Lane line generation method and device, computer device and storage medium

Abstract

The application relates to a lane line generation method and device and a computer device. The method comprises the following steps: acquiring lane line point cloud data and lane line point cloud data corresponding to lane line identification respectively, and determining target lane line point cloud data corresponding to the same lane line identification; performing point cloud position conversion on the target lane line point cloud data according to target position conversion information to obtain to-be-fitted lane line point cloud data; determining a target fitting equation from each candidate fitting equation, and performing equation parameter calculation on the target fitting equation by using the to-be-fitted lane line point cloud data to obtain fitting equation parameters; when the fitting equation parameters meet a preset fitting completion condition, determining a target lane line fitting line based on the target fitting equation and the fitting equation parameters; performing equidistance sampling on the target lane line fitting line to obtain each equidistance sampling point, and connecting the equidistance sampling points to obtain a lane line corresponding to the lane line identification. The method can improve the accuracy of lane line fitting.

Description

Technical Field

[0001] This application relates to the field of high-precision mapping, and in particular to a method, apparatus, computer device, storage medium, and computer program product for generating lane lines. Background Technology

[0002] During the operation of autonomous vehicles, high-precision maps are required to obtain traffic information, such as information on roads, traffic signs, lane lines, obstacles, pedestrians, and other traffic elements. This traffic information is then used to control the vehicle's steering, speed, path planning, and lane changes. Therefore, the accuracy of high-precision maps, especially the accuracy of lane lines, directly affects the safe driving of autonomous vehicles. Existing high-precision maps... Figure 1 Typically, this is a crowdsourced semantic map, where lane lines are generated by curve fitting of sampled lane lines. However, existing curve fitting methods sometimes fail to fit a particular lane line, resulting in the inability to generate a fitted curve and consequently, low accuracy in fitting lane lines in high-precision maps. Summary of the Invention

[0003] Therefore, it is necessary to provide a lane line generation method, apparatus, computer equipment, computer-readable storage medium, and computer program product that can improve the fitting accuracy of lane lines, addressing the aforementioned technical problems.

[0004] Firstly, this application provides a lane line generation method. The method includes:

[0005] Obtain the lane line point cloud data and the lane line identifiers corresponding to each lane line point cloud data, and determine the target lane line point cloud data corresponding to the same lane line identifier from each lane point cloud data.

[0006] The target lane line point cloud data is transformed according to the target position transformation information to obtain the lane line point cloud data to be fitted.

[0007] The target fitting equation is determined from the candidate fitting equations, and the equation parameters of the target fitting equation are calculated using the point cloud data of the lane line to be fitted, so as to obtain the fitting equation parameters.

[0008] When the fitting equation parameters meet the preset fitting completion conditions, the target lane line fitting line is determined based on the target fitting equation and the fitting equation parameters.

[0009] The target lane line fitting line is sampled at equal intervals to obtain each equal sampling point. The lane lines corresponding to the lane line labels are then connected.

[0010] Secondly, this application also provides a lane line generation device. The device includes:

[0011] The acquisition module is used to acquire each lane line point cloud data and the lane line identifier corresponding to each lane line point cloud data, and to determine the target lane line point cloud data corresponding to the same lane line identifier from each lane point cloud data.

[0012] The conversion module is used to convert the point cloud data of the target lane line according to the target position conversion information to obtain the point cloud data of the lane line to be fitted.

[0013] The fitting equation determination module is used to determine the target fitting equation from various candidate fitting equations. It uses the point cloud data of the lane line to be fitted to calculate the equation parameters of the target fitting equation and obtain the fitting equation parameters.

[0014] The fitting line generation module is used to determine the target lane line fitting line based on the target fitting equation and the fitting equation parameters when the fitting equation parameters meet the preset fitting completion conditions.

[0015] The lane line generation module is used to perform equidistant sampling on the target lane line fitting line to obtain each equidistant sampling point. The equidistant sampling points are then connected to obtain the lane line corresponding to the lane line identifier.

[0016] 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 perform the following steps:

[0017] Obtain the lane line point cloud data and the lane line identifiers corresponding to each lane line point cloud data, and determine the target lane line point cloud data corresponding to the same lane line identifier from each lane point cloud data.

[0018] The target lane line point cloud data is transformed according to the target position transformation information to obtain the lane line point cloud data to be fitted.

[0019] The target fitting equation is determined from the candidate fitting equations, and the equation parameters of the target fitting equation are calculated using the point cloud data of the lane line to be fitted, so as to obtain the fitting equation parameters.

[0020] When the fitting equation parameters meet the preset fitting completion conditions, the target lane line fitting line is determined based on the target fitting equation and the fitting equation parameters.

[0021] The target lane line fitting line is sampled at equal intervals to obtain each equal sampling point. The lane lines corresponding to the lane line labels are then connected.

[0022] 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, performs the following steps:

[0023] Obtain the lane line point cloud data and the lane line identifiers corresponding to each lane line point cloud data, and determine the target lane line point cloud data corresponding to the same lane line identifier from each lane point cloud data.

[0024] The target lane line point cloud data is transformed according to the target position transformation information to obtain the lane line point cloud data to be fitted.

[0025] The target fitting equation is determined from the candidate fitting equations, and the equation parameters of the target fitting equation are calculated using the point cloud data of the lane line to be fitted, so as to obtain the fitting equation parameters.

[0026] When the fitting equation parameters meet the preset fitting completion conditions, the target lane line fitting line is determined based on the target fitting equation and the fitting equation parameters.

[0027] The target lane line fitting line is sampled at equal intervals to obtain each equal sampling point. The lane lines corresponding to the lane line labels are then connected.

[0028] Fifthly, this application also provides a computer program product. The computer program product includes a computer program that, when executed by a processor, performs the following steps:

[0029] Obtain the lane line point cloud data and the lane line identifiers corresponding to each lane line point cloud data, and determine the target lane line point cloud data corresponding to the same lane line identifier from each lane point cloud data.

[0030] The target lane line point cloud data is transformed according to the target position transformation information to obtain the lane line point cloud data to be fitted.

[0031] The target fitting equation is determined from the candidate fitting equations, and the equation parameters of the target fitting equation are calculated using the point cloud data of the lane line to be fitted, so as to obtain the fitting equation parameters.

[0032] When the fitting equation parameters meet the preset fitting completion conditions, the target lane line fitting line is determined based on the target fitting equation and the fitting equation parameters.

[0033] The target lane line fitting line is sampled at equal intervals to obtain each equal sampling point. The lane lines corresponding to the lane line labels are then connected.

[0034] The aforementioned lane line generation method, apparatus, computer equipment, storage medium, and computer program product obtain lane line point cloud data corresponding to the same lane line identifier from various lane point cloud data, and perform point cloud position transformation on the lane line point cloud data to obtain lane line point cloud data to be fitted for the complete lane lines corresponding to the same lane line identifier. Then, a target fitting equation is determined from various candidate fitting equations, and the fitting equation parameters corresponding to the target fitting equation are calculated using the lane line point cloud data to be fitted, resulting in the target fitting equation and fitting equation parameters corresponding to the lane line point cloud data to be fitted. By determining that the fitting equation parameters meet the preset fitting completion conditions, the target lane line fitting line obtained using the target fitting equation and fitting equation parameters meets the fitting requirements. This makes the target lane fitting line obtained by fitting the lane line point cloud data corresponding to the same lane line identifier more accurate, avoiding the situation where fitting the complete lane lines corresponding to the same lane line identifier fails, thereby improving the accuracy of the fitted lane lines. Attached Figure Description

[0035] Figure 1 This is a diagram illustrating the application environment of a lane line generation method in one embodiment.

[0036] Figure 2 This is a flowchart illustrating a lane line generation method in one embodiment;

[0037] Figure 3 This is a schematic diagram of the filtered point cloud processing flow in one embodiment;

[0038] Figure 4 This is a schematic diagram of the lane line generation process in one embodiment;

[0039] Figure 5 This is a schematic diagram of lane line fitting in one embodiment;

[0040] Figure 6 This is a structural block diagram of a lane line generation device in one embodiment;

[0041] Figure 7 This is an internal structural diagram of a computer device in one embodiment;

[0042] Figure 8 This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation

[0043] 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.

[0044] The lane line generation method provided in this application embodiment can be applied to, for example, Figure 1 In the application environment shown, the vehicle-mounted terminal 102 communicates with the server 104 via a network. A data storage system can store the data that the server 104 needs to process. The data storage system can be integrated onto the server 104 or placed on the cloud or other network servers. The server 104 can obtain lane line point cloud data and lane line identifiers corresponding to each lane line point cloud data through the vehicle-mounted terminal, and determine the target lane line point cloud data corresponding to the same lane line identifier from each lane point cloud data; the server 104 performs point cloud position transformation on the target lane line point cloud data according to the target position transformation information to obtain the lane line point cloud data to be fitted; the server 104 determines the target fitting equation from each candidate fitting equation, and uses the lane line point cloud data to be fitted to calculate the equation parameters of the target fitting equation to obtain the fitting equation parameters; when the terminal 102 detects that the fitting equation parameters meet the preset fitting completion conditions, it determines the target lane line fitting line based on the target fitting equation and the fitting equation parameters; the server 104 performs equidistant sampling on the target lane line fitting line to obtain each equidistant sampling point, and connects each equidistant sampling point to obtain the lane line corresponding to the lane line identifier. Server 104 can also send the lane markings corresponding to the lane markings to the terminal for display. Server 104 can be implemented using a standalone server or a server cluster consisting of multiple servers.

[0045] In one embodiment, such as Figure 2 As shown, a lane line generation method is provided, which can be applied to... Figure 1 Taking a server as an example, it can be understood that this method can also be applied to a terminal, and also to a system that includes both a terminal and a server, and is implemented through the interaction between the terminal and the server. In this embodiment, the method includes the following steps:

[0046] Step 202: Obtain the lane line point cloud data and the lane line identifiers corresponding to each lane line point cloud data, and determine the target lane line point cloud data corresponding to the same lane line identifier from each lane point cloud data.

[0047] Lane line point cloud data refers to the collected point cloud data representing lane lines. Lane line labels refer to the label information corresponding to lane lines; different lane lines have different label information, while the lane line labels are the same for all lane line segments in the point cloud data corresponding to the same lane line. Target lane line point cloud data refers to lane line point cloud data with the same lane line labels.

[0048] Specifically, the vehicle is equipped with a point cloud acquisition device. During operation, the vehicle collects point cloud data for each lane line using this device. This data can be point cloud data for multiple lane segments corresponding to multiple complete lane lines, and generates lane line identifiers for each lane line point cloud. Lane line identifiers are identical for lane line point cloud data. The onboard terminal sends each lane line point cloud data and its corresponding lane line identifier to a server. The server determines the target lane line point cloud data from the various lane line point cloud data based on the identical lane line identifiers. Alternatively, the vehicle terminal can save multiple lane line point cloud data corresponding to the same lane line identifier separately according to the lane line identifier, and send each lane line point cloud data corresponding to each lane line identifier to the server separately. The server then retrieves the target lane line point cloud data based on the lane line identifier.

[0049] Step 204: Perform point cloud position transformation on the target lane line point cloud data according to the target position transformation information to obtain the lane line point cloud data to be fitted.

[0050] The target location transformation information refers to the transformation parameters used to transform the coordinate system of the point cloud data. The lane line point cloud data to be fitted refers to the lane line point cloud data under a unified coordinate system, and lane line fitting is performed using the lane line point cloud data to be fitted.

[0051] Specifically, the vehicle is also equipped with a pose sensor to perceive the vehicle's pose information during driving. The onboard terminal sends the pose information and lane line point cloud data to the server. The server determines the target lane line point cloud data and its corresponding pose information, and then optimizes the pose information using a pre-set pose optimization algorithm. The server obtains pre-set target position transformation information and uses the optimized position and target position transformation information to perform coordinate transformation calculations on the lane line point cloud data to obtain lane line point cloud data to be fitted in a unified coordinate system.

[0052] Step 206: Determine the target fitting equation from the candidate fitting equations, and use the point cloud data of the lane line to be fitted to calculate the equation parameters of the target fitting equation to obtain the fitting equation parameters.

[0053] Here, candidate fitting equations refer to pre-stored fitting equations used for lane line fitting. The target fitting equation is the equation selected from among the candidate fitting equations for lane line fitting; it can be a curve equation, a straight line equation, etc. Fitting equation parameters refer to the equation parameters corresponding to the target fitting equation, calculated using the point cloud data of the lane line to be fitted.

[0054] Specifically, the server pre-stores multiple candidate fitting equations, including equations representing different line shapes, such as curve equations representing different curves and line equations representing different straight lines. The server can randomly select one candidate fitting equation as the target fitting equation, or it can determine the target fitting equation from among the candidate fitting equations according to a preset usage order. The server uses the point cloud data of the lane line to be fitted to calculate the equation parameters of the target fitting equation, thus obtaining the fitting equation parameters corresponding to the target fitting equation.

[0055] Step 208: When the fitting equation parameters meet the preset fitting completion conditions, determine the target lane line fitting line based on the target fitting equation and the fitting equation parameters.

[0056] Among them, the preset fitting completion conditions refer to the judgment conditions for whether the pre-set fitting equation parameters meet the fitting requirements. The target lane line fitting line refers to the fitting line generated by the fitting equation parameters and the target fitting equation.

[0057] Specifically, the server uses the real coordinate information of the lane line point cloud data to be fitted to calculate the fitting coordinate information corresponding to the target fitting equation through the fitting equation parameters and the target fitting equation. The fitting coordinate information represents the coordinate values ​​of the lane line point cloud data to be fitted, calculated using the fitting equation parameters and the target fitting equation. Then, the error between the real coordinate information and the fitting coordinate information of the lane line point cloud data to be fitted is calculated to obtain the coordinate error information corresponding to the target fitting equation. The server obtains a pre-set coordinate error information threshold. When the coordinate error information is detected to be less than the coordinate error information threshold, it is considered that the fitting equation parameters meet the preset fitting completion conditions, and the target lane line fitting line is generated through the target fitting equation and the fitting equation parameters.

[0058] In one embodiment, the server obtains a pre-set number of times the parameters of the target fitting equation can be optimized, calculates the initial fitting equation parameters using the point cloud data of the lane line to be fitted and the target fitting equation, and calculates the coordinate error information using the initial fitting equation parameters. When the coordinate error information exceeds the coordinate error information threshold, the initial fitting equation parameters are optimized within the allowed number of parameter optimizations until the optimized fitting equation parameters meet the preset fitting completion conditions.

[0059] Step 210: Perform equidistant sampling on the target lane line fitting line to obtain each equidistant sampling point, and connect each equidistant sampling point to obtain the lane line corresponding to the lane line identifier.

[0060] Lane lines refer to the lines that represent lanes on a map, and are composed of line segments connected by equidistant sampling points.

[0061] Specifically, the server performs equidistant sampling on the target lane line fitting line to obtain each equidistant sampling point, then connects each equidistant sampling point to obtain the lane line corresponding to the lane line identifier, and then sends the lane line to the terminal for display.

[0062] In the lane line generation method described above, lane line point cloud data corresponding to the same lane line identifier is obtained from various lane point cloud data sets. This is followed by point cloud position transformation to obtain lane line point cloud data for the complete lane lines corresponding to the same lane line identifier. Then, a target fitting equation is determined from various candidate fitting equations. The fitting equation parameters corresponding to the target fitting equation are calculated using the lane line point cloud data for the lane line to be fitted, resulting in the target fitting equation and fitting equation parameters. When the fitting equation parameters meet the preset fitting completion conditions, the target lane line fitting line obtained using the target fitting equation and fitting equation parameters meets the fitting requirements. This ensures that the target lane fitting line obtained by fitting the lane line point cloud data corresponding to the same lane line identifier is more accurate, avoiding the failure to fit the complete lane lines corresponding to the same lane line identifier, thereby improving the accuracy of the fitted lane lines.

[0063] In one embodiment, step 208, when the fitting equation parameters meet the preset fitting completion conditions, determines the target lane line fitting line based on the target fitting equation and the fitting equation parameters, including:

[0064] If the fitting equation parameters do not meet the preset fitting completion conditions, the process returns to the step of determining the target fitting equation from each candidate fitting equation until the fitting equation parameters are detected to meet the preset fitting completion conditions. Then, the target lane line fitting line is determined based on the target fitting equation and the fitting equation parameters.

[0065] Specifically, when the server detects that the coordinate error information calculated based on the fitting equation parameters consistently exceeds the coordinate error information threshold within the parameter optimization iterations, it considers that the fitting equation parameters do not meet the preset fitting completion conditions. That is, the currently used candidate fitting equation does not meet the fitting requirements. The server then returns to the step of determining the target fitting equation from among the candidate fitting equations. The server can either exclude the currently used fitting equation from the candidate fitting equations and randomly determine a new target fitting equation from the remaining candidate fitting equations. Alternatively, the server can redetermine a new target fitting equation from among the candidate fitting equations according to the pre-set order of their use, calculate the fitting equation parameters corresponding to the target fitting equation using the point cloud data of the lane line to be fitted, and determine whether the fitting equation parameters meet the preset fitting completion conditions. If they do, the new target fitting equation and its corresponding fitting equation parameters are used to determine the target lane line fitting line, and a lane line corresponding to the lane line identifier is generated based on the target lane line fitting line. If the conditions are not met, the target fitting equation is reconfirmed among the candidate fitting equations until the fitting equation parameters meet the preset fitting completion conditions. The target lane line fitting line is then determined based on the target fitting equation and its corresponding fitting equation parameters, and a lane line corresponding to the lane line identifier is generated based on the target lane line fitting line.

[0066] In one specific embodiment, the server has a pre-set library of fitting equations, storing multiple candidate fitting equations, such as univariate polynomial equations, spline interpolation, exponential functions, power functions, etc. The univariate polynomial equation could be: y = k0 + k1 * x 1 +k2*x 2 +k3*x 3 +……, where k0, k1, etc. represent the parameters of the fitting equation, x, etc. represent the x-coordinates of the 3D points in the point cloud data, and y represents the y-coordinates calculated using the x-coordinates of the 3D points in the point cloud data. Multiple candidate fitting equations can be pre-sorted randomly or sorted by curve type, such as linear equations, simple curve equations, and complex curve equations. The server can retrieve the target fitting equation from the fitting equation library according to the order in which the fitting equations are used.

[0067] In one embodiment, such as Figure 3 As shown, a schematic diagram of a filtered point cloud processing workflow is provided; the method further includes:

[0068] Step 302: When the fitting equation parameters corresponding to each candidate fitting equation do not meet the preset fitting completion conditions, downsampling and filtering are performed on the lane line point cloud data to be fitted to obtain filtered lane line point cloud data.

[0069] Step 304: Determine the target fitting equation from each candidate fitting equation, and use the filtered lane line point cloud data to calculate the equation parameters of the target fitting equation to obtain the parameters of the filtered fitting equation.

[0070] Step 306: When the parameters of the filtered fitting equation meet the preset fitting completion conditions, determine the filtered lane line fitting line based on the target fitting equation and the parameters of the filtered fitting equation.

[0071] Step 308: Perform equidistant sampling on the fitted lines of the filtered lane lines to obtain each filtered equidistant sampling point, and connect the filtered equidistant sampling points to obtain the lane lines corresponding to the lane line identifiers.

[0072] Downsampling filtering refers to filtering out clusters of 3D points in the lane line point cloud data to be fitted. Filtered lane line point cloud data refers to the point cloud data after downsampling filtering. Filtered fitting equation parameters refer to the fitting equation parameters calculated using the filtered lane line point cloud data. Filtered lane line fitting line refers to the fitting line generated based on the target fitting equation and the filtered lane line point cloud data.

[0073] Specifically, when the server detects that the parameters of the fitting equations corresponding to all candidate fitting equations in the fitting equation library do not meet the preset fitting completion conditions, it performs downsampling filtering on the lane line point cloud data to be fitted. That is, it filters out point cloud data that exhibit clustering in the lane line point cloud data to be fitted, retaining one or a few point cloud data, resulting in filtered lane line point cloud data. Then, the server uses the filtered lane line point cloud as the lane line point cloud data to be fitted and returns to the step of determining the target fitting equation from each candidate fitting equation. It then uses the filtered lane line point cloud data to calculate the equation parameters of the target fitting equation, obtaining the filtered fitting equation parameters.

[0074] The server determines whether the parameters of the filtered fitting equations meet the preset fitting completion conditions. If they do, the server generates a filtered lane line fitting line based on the target fitting equation and the parameters of the filtered fitting equations. The server then performs equidistant sampling on the filtered lane line fitting line to obtain each filtered equidistant sampling point. The server can then optimize the sampling points using methods such as Bezier interpolation and nonlinear constraint optimization to obtain optimized sampling points. These optimized sampling points are then connected to obtain the lane line corresponding to the smooth lane line identifier. If the conditions are not met, it is considered that all candidate fitting equations do not meet the fitting requirements of the filtered lane line point cloud data, and a straight line is directly used as the lane line fitting line for subsequent adjustments via the management terminal.

[0075] In this embodiment, when it is detected that none of the candidate fitting equations meet the fitting requirements of the lane line point cloud data to be fitted, the lane line point cloud data to be fitted is downsampled and filtered to obtain filtered lane line point cloud data, which can improve the fitting success rate of the filtered lane line point cloud data.

[0076] In one embodiment, step 208, when the fitting equation parameters meet the preset fitting completion conditions, determines the target lane line fitting line based on the target fitting equation and the fitting equation parameters, including:

[0077] The fitting error is calculated based on the target fitting equation and the fitting equation parameters to obtain the fitting error information, and the number of fitting parameter iterations corresponding to the fitting equation parameters is obtained.

[0078] When the number of iterations of the fitting parameters does not exceed the preset threshold for the number of iterations of the fitting parameters, and the fitting error result does not exceed the preset error threshold, it is determined that the target fitting equation parameters meet the preset fitting completion conditions, and the target lane line fitting line is determined based on the target fitting equation and the fitting equation parameters.

[0079] The fitting error information refers to the error between the lane line point cloud coordinates calculated using the target fitting equation and the fitting equation parameters and the actual lane line point cloud coordinates. The fitting parameter iteration count refers to the number of times the fitting equation parameters are calculated iteratively using the target fitting equation to calculate the fitting equation parameters using the coordinates of each 3D point in the lane line point cloud data to be fitted. The fitting parameter iteration count threshold is a pre-set number of iterations allowed to calculate the fitting equation parameters using the target fitting equation. It serves as a stopping condition for calculating the fitting equation parameters, indicating that calculation stops when the number of fitting parameter iterations reaches the fitting parameter iteration count threshold and the fitting error information corresponding to the fitting equation parameters still exceeds the preset error threshold.

[0080] Specifically, the server replaces the unknown parameters in the target fitting equation with pre-set initial values ​​for the fitting equation parameters to obtain a calculable initial target fitting equation. Then, the server obtains the x-coordinates of each 3D point in the lane line point cloud data to be fitted, inputs these x-coordinates into the initial target fitting equation for calculation, obtains the calculated y-coordinates of the 3D points, and calculates the difference between the calculated and actual y-coordinates to obtain the initial coordinate error information for each 3D point. Based on this initial coordinate error information, the initial fitting equation parameters in the initial target fitting equation are iteratively updated within a fitting parameter iteration threshold, and the fitting error information corresponding to each updated fitting equation parameter is calculated. The fitting error information corresponding to each updated fitting equation parameter is obtained by calculating the coordinate error information of the y-coordinates of some 3D points. When the calculated fitting error information consistently exceeds a preset error threshold, it is determined that the target fitting equation parameters do not meet the preset fitting completion conditions. The target fitting equation is then re-determined from the candidate fitting equations and recalculated. When the calculated fitting error information is detected to be less than the preset error threshold, the calculation stops. The fitting equation parameters updated at this time are used as the fitting equation parameters corresponding to the target fitting equation. It is determined that the target fitting equation parameters meet the preset fitting completion conditions. Based on the target fitting equation and the fitting equation parameters, the target lane line fitting line corresponding to the lane line point cloud data to be fitted is generated.

[0081] In one specific embodiment, when the server calculates the fitting equation parameters corresponding to the target fitting equation using the lane line point cloud data to be fitted, it considers the target fitting equation to have converged successfully. Then, it determines whether the target fitting equation and the fitting equation parameters meet the preset fitting completion condition, that is, whether the fitting error information corresponding to the fitting equation parameters exceeds the preset error threshold. When the fitting error parameter exceeds the preset error threshold, the target fitting equation and the fitting equation parameters do not meet the preset fitting completion condition. If the fitting equation parameters corresponding to the target fitting equation calculated by the server using the lane line point cloud data to be fitted oscillate and cannot approach a constant value, it is considered that the target fitting equation has failed to converge, and it is determined that the target fitting equation does not meet the preset fitting completion condition.

[0082] In this embodiment, by using the number of iterations of the fitting parameters and the fitting error information, it is possible to accurately determine whether the target fitting equation and the fitting equation parameters meet the fitting requirements, thereby making the target lane line fitted using the target fitting equation and fitting equation parameters that meet the fitting requirements more accurate.

[0083] In one embodiment, before step 202, which involves obtaining the lane line point cloud data and the lane line identifiers corresponding to each lane line point cloud data, the method further includes:

[0084] Acquire point cloud data for each lane line corresponding to the target time period;

[0085] Based on the point cloud data of each lane line, identify the same type of lane line to obtain the same type of lane line point cloud data in each lane line point cloud data.

[0086] Generate lane line identifiers corresponding to the same type of lane line point cloud data.

[0087] The target time period refers to the time period during which lane line point cloud data is collected. Lane line point cloud data of the same type refers to lane line point cloud data belonging to the same complete lane line, including lane line point cloud data corresponding to multiple lane line segments of the same lane line.

[0088] Specifically, the server acquires at least two lane line point cloud data sets corresponding to the target time period sent by the vehicle terminal. Each lane line point cloud data set can be multiple frames of point cloud data collected by the point cloud acquisition device within the target time period. Each frame of point cloud data can include at least one lane line point cloud data set, representing the point cloud data corresponding to a portion of a complete lane line. Based on the timestamps of the multiple frames of lane line point cloud data, the server uses a pre-trained deep learning network to identify similar lane line point cloud data sets between consecutive frames, obtaining similar lane line point cloud data sets within each set. Then, the server generates identical lane line identifiers for the similar lane line point cloud data sets, obtaining the lane line identifiers corresponding to each set of lane line point cloud data sets.

[0089] In one specific embodiment, the vehicle terminal controls the vehicle to start moving and sends a point cloud data acquisition command to the point cloud acquisition device. The point cloud acquisition device responds to the command and acquires point cloud data of the vehicle's surrounding environment at a preset acquisition frequency, obtaining multiple frames of point cloud data and sending them to the vehicle terminal. The vehicle terminal can use a pre-trained deep learning network, such as a lane line tracking algorithm, to identify lane lines in the point cloud data. During the acquisition process, the vehicle terminal identifies similar lane line point cloud data in consecutive frames, generating identical lane line identifiers for each type of lane line point cloud data. The vehicle terminal then sends each lane line point cloud data and its corresponding lane line identifier to the server.

[0090] In this embodiment, by identifying similar lane line point cloud data and generating the same lane line identifier for similar lane line point cloud data, it is possible to quickly obtain similar lane line point cloud data through the lane line identifier, thereby avoiding omissions in lane line point cloud data and ensuring the integrity of lane line fitting.

[0091] In one embodiment, step 204, which involves transforming the target lane line point cloud data according to the target position transformation information to obtain the lane line point cloud data to be fitted, includes:

[0092] Based on the target position transformation information, the point cloud positions of each target lane line point cloud data of the same type in the target lane line point cloud data are transformed to obtain the lane line point cloud data of the same type to be fitted.

[0093] The lane line point cloud data to be fitted is obtained by stitching together the lane line point cloud data based on the coordinate information corresponding to the lane line point cloud data of the same type to be fitted.

[0094] Among them, target location transformation information refers to the transformation parameters for coordinate system transformation. Target similar lane line point cloud data refers to lane line point cloud data belonging to the same lane line within the target lane line point cloud data. Similar lane line point cloud data to be fitted refers to similar lane line point cloud data after coordinate system transformation.

[0095] Specifically, the server acquires vehicle pose information and lane line point cloud data for the target time period sent by the in-vehicle terminal. The server optimizes the pose information using filtering algorithms, such as round-trip Kalman filtering, to obtain optimized pose information. Since the target lane line point cloud data is in the vehicle coordinate system at the time of acquisition, and the origin of the vehicle coordinate system moves during vehicle movement, a unified coordinate system transformation is required. The server acquires pre-set target position transformation information, i.e., the extrinsic parameters of the LiDAR or camera, and uses the extrinsic parameters and optimized pose information to transform the lane line point cloud data of each target from their respective vehicle coordinate system to the world coordinate system. At this point, the 3D points of each target's lane line point cloud data are in the same coordinate system, and the lane line labels corresponding to each target's lane line remain unchanged, resulting in the lane line point cloud data to be fitted.

[0096] Then, based on the 3D point coordinates of the lane line point cloud data to be fitted, the server stitches together the individual lane line point cloud data from each lane line point cloud data to obtain the complete lane line point cloud data to be fitted. Alternatively, the server can determine the first segment of lane line point cloud data from the lane line point cloud data to be fitted based on the point cloud coordinates, and transform the other lane line point cloud data to the coordinate system of the first segment to obtain the lane line point cloud data to be fitted.

[0097] In this embodiment, the point cloud data of the target lane line is transformed to obtain the point cloud data of the same type of lane line to be fitted. Then, the point cloud data of each lane line is stitched together according to the point cloud coordinates to obtain the point cloud data of the lane line to be fitted corresponding to the complete lane line, thereby ensuring the integrity of the lane line fitting.

[0098] In one specific embodiment, such as Figure 4 The diagram illustrates a process for lane line generation. The vehicle-mounted terminal controls the vehicle to begin driving. During this process, point cloud data of the surrounding environment is collected via a point cloud acquisition device, and the vehicle's pose information is collected via a pose sensor. The vehicle-mounted terminal uses a trained deep learning network to identify each frame of lane line point cloud data. Based on the lane line point cloud data from consecutive frames, lane line point cloud data belonging to the same lane line are identified. The same lane line identifier is generated for lane line point cloud data belonging to the same lane line, thus each frame of collected lane line point cloud data corresponds to one lane line identifier. The vehicle-mounted terminal sends the pose information, lane line point cloud data, and their corresponding lane line identifiers to the server for lane line generation.

[0099] The server uses a round-trip Kalman filter algorithm to optimize the pose information, obtaining the optimized vehicle pose. Then, based on lane line identifiers, the server saves the lane point cloud data corresponding to the same lane line identifier, using this data as the target lane point cloud data. Next, based on the optimized vehicle pose and sensor extrinsic parameters (e.g., LiDAR or camera), the server performs coordinate transformation on the target lane point cloud data and stitches the data together to obtain the lane point cloud data to be fitted. Alternatively, the server can first transform all lane point cloud data in each frame to the world coordinate system based on the optimized vehicle pose and sensor extrinsic parameters, and save the corresponding lane line identifiers. Then, based on the lane line identifiers, the server uses the lane point cloud data corresponding to the same lane line identifier as the lane point cloud data to be fitted.

[0100] The server determines the target fitting equation from the candidate fitting equations in the fitting equation library. It then calculates the target fitting equation using the lane line point cloud data to be fitted, obtaining the fitting equation parameters, fitting error information, and the number of fitting parameter iterations corresponding to the target fitting equation. At this point, the target fitting equation has successfully converged. Alternatively, the server can input the lane line point cloud data to be fitted into a nonlinear optimization library, such as Ceres (a library for nonlinear optimization), and send the target fitting equation to the library. This allows the nonlinear fitting equation to calculate based on the target fitting equation, obtaining the fitting equation parameters, fitting error information, and the number of fitting parameter iterations corresponding to the target fitting equation output by the nonlinear optimization library. When it is detected that the number of fitting parameter iterations does not exceed the fitting parameter iteration threshold and the fitting error information does not exceed the error threshold, the target fitting equation and fitting equation parameters are determined to meet the fitting completion conditions. The target lane line fitting line is then generated based on the target fitting equation and fitting equation parameters.

[0101] The target lane line fitting line is sampled at equal intervals to obtain each equal sampling point, and the equal sampling points are connected to form the lane line corresponding to the lane line identifier, which can be the lane line in the crowdsourced semantic map.

[0102] When the detected fitting error information consistently exceeds the error threshold, it is determined that the target fitting equation and the fitting equation parameters do not meet the fitting completion conditions. The server executes the instruction to adjust the fitting equation and returns to the step of determining the target fitting equation from each candidate fitting equation. The fitting continues using the lane line point cloud data to be fitted until the target fitting equation is detected to have converged successfully. The lane line corresponding to the lane line identifier is generated, and a lane line semantic map is generated based on the lane line.

[0103] In one specific embodiment, such as Figure 5 The diagram shown illustrates a lane line fitting method. Figure 5 -A indicates the generated semantic map of a certain lane line segment; Figure 5 -B indicates a lane line semantic map that failed to fit a certain lane line segment; Figure 5 -C indicates a lane line semantic map that has been successfully fitted for a certain lane line segment, meaning... Figure 5 -B is the fitting result of the lane line segment after adjustment by the fitting equation.

[0104] 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.

[0105] Based on the same inventive concept, this application also provides a lane line generation apparatus for implementing the lane line generation method described above. The solution provided by this apparatus is similar to the implementation described in the above method; therefore, the specific limitations of one or more lane line generation apparatus embodiments provided below can be found in the limitations of the lane line generation method described above, and will not be repeated here.

[0106] In one embodiment, such as Figure 6As shown, a lane line generation device 600 is provided, including: an acquisition module 602, a conversion module 604, a fitting equation determination module 606, a fitting line generation module 608, and a lane line generation module 610, wherein:

[0107] The acquisition module 602 is used to acquire each lane line point cloud data and the lane line identifier corresponding to each lane line point cloud data, and to determine the target lane line point cloud data corresponding to the same lane line identifier from each lane point cloud data.

[0108] The conversion module 604 is used to convert the point cloud data of the target lane line according to the target position conversion information to obtain the point cloud data of the lane line to be fitted.

[0109] The fitting equation determination module 606 is used to determine the target fitting equation from each candidate fitting equation, and to calculate the equation parameters of the target fitting equation using the point cloud data of the lane line to be fitted, so as to obtain the fitting equation parameters.

[0110] The fitting line generation module 608 is used to determine the target lane line fitting line based on the target fitting equation and the fitting equation parameters when the fitting equation parameters meet the preset fitting completion conditions.

[0111] The lane line generation module 610 is used to perform equidistant sampling on the target lane line fitting line to obtain each equidistant sampling point, and connect each equidistant sampling point to obtain the lane line corresponding to the lane line identifier.

[0112] In one embodiment, the fitted line generation module 608 includes:

[0113] The detection unit is used to return to the step of determining the target fitting equation from each candidate fitting equation when the fitting equation parameters do not meet the preset fitting completion conditions, until the fitting equation parameters are detected to meet the preset fitting completion conditions, and then determine the target lane line fitting line based on the target fitting equation and the fitting equation parameters.

[0114] In one embodiment, the lane line generating device 600 further includes:

[0115] The filtering unit is used to downsample and filter the lane line point cloud data to be fitted when the fitting equation parameters corresponding to each candidate fitting equation do not meet the preset fitting completion conditions, thereby obtaining filtered lane line point cloud data; determine the target fitting equation from each candidate fitting equation, and use the filtered lane line point cloud data to calculate the equation parameters of the target fitting equation, thereby obtaining the filtered fitting equation parameters; when the filtered fitting equation parameters meet the preset fitting completion conditions, determine the filtered lane line fitting line based on the target fitting equation and the filtered fitting equation parameters; perform equidistant sampling on the filtered lane line fitting line to obtain each filtered equidistant sampling point, and connect each filtered equidistant sampling point to obtain the lane line corresponding to the lane line identifier.

[0116] In one embodiment, the fitted line generation module 608 includes:

[0117] The fitting completion unit is used to calculate the fitting error based on the target fitting equation and the fitting equation parameters, obtain the fitting error information, and obtain the fitting parameter iteration number corresponding to the fitting equation parameters. When the fitting parameter iteration number does not exceed the preset fitting parameter iteration number threshold and the fitting error result does not exceed the preset error threshold, it is determined that the target fitting equation parameters meet the preset fitting completion condition, and the target lane line fitting line is determined based on the target fitting equation and the fitting equation parameters.

[0118] In one embodiment, the lane line generating device 600 further includes:

[0119] The lane line recognition unit is used to acquire point cloud data of each lane line corresponding to the target time period; to perform lane line recognition of the same type based on each lane line point cloud data, to obtain lane line point cloud data of the same type in each lane line point cloud data; and to generate lane line labels corresponding to the lane line point cloud data of the same type.

[0120] In one embodiment, the conversion module 604 includes:

[0121] The stitching unit is used to perform point cloud position transformation on each target lane line point cloud data of the same type according to the target position transformation information to obtain the lane line point cloud data of the same type to be fitted; and to stitch the lane line point clouds based on the coordinate information corresponding to the lane line point cloud data of the same type to be fitted to obtain the lane line point cloud data to be fitted.

[0122] Each module in the aforementioned lane line 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.

[0123] In one embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as follows: Figure 7 As shown, the computer device includes a processor, memory, input / output (I / O) interfaces, and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides the environment for the operating system and computer programs stored in the non-volatile storage media. The database stores lane line point cloud data. The I / O interfaces are used for exchanging information between the processor and external devices. The communication interface is used for communicating with external terminals via a network. When executed by the processor, the computer program implements a lane line generation method.

[0124] In one embodiment, a computer device is provided, which may be a terminal, and its internal structure diagram may be as follows: Figure 8 As shown, the computer device includes a processor, memory, input / output interface, communication interface, display unit, and input device. The processor, memory, and input / output interface are connected via a system bus, and the communication interface, display unit, and input device are also connected to the system bus via the input / output interface. The processor provides computing and control capabilities. The memory includes a non-volatile storage medium and internal memory. The non-volatile storage medium 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 medium. The input / output interface is used for exchanging information between the processor and external devices. 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 executed by the processor, the computer program implements a lane line generation method. The display unit of the computer device is used to form a visually visible image. It can be a display screen, a projection device, or a virtual reality imaging device. The display screen can be an LCD screen or an e-ink screen. The input device of the computer device can be a touch layer covering the display screen, or buttons, trackballs, or touchpads set on the casing of the computer device, or external keyboards, touchpads, or mice, etc.

[0125] Those skilled in the art will understand that Figure 7-8The 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.

[0126] In one embodiment, a computer device is also 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 method embodiments.

[0127] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon that, when executed by a processor, implements the steps in the above method embodiments.

[0128] 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.

[0129] 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, and the collection, use and processing of the relevant data shall comply with the relevant laws, regulations and standards of the relevant countries and regions.

[0130] 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.

[0131] 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.

[0132] 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 method for generating lane lines, characterized in that, The method includes: Obtain each lane line point cloud data and the lane line identifier corresponding to each lane line point cloud data, and determine the target lane line point cloud data corresponding to the same lane line identifier from each lane line point cloud data; The target lane line point cloud data is transformed according to the target position transformation information to obtain the lane line point cloud data to be fitted. The target fitting equation is determined from the candidate fitting equations, and the equation parameters of the target fitting equation are calculated using the point cloud data of the lane line to be fitted, so as to obtain the fitting equation parameters. When the fitting equation parameters do not meet the preset fitting completion condition, the process returns to the step of determining the target fitting equation from each candidate fitting equation until the fitting equation parameters are detected to meet the preset fitting completion condition. Then, the target lane line fitting line is determined based on the target fitting equation and the fitting equation parameters. The target lane line fitting line is sampled at equal intervals to obtain each equal sampling point. The equal sampling points are then connected to obtain the lane line corresponding to the lane line identifier. When the fitting equation parameters corresponding to each candidate fitting equation do not meet the preset fitting completion condition, the lane line point cloud data to be fitted is downsampled and filtered to obtain filtered lane line point cloud data. The target fitting equation is determined from the candidate fitting equations, and the filtered lane line point cloud data is used to calculate the equation parameters of the target fitting equation to obtain the filtered fitting equation parameters. When the parameters of the filtered fitting equation meet the preset fitting completion condition, the filtered lane line fitting line is determined based on the target fitting equation and the parameters of the filtered fitting equation. The filtered lane line fitting line is sampled at equal intervals to obtain each filtered equal-interval sampling point. The filtered equal-interval sampling points are then connected to obtain the lane line corresponding to the lane line identifier.

2. The method according to claim 1, characterized in that, The step of determining the target lane line fitting line based on the target fitting equation and the fitting equation parameters includes: Using the real coordinate information of the lane line point cloud data to be fitted, the fitting coordinate information corresponding to the target fitting equation is calculated through the fitting equation parameters and the target fitting equation; the fitting coordinate information represents the coordinate value of the lane line point cloud data to be fitted obtained by the fitting equation parameters and the target fitting equation. Calculate the error between the true coordinate information and the fitted coordinate information of the lane line point cloud data to be fitted, and obtain the coordinate error information corresponding to the target fitting equation; When the coordinate error information does not exceed the coordinate error information threshold, it is determined that the fitting equation parameters meet the preset fitting completion conditions, and the target lane line fitting line is generated through the target fitting equation and the fitting equation parameters.

3. The method according to claim 1, characterized in that, When the fitting equation parameters meet the preset fitting completion conditions, determining the target lane line fitting line based on the target fitting equation and the fitting equation parameters includes: The fitting error is calculated based on the target fitting equation and the fitting equation parameters to obtain fitting error information, and the number of fitting parameter iterations corresponding to the fitting equation parameters is obtained. When the number of iterations of the fitting parameters does not exceed the preset threshold for the number of iterations of the fitting parameters, and the fitting error result does not exceed the preset error threshold, it is determined that the target fitting equation parameters meet the preset fitting completion condition, and the target lane line fitting line is determined based on the target fitting equation and the fitting equation parameters.

4. The method according to claim 1, characterized in that, Before acquiring the point cloud data of each lane line and the lane line identifiers corresponding to the point cloud data of each lane line, the method further includes: Acquire point cloud data for each lane line corresponding to the target time period; Based on the point cloud data of each lane line, similar lane lines are identified to obtain similar lane line point cloud data in the point cloud data of each lane line. Generate lane line identifiers corresponding to the same type of lane line point cloud data.

5. The method according to claim 1, characterized in that, The step of performing point cloud position transformation on the target lane line point cloud data according to the target position transformation information to obtain the lane line point cloud data to be fitted includes: According to the target position conversion information, the point cloud position of each target lane line point cloud data of the same type in the target lane line point cloud data is converted to obtain the same type lane line point cloud data to be fitted. Based on the coordinate information corresponding to the lane line point cloud data to be fitted, the lane line point cloud is stitched together to obtain the lane line point cloud data to be fitted.

6. A lane line generating device, characterized in that, The device includes: The acquisition module is used to acquire each lane line point cloud data and the lane line identifiers corresponding to the lane line point cloud data respectively, and to determine the target lane line point cloud data corresponding to the same lane line identifier from each lane line point cloud data. The conversion module is used to convert the point cloud data of the target lane line according to the target position conversion information to obtain the point cloud data of the lane line to be fitted. The fitting equation determination module is used to determine the target fitting equation from each candidate fitting equation, and to calculate the equation parameters of the target fitting equation using the point cloud data of the lane line to be fitted, so as to obtain the fitting equation parameters. The fitting line generation module is used to return to the step of determining the target fitting equation from each candidate fitting equation when the fitting equation parameters do not meet the preset fitting completion conditions, until the fitting equation parameters are detected to meet the preset fitting completion conditions, and then determine the target lane line fitting line based on the target fitting equation and the fitting equation parameters. The lane line generation module is used to perform equidistant sampling on the target lane line fitting line to obtain each equidistant sampling point, and connect each equidistant sampling point to obtain the lane line corresponding to the lane line identifier. The lane line generation module is further configured to: downsample and filter the lane line point cloud data to be fitted to obtain filtered lane line point cloud data when the fitting equation parameters corresponding to each candidate fitting equation do not meet the preset fitting completion condition; determine a target fitting equation from each candidate fitting equation; calculate the equation parameters of the target fitting equation using the filtered lane line point cloud data to obtain filtered fitting equation parameters; when the filtered fitting equation parameters meet the preset fitting completion condition, determine a filtered lane line fitting line based on the target fitting equation and the filtered fitting equation parameters; perform equidistant sampling on the filtered lane line fitting line to obtain each filtered equidistant sampling point; and connect the filtered equidistant sampling points to obtain the lane line corresponding to the lane line identifier.

7. 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.

8. 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.

9. 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