANSAC (Random Sample Consensus) algorithm, removing outliers and retaining inliers. Then, it can use the least squares method to fit the inliers, obtaining the straight line y = a + bx. Here, x and y correspond to the x-axis position information of the data point in the world coordinate system. w and y w ; a represents the intercept, and b represents the slope, which are two parameters to be determined. For N sets of data (xi, yi) obtained from measurements of equal precision, i = 1, 2, ..., N, the values ​​of xi are considered accurate, and all errors are only related to yi. Therefore, when estimating parameters using the least squares method, the weighted sum of squares of the deviations of the observed values ​​yi should be minimized. For linear fitting of measurements of equal precision, the following formula can be minimized:

[0094]

[0095] Taking the partial derivative of the above equation, we obtain the following system of equations:

[0096]

[0097]

[0098] Solving the above system of equations yields the optimal estimates of the straight line parameters a and b. Based on the location information of the data points corresponding to this lane feature, the location information of the region where the straight line y = a + bx is located can be determined. This region (straight line region) corresponds to the location of the lane feature. Therefore, the location information of this straight line region can be used as the target location information corresponding to the lane feature, serving as high-precision map data to generate an identifier (straight lane line) corresponding to the lane feature category at that location on the map. Understandably, for dashed line lane features, the data points corresponding to this lane feature can be further clustered, with each group of data points corresponding to a segment of dashed line. Then, straight line fitting can be performed on the data points corresponding to each segment of dashed line to obtain multiple shorter straight line regions. These straight line regions are combined to form the region where the lane feature is located.

[0099] In one embodiment, after generating the map corresponding to the target road surface in step 105, the method may further include a map change judgment and update process, specifically including the following steps: obtaining a historical map of the target road surface; determining the degree of difference between the historical map and the currently generated map based on the historical location information of each lane element contained in the historical map and the target location information of each lane element contained in the currently generated map; and determining the target map to be used from the historical map and the currently generated map based on the degree of difference.

[0100] In practice, the computer equipment can determine whether the map data of the target road surface has changed and needs to be updated based on the difference between the historical map of the target road surface (such as a historical map generated based on historical image data collected by the target camera or other methods) and the map generated in step 105 (the currently generated map). Specifically, the computer equipment can match the historical location information of each lane element contained in the historical map with the historical location information of each lane element in the currently generated map to obtain the difference.

[0101] For example, for each lane feature in the currently generated map, the currently matched lane feature can be used as the lane feature to be matched, and the location information of the lane feature to be matched in the currently generated map can be used as the location information to be matched. Then, based on the location information to be matched, a neighborhood search is performed in the historical map to determine the target lane feature that is adjacent to the lane feature to be matched and belongs to the same lane feature category. After that, the image data of the area where the target lane feature is located can be determined according to the location information of the target lane feature, and the image data of the area corresponding to the location information to be matched of the lane feature to be matched can be compared, and the area intersection-union ratio (IoU) of the two can be calculated. The distance between the center points of the two regions can also be calculated. The degree of difference can be determined based on the area intersection-union ratio and the distance between the center points of the regions. Specifically, if the area intersection-union ratio of the two image data is greater than a preset threshold (e.g., IoU>0.6), and the distance between the center points of the two regions is less than a preset threshold (e.g., less than 0.3m), it can be determined that the degree of difference between the historical map and the currently generated map does not meet the preset condition, that is, the historical map and the currently generated map can be regarded as unchanged, so the historical map can be determined as the target map to be used, that is, the map is not updated. If the area intersection ratio is less than or equal to a preset threshold, and / or the distance between the center points of the two areas is greater than or equal to a preset threshold, then the difference is determined to meet the preset conditions. Therefore, the currently generated map can be determined as the target map to be used, that is, the map is updated.

[0102] Understandably, if a neighborhood search is performed in the historical map based on the location information to be matched, and no target lane element adjacent to the lane element to be matched and belonging to the same lane element category is found, that is, the historical map may not contain lane elements of the same category as the lane element to be matched, it indicates that the map data of the target road surface has changed, and the currently generated map can be used as the map to be updated. In addition, computer equipment can also use other methods to determine the degree of difference between the historical map and the currently generated map. For example, for each lane element category, a first number of lane elements of that category can be determined in the historical map of the target road surface, and a second number of lane elements of that category can be determined in the currently generated map. Then, the first number and the second number are compared. If the first number and the second number of lane elements corresponding to each lane element category are the same, the aforementioned step of determining the degree of difference based on the location information of each lane element can be further executed; if the first number and the second number of lane elements corresponding to at least one lane element category are different, it can be determined that the degree of difference between the historical map and the currently generated map meets the condition, that is, the map data has changed, and a map update is required.

[0103] Optionally, different comparison methods can be set for different lane feature categories. For example, the area intersection-union ratio (IUU) method described above can be used for non-linear lane features (which have a corresponding minimum bounding rectangle during map generation). The IUU of the minimum bounding rectangle and the distance to its center point can be directly calculated to determine the degree of difference. For linear lane features, the ICP (Point Cloud Registration) algorithm can be used to calculate the minimum error. When the error exceeds a preset threshold (e.g., greater than 0.3m), the difference condition is met, and the map is considered to have changed and needs to be updated.

[0104] In this embodiment, the difference between the historical map and the currently generated map is determined based on the historical location information of each lane element in the historical map and the location information of each lane element in the currently generated map. This difference allows for a decision on whether to update the map. Consequently, the map can be updated promptly when lane information on the target road surface changes, improving map update efficiency.

[0105] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0106] Based on the same inventive concept, this application also provides a map generation apparatus for implementing the map generation method described above. The solution provided by this apparatus is similar to the implementation described in the above method; therefore, the specific limitations in one or more map generation apparatus embodiments provided below can be found in the limitations of the map generation method described above, and will not be repeated here.

[0107] In one embodiment, such as Figure 4 As shown, a map generation device 400 is provided, including: a first acquisition module 401, an identification module 402, a first determination module 403, a clustering module 404, and a generation module 405, wherein:

[0108] The first acquisition module 401 is used to acquire image data of the target road surface collected by the target camera.

[0109] The recognition module 402 is used to identify the lane feature category to which each data point in the image data belongs by using an image segmentation algorithm; each data point corresponds to a pixel in the image data.

[0110] The first determining module 403 is used to determine the second position information of the data points corresponding to each lane element category in the world coordinate system based on the parameter information of the target camera and the first position information of the data points corresponding to each lane element category in the image data.

[0111] The clustering module 404 is used to cluster the data points corresponding to the lane element category for each lane element category based on the second location information of the data points corresponding to the lane element category, and obtain the target location information of the lane element of the lane element category based on the second location information of the clustered data points.

[0112] The generation module 405 is used to generate a map corresponding to the target road surface based on the target location information of each lane element and the lane element category of each lane element.

[0113] In one embodiment, the first determining module 403 is specifically used to: obtain parameter information of the target camera, the parameter information including the installation angle information of the target camera relative to the target road surface, and the installation angle information and position information of the target camera in the world coordinate system; and perform coordinate transformation on the first position information of the data points corresponding to each lane element category in the image data according to the parameter information of the target camera, to obtain the second position information of the data points corresponding to each lane element category in the world coordinate system.

[0114] In one embodiment, the clustering module 404 is specifically used to: determine the data points corresponding to each lane element of the lane element category based on the second location information of the clustered data points; and for each lane element, determine the target location information of the lane element based on the second location information of the data points corresponding to the lane element and the location information determination strategy corresponding to the lane element category.

[0115] In one embodiment, the clustering module 404 is specifically used to: determine the minimum bounding rectangle corresponding to the lane element based on the second location information of the data points corresponding to the lane element; and determine the location information of the minimum bounding rectangle as the location information of the lane element.

[0116] In one embodiment, the clustering module 404 is specifically used to: fit the data points corresponding to the lane elements based on the second location information of the data points corresponding to the lane elements to obtain the straight line region corresponding to the lane elements; and determine the location information of the straight line region as the target location information of the lane elements.

[0117] In one embodiment, the apparatus further includes a second acquisition module, a second determination module, and a third determination module, wherein:

[0118] The second acquisition module is used to acquire a historical map of the target road surface.

[0119] The second determining module is used to determine the degree of difference between the historical map and the currently generated map based on the historical location information of each lane element contained in the historical map and the target location information of each lane element contained in the currently generated map.

[0120] The third determining module is used to determine the target map to be used from the historical map and the currently generated map based on the degree of difference.

[0121] Each module in the aforementioned map generation device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can call and execute the operations corresponding to each module.

[0122] In one embodiment, a computer device is provided, which may be a terminal, and its internal structure diagram may be as follows: Figure 5 As shown, the computer device includes a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, mobile cellular networks, NFC (Near Field Communication), or other technologies. When the computer program is executed by the processor, it implements a map generation method. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad mounted on the computer device casing, or an external keyboard, touchpad, or mouse.

[0123] Those skilled in the art will understand that Figure 5 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0124] In one embodiment, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above-described method embodiments.

[0125] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the steps in the above method embodiments.

[0126] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above method embodiments.

[0127] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties.

[0128] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0129] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0130] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A map generation method, characterized in that, The method includes: Acquire image data of the target road surface captured by the target camera; An image segmentation algorithm is used to identify the lane feature category to which each data point in the image data belongs; each data point corresponds to a pixel in the image data; Based on the parameter information of the target camera and the first position information of the data points corresponding to each lane element category in the image data, the second position information of the data points corresponding to each lane element category in the world coordinate system is determined. For each lane element category, based on the second location information of the data points corresponding to the lane element category, the data points corresponding to the lane element category are clustered, and based on the second location information of the clustered data points, the data points corresponding to each lane element of the lane element category are determined. For each lane element, the target location information of the lane element is determined based on the second location information of the data point corresponding to the lane element and the location information corresponding to the lane element category. The location information determination strategy includes: For the linear lane feature, based on the second position information of the data points corresponding to the lane feature, the RANSAC algorithm and the least squares method are used to perform straight line fitting on the data points corresponding to the lane feature to obtain the straight line region corresponding to the lane feature, and the position information corresponding to the straight line region is determined as the target position information of the lane feature. For non-linear lane features, the minimum bounding rectangle of each lane feature is calculated based on the second position information of the data points corresponding to the lane feature, and the position information of the minimum bounding rectangle is determined as the target position information of the lane feature. Based on the target location information of each lane element and the lane element category of each lane element, a map corresponding to the target road surface is generated.

2. The method according to claim 1, characterized in that, The determination of the second position information of the data points corresponding to each lane element category in the world coordinate system based on the parameter information of the target camera and the first position information of the data points corresponding to each lane element category in the image data includes: Obtain the parameter information of the target camera, including the installation angle information of the target camera relative to the target road surface, and the installation angle information and position information of the target camera in the world coordinate system; Based on the parameter information of the target camera, the first position information of the data points corresponding to each lane element category in the image data is transformed by coordinate transformation to obtain the second position information of the data points corresponding to each lane element category in the world coordinate system.

3. The method according to claim 1, characterized in that, The clustering of data points corresponding to the lane feature categories includes: After clustering data points identified as belonging to the same category, the data points in the same group are used as the data points corresponding to the same lane feature.

4. The method according to claim 1 or 3, characterized in that, The method further includes: The rotating caliper algorithm is used to calculate the minimum bounding rectangle corresponding to each lane element.

5. The method according to claim 1, characterized in that, After generating the map corresponding to the target road surface based on the target location information and lane feature category of each lane feature, the method further includes: Obtain a historical map of the target road surface; The degree of difference between the historical map and the currently generated map is determined based on the historical location information of each lane element contained in the historical map and the target location information of each lane element contained in the currently generated map. Based on the degree of difference, the target map to be used is determined from the historical map and the currently generated map.

6. A map generation device, characterized in that, The device includes: The first acquisition module is used to acquire image data of the target road surface collected by the target camera; The recognition module is used to identify the lane feature category to which each data point in the image data belongs by using an image segmentation algorithm; each data point corresponds to a pixel in the image data; The first determining module is used to determine the second position information of the data points corresponding to each lane element category in the world coordinate system based on the parameter information of the target camera and the first position information of the data points corresponding to each lane element category in the image data. The clustering module is used to cluster the data points corresponding to the lane element category for each lane element category based on the second location information of the data points corresponding to the lane element category, and to determine the data points corresponding to each lane element of the lane element category based on the second location information of the clustered data points. For each lane element, the target location information of the lane element is determined based on the second location information of the data point corresponding to the lane element and the location information corresponding to the lane element category. The location information determination strategy includes: For the linear lane feature, based on the second position information of the data points corresponding to the lane feature, the RANSAC algorithm and the least squares method are used to perform straight line fitting on the data points corresponding to the lane feature to obtain the straight line region corresponding to the lane feature, and the position information corresponding to the straight line region is determined as the target position information of the lane feature. For non-linear lane features, the minimum bounding rectangle of each lane feature is calculated based on the second position information of the data points corresponding to the lane feature, and the position information of the minimum bounding rectangle is determined as the target position information of the lane feature. The generation module is used to generate a map corresponding to the target road surface based on the target location information of each lane element and the lane element category of each lane element.

7. The apparatus according to claim 6, characterized in that, The first determining module is specifically used for: Obtain the parameter information of the target camera, including the installation angle information of the target camera relative to the target road surface, and the installation angle information and position information of the target camera in the world coordinate system; Based on the parameter information of the target camera, the first position information of the data points corresponding to each lane element category in the image data is transformed by coordinate transformation to obtain the second position information of the data points corresponding to each lane element category in the world coordinate system.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 5.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 5.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 5.

Images