Map vector fragment generation method and device, equipment and medium

Abstract

The invention relates to the field of map generation, in particular to a map vector fragment generation method and device, equipment and a medium. According to the method, firstly, the historical fusion fragments and the sensing fragments corresponding to the sampling points are fused, so that the accuracy of map generation is ensured. And secondly, the distances between the sampling points are obtained in sequence, and the key second sampling point is determined according to the threshold value, so that the important sampling points can be processed in a targeted manner, the indifference processing on all the sampling points is avoided, and the data volume needing to be processed is reduced. Judging whether a preset condition is met or not when the vehicle passes through the second sampling point, and performing fitting processing and outputting a target vector fragment only when the condition is met; therefore, according to the method and the device, the perception fragment obtained by each sampling point in the driving process of the vehicle is fused with the previously obtained historical fusion fragment, so that manual intervention can be reduced. The problem that an existing high-precision map making process consumes a large amount of time and energy is solved.

Description

Technical Field

[0001] The present invention relates to the field of map generation, and in particular to a method, device, equipment and medium for generating map vector fragments. Background Art

[0002] High-precision maps play a key role in the field of autonomous driving. They are an important part of the autonomous driving system and can provide centimeter-level road information for self-positioning, path planning and other downstream tasks. Compared with traditional navigation maps, they are more accurate and have higher data dimensions. They not only cover road data such as lane line position, type, width, slope and curvature, but also include traffic sign information such as speed limit signs and traffic lights.

[0003] High-precision map vector fragments are the cornerstone of building crowdsourced high-precision maps. The traditional method of building high-precision map vector fragments relies on high-precision map data collection vehicles to collect road data, and manually annotates the collected lidar point clouds to achieve high-precision map production, which results in a large consumption of time and energy costs. Summary of the invention

[0004] In view of this, an embodiment of the present invention provides a method, device, equipment and medium for generating map vector fragments to solve the problem that the existing high-precision map production process consumes a lot of time and energy costs.

[0005] In a first aspect, an embodiment of the present invention provides a method for generating a map vector fragment, the method comprising:

[0006] Acquire a first perception segment obtained by processing perception information of a current road section when the vehicle passes through a first sampling point, wherein the first perception segment includes a plurality of first continuous targets;

[0007] Acquire a historical fusion segment, and fuse the first continuous target in the first perception segment with the historical continuous target in the historical fusion segment to obtain a first fusion segment, wherein if the previous sampling point of the first sampling point is the starting sampling point of the vector segment, then the historical fusion segment acquired at the first sampling point is the perception segment obtained by the vehicle at the previous sampling point;

[0008] Sequentially acquiring first accumulated distances between sampling points after the first sampling point and the first sampling point until a second sampling point whose distance is greater than or equal to a first threshold is obtained;

[0009] A second perception segment is obtained by processing the perception information of the current road section when the vehicle passes through the second sampling point, the second perception segment is fused with the first fusion segment to obtain a second fusion segment, and it is determined whether a preset condition is satisfied when the vehicle passes through the second sampling point. If the preset condition is satisfied, the second fusion segment is fitted to obtain a target vector segment.

[0010] Furthermore, the obtaining of the first perception segment obtained by processing the perception information of the current road section when the vehicle passes the first sampling point includes:

[0011] Acquire forward-looking perception data obtained by processing a road section image when the vehicle passes through the first sampling point, wherein the road section image is an image of a current road section, and the forward-looking perception data includes a plurality of continuous targets;

[0012] Determine a cut-off point of the continuous target by using each of the plurality of continuous targets, and determine a cut-off point of the continuous target based on the cut-off point of the continuous target and the cut-off length of the fragment;

[0013] The corresponding continuous target is intercepted using the interception starting point and the interception end point of each continuous target to obtain a first continuous target, and the first perception segment is constructed using the first continuous target and the first sampling point.

[0014] Further, the first continuous target in the first perception segment is fused with the historical continuous target in the historical fusion segment to obtain a first fusion segment, including:

[0015] Matching the first continuous target in the first perception segment with all the continuous targets in the historical fusion segment to obtain a second continuous target matched with the first continuous target, wherein the second continuous target is the continuous target with the shortest lateral distance to the first continuous target in the historical fusion segment;

[0016] Fusing the first continuous target with the second continuous target to obtain a fused continuous target;

[0017] A first fusion segment is constructed using the fused continuous target, the third continuous target, the fourth continuous target, all sampling points in the historical fusion segment, and the first sampling point, wherein the third continuous target is a continuous target in the historical fusion segment except the second continuous target, and the fourth continuous target is a continuous target in the first perception segment that is not matched to the second continuous target.

[0018] Furthermore, the fusing the first continuous target with the second continuous target to obtain a fused continuous target includes:

[0019] Deleting the end point of the second continuous target;

[0020] The starting point of the second continuous target, the starting point and the ending point of the first continuous target are taken as target points, and fusion is performed based on the target points to obtain the fused continuous target.

[0021] Furthermore, the determining whether a preset condition is satisfied when the vehicle passes through the second sampling point includes:

[0022] Acquire a second accumulated distance when the vehicle passes through a second sampling point, wherein the second accumulated distance is accumulated and calculated based on the end sampling point of the previous vector segment;

[0023] According to whether the second accumulated distance is greater than or equal to a second threshold, it is determined whether a preset condition is satisfied when the vehicle passes through the second sampling point, wherein if the second accumulated distance is greater than or equal to the second threshold, it is determined that the preset condition is satisfied when the vehicle passes through the second sampling point; or, if the second accumulated distance is less than the second threshold, it is determined that the preset condition is not satisfied when the vehicle passes through the second sampling point.

[0024] Further, the second fused segment is subjected to fitting processing to obtain a target vector segment, including:

[0025] For each continuous target in the second fusion segment, a sliding window is created according to a preset window length;

[0026] Sliding the sliding window in sequence, and taking the target points of the continuous targets in the sliding window as control points;

[0027] Performing third-order B-spline fitting using the control points to obtain a fitting curve, wherein the fitting curve includes a plurality of fitting points;

[0028] Calculating the curvature of each position based on the fitting points in the fitting curve;

[0029] Obtaining a target sampling interval corresponding to the curvature, and downsampling the fitting points using the target sampling interval to obtain fitting sampling points, wherein the fitting sampling points are used to represent the final state of the continuous target;

[0030] The continuous target within the second threshold is constructed by using the fitting sampling points, and the target vector segment is constructed based on the continuous target within the second threshold and all the sampling points within the second threshold.

[0031] Further, obtaining the target sampling interval corresponding to the curvature includes:

[0032] comparing the curvature with a preset curvature;

[0033] If the curvature is greater than or equal to the preset curvature, the first sampling interval is used as the target sampling interval; or, if the curvature is less than the preset curvature, the second sampling interval is used as the target sampling interval, wherein the first sampling interval is less than the second sampling interval.

[0034] In a second aspect, an embodiment of the present invention provides a device for generating a map vector segment, the device comprising:

[0035] an acquisition module, configured to acquire a first perception segment obtained by processing the perception information of the current road section when the vehicle passes through the first sampling point, wherein the first perception segment includes a plurality of first continuous targets;

[0036] a fusion module, configured to obtain a historical fusion segment, and fuse the first continuous target in the first perception segment with the historical continuous target in the historical fusion segment to obtain a first fusion segment, wherein if the previous sampling point of the first sampling point is the starting sampling point of the vector segment, the historical fusion segment obtained at the first sampling point is the perception segment obtained by the vehicle at the previous sampling point;

[0037] A traversal module, used for sequentially acquiring a first accumulated distance between the first sampling point and the sampling points after the first sampling point, until a second sampling point whose distance is greater than or equal to a first threshold is obtained;

[0038] a processing module, used to obtain a second perception segment obtained by processing the perception information of the current road section when the vehicle passes through the second sampling point, fuse the second perception segment with the first fusion segment to obtain a second fusion segment, and determine whether a preset condition is met when the vehicle passes through the second sampling point; if the preset condition is met, the second fusion segment is fitted to obtain a target vector segment.

[0039] In a third aspect, an embodiment of the present invention provides a computer device, comprising: a memory and a processor, the memory and the processor being communicatively connected to each other, the memory storing computer instructions, and the processor executing the method of the first aspect or any corresponding embodiment thereof by executing the computer instructions.

[0040] In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium having computer instructions stored thereon, the computer instructions being used to enable a computer to execute the method of the first aspect or any corresponding embodiment thereof.

[0041] This application first fuses the historical fusion fragments and the perception fragments corresponding to the sampling points to ensure the accuracy of the generated map. Secondly, by obtaining the distances between the sampling points in sequence and determining the key second sampling points according to the threshold, it is possible to process important sampling points in a targeted manner, avoid indiscriminate processing of all sampling points, and reduce the amount of data that needs to be processed. Then when the vehicle passes the second sampling point, it is determined whether the preset conditions are met, and only when the conditions are met does the fitting process occur and the target vector fragment is output. Therefore, this application can reduce manual intervention by fusing the perception fragments obtained at each sampling point during the vehicle's driving process with the previously obtained historical fusion fragments. It solves the problem that the existing high-precision map production process consumes a lot of time and energy costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0043] Figure 1 is a flowchart of a method for generating a map vector fragment according to some embodiments of the present invention;

[0044] Figure 2 is a flowchart of a method for generating a map vector fragment according to some embodiments of the present invention;

[0045] Figure 3 is a schematic diagram of a continuous target detection range according to some embodiments of the present invention;

[0046] Figure 4 is a schematic diagram of the effect of continuous target matching according to some embodiments of the present invention;

[0047] Figure 5 is a schematic diagram of the effect of continuous target fusion according to some embodiments of the present invention

[0048] Figure 6 is a schematic diagram of the position relationship of sampling points according to some embodiments of the present invention;

[0049] Figure 7 is a schematic diagram of the effect of continuous target fitting according to some embodiments of the present invention;

[0050] Figure 8 is a structural block diagram of a device for generating a map vector segment according to an embodiment of the present invention;

[0051] Fig. 9It is a schematic diagram of the hardware structure of a computer device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0052] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0053] According to an embodiment of the present invention, a method, apparatus, device and medium for generating a map vector fragment are provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0054] In this embodiment, a method for generating a map vector segment is provided. Figure 1 is a flow chart of a method for generating a map vector fragment according to an embodiment of the present invention. Figure 1 As shown, the process includes the following steps:

[0055] Step S101, obtaining a first perception segment obtained by processing perception information of a current road section when a vehicle passes a first sampling point, wherein the first perception segment includes a plurality of first continuous targets.

[0056] In an embodiment of the present application, the vehicle is equipped with a device capable of accurately recording time and a camera for collecting images of road sections. When the vehicle passes the first sampling point at time T1, the camera is triggered to start collecting images of the current road section. The camera continuously captures images at a certain frame rate (for example, 30 frames per second) for a period of time (such as 2 seconds) to obtain sufficient forward-looking perception data for the current road section. Use image recognition and processing algorithms, such as edge detection, to detect and extract first continuous targets such as roadside edges, solid lane lines, and dashed lane lines in the image. Mark and classify the detected first continuous targets. Combine the processed image data containing multiple first continuous targets into a first perception segment.

[0057] It should be noted that the first sampling point is a sampling point specifically for the real-time vehicle posture output by the vehicle-mounted INS system. By providing the vehicle position and posture, the relative perception information relative to the vehicle coordinate system at the current moment can be associated with the absolute posture of the world system to construct a vector fragment in the world system. At the same time, the use of sampling points ensures the real-time performance of the calculation while ensuring the quality of the vector fragment.

[0058] Step S102, obtaining a historical fusion segment, and fusing the first continuous target in the first perception segment with the historical continuous target in the historical fusion segment to obtain a first fusion segment, wherein, if the previous sampling point of the first sampling point is the starting sampling point of the vector segment, then the historical fusion segment obtained at the first sampling point is the perception segment obtained by the vehicle at the previous sampling point.

[0059] In an embodiment of the present application, a historical fusion segment is obtained, and the historical fusion segment includes relevant information of the historical continuous targets therein, such as position, features, etc. Then the coordinate system of the first continuous target in the first perception segment and the historical continuous target in the historical fusion segment is unified to ensure that they are compared and fused in the same coordinate system. According to the similarity metric, the first continuous target is matched with the historical continuous target. For the successfully matched continuous targets, they are fused according to the preset fusion rules, and for the unmatched continuous targets, they are directly added to the fusion results. The fused results are sorted and optimized to remove duplicate or abnormal data. The continuous target after fusion and all posture sampling points are used as the first fusion segment.

[0060] Step S103, sequentially obtaining first accumulated distances between sampling points after the first sampling point and the first sampling point, until a second sampling point whose first accumulated distance is greater than or equal to a first threshold is obtained.

[0061] In the embodiment of the present application, the position information of the first sampling point, such as the coordinate value, is determined, and then a variable is initialized to store the currently calculated distance, with an initial value of 0, and a variable is initialized to mark the sampling point currently being processed.

[0062] In particular, when the system is initialized, the first valid pose point output by the INS system can be directly used as the first sampling point.

[0063] After the first sampling point, traverse each pose point output by the INS system, accumulate the distances between adjacent pose points, obtain the first accumulated distance between them and the first sampling point, and compare the calculated first accumulated distance with the first threshold. If the currently calculated first accumulated distance is less than the first threshold, continue to process the next pose point. If the currently calculated first accumulated distance is greater than or equal to the first threshold, mark the current pose point as the second sampling point and end the loop. Finally, output the information of the second sampling point.

[0064] For example, assuming that the coordinates of the first sampling point are (10, 20), the coordinates of the subsequent pose points are (12, 22), (15, 25), and (20, 30), and the first threshold is 10. First, calculate the distance between the first subsequent pose point (12, 22) and (10, 20). Assuming the distance is 2.83, which is less than 10, continue to calculate the next one. Calculate the distance between the second subsequent pose point (15, 25) and the first subsequent pose point (12, 22). Assuming the distance is 4.24, 4.24+2.83 is less than 10, continue to calculate the next one. Calculate the distance between the third subsequent pose point (20, 30) and the second subsequent pose point (15, 25). Assuming the distance is 7.07, 4.24+2.83+7.07 is greater than 10, then (20, 30) is the second sampling point.

[0065] Step S104, obtaining a second perception segment obtained by processing the perception information of the current road section when the vehicle passes through the second sampling point, fusing the second perception segment with the first fusion segment to obtain a second fusion segment, and judging whether a preset condition is met when the vehicle passes through the second sampling point. If the preset condition is met, fitting the second fusion segment to obtain a target vector segment.

[0066] In an embodiment of the present application, when the vehicle travels to the second sampling point, various types of perception information of the current road section will be processed. These perception information include the geometry of the road, traffic signs and markings, the location of surrounding vehicles and obstacles, etc. By analyzing and integrating these information, a second perception segment is obtained. Then, the newly generated second perception segment is fused with the first fused segment previously obtained and processed at the first sampling point. The fusion process aims to integrate the information of different sampling points so that the second fused segment can contain more extensive and more accurate road section features. Subsequently, it is determined whether the preset conditions are met when the vehicle passes the second sampling point, where the preset condition is whether the second cumulative distance is reached. If the distance traveled by the vehicle has reached the preset second cumulative distance since the end sampling point of the previous vector segment, then it is considered that the preset conditions are met. When the preset conditions are met, the second fused segment will be fitted to obtain a smoother and more accurate target vector segment.

[0067] This application first fuses the historical fusion fragments and the perception fragments corresponding to the sampling points to ensure the accuracy of the generated map. Secondly, by obtaining the distances between the sampling points in sequence and determining the key second sampling points according to the threshold, it is possible to process important sampling points in a targeted manner, avoid indiscriminate processing of all sampling points, and reduce the amount of data that needs to be processed. Then when the vehicle passes the second sampling point, it is determined whether the preset conditions are met, and only when the conditions are met does the fitting process occur and the target vector fragment is output. Therefore, this application can reduce manual intervention by fusing the perception fragments obtained at each sampling point during the vehicle's driving process with the previously obtained historical fusion fragments. It solves the problem that the existing high-precision map production process consumes a lot of time and energy costs.

[0068] Figure 2 is a flow chart of a method for generating a map vector fragment according to an embodiment of the present invention. Figure 2 As shown, the process includes the following steps:

[0069] Step S201, obtaining a first perception segment obtained by processing perception information of a current road section when a vehicle passes a first sampling point, wherein the first perception segment includes a plurality of first continuous targets.

[0070] In the embodiment of the present application, obtaining a first perception segment obtained by sampling the current road section when the vehicle passes the first sampling point includes the following steps A1-A3:

[0071] Step A1, obtaining forward-looking perception data obtained by processing a road section image when the vehicle passes through a first sampling point, wherein the road section image is an image of a current road section, and the forward-looking perception data includes a plurality of continuous targets.

[0072] Specifically, when the vehicle passes the first sampling point, the front-view camera is triggered to start collecting images of the current road section. Multiple images can be collected continuously to ensure that clear and complete road section information is obtained. The collected image data is transmitted to a computing device on the vehicle for processing. A wired or wireless connection method can be used.

[0073] The captured image is then preprocessed to improve the accuracy of target detection. The preprocessed image is processed using the selected target detection algorithm to detect multiple continuous targets in the image. For example, a deep learning model can be used to infer the image, identify objects such as vehicles, pedestrians, traffic signs, and determine their positions and bounding boxes in the image. For each continuous target detected, its relevant feature information is extracted. This may include features such as the location, size, shape, and color of the target. These features can be extracted using image processing techniques and algorithms. Finally, the detected forward-looking perception data (including multiple continuous targets and their feature information) is sent to the vehicle-side processor on the one hand for the subsequent generation step of the target vector fragment of this patent, and on the other hand is stored in the storage device on the vehicle, or uploaded to the cloud server through a network connection for storage and subsequent analysis.

[0074] Step A2, using each of the multiple continuous targets to determine the interception starting point of the continuous target, and determining the interception end point of the continuous target based on the interception starting point of the continuous target and the segment interception length.

[0075] Specifically, select any lane line L from multiple first continuous targets as the operation object. Set the starting position of the traversal to x=0. Starting from x=0, check the sampling points in order from small to large. For each sampling point, substitute its relevant data into the following formula for judgment:

[0076] In the formula, θ fov Refers to the camera's field of view. s ,y s are the horizontal and vertical coordinates of the sampling point respectively. Figure 3 As shown, when the first sampling point that makes the formula valid is encountered, it is marked and determined as the interception starting point p of the continuous target. s (x s ,y s ).

[0077] Then, determine the next sampling point of the current sampling point. The specific process is as follows: From the first sampling point T i Start traversing the pose points of the INS system, accumulating the distances between adjacent pose points, and obtaining the distance between each pose point and the first sampling point T i The accumulated distance between them is calculated until the accumulated distance reaches or exceeds the first threshold, and the first sampling point T is found. i The next sampling point T j , record the accumulated T at this time i With T j The accumulated distance between them is recorded as d, and d is used as the first sampling point T iThe length of the lane line that needs to be intercepted for the continuous target in the corresponding perception segment (the so-called segment interception length). Figure 3 As shown, to intercept the starting point p s (x s ,y s ), intercept the length d on the current continuous target, and determine the horizontal coordinate of the end point as: x e =x s +d, the corresponding point is the interception end point p of the continuous target e (x e ,y e ). The information of the determined starting point and ending point is sorted and output for subsequent processing and use. For example, assuming that the coordinates of the starting point on a lane line are (10, 20), and the cumulative distance from the current sampling point to the next sampling point is d = 20. Then the length of the lane line to be intercepted is also d, so as to determine the interception end point of the lane line.

[0078] Step A3, using the interception starting point and the interception end point of each continuous target to intercept the corresponding continuous target, to obtain a first continuous target, and using the first continuous target and the first sampling point to construct a first perception segment.

[0079] In an embodiment of the present application, the corresponding continuous target is intercepted by the interception starting point and the interception end point of each continuous target, and the intercepted continuous target is used as the first continuous target. Finally, all the first continuous targets and the first sampling point constitute the first perception segment.

[0080] Step S202, obtain a historical fusion segment, and fuse the first continuous target in the first perception segment with the historical continuous target in the historical fusion segment to obtain a first fusion segment, wherein, if the previous sampling point of the first sampling point is the starting sampling point of the vector segment, then the historical fusion segment obtained at the first sampling point is the perception segment obtained by the vehicle at the previous sampling point.

[0081] In the embodiment of the present application, when the vehicle reaches the current first sampling point, the historical fusion segment formed at the previous sampling point is obtained. Then, the first continuous target in the current first perception segment is compared and integrated with the historical continuous target in the historical fusion segment of the previous sampling point. In this way, the consistency of the data can be maintained, so that the information obtained from different sampling points can be connected and supplemented in an orderly manner. At the same time, by fusing with the new first continuous target, the fusion result of the previous sampling point can be expanded to obtain the fusion result at the current sampling point.

[0082] In the embodiment of the present application, the first continuous target in the first perception segment is fused with the historical continuous target in the historical fusion segment to obtain the first fusion segment, including the following steps B1-B3:

[0083] Step B1, matching the first continuous target in the first perception segment with all continuous targets in the historical fusion segment to obtain a second continuous target matched with the first continuous target, wherein the second continuous target is the continuous target in the historical fusion segment that is closest to the first continuous target in lateral distance.

[0084] Specifically, first, the data of all continuous targets in the historical fusion segment are read from the relevant storage or data source, including information such as their position, direction, shape, etc. At the same time, the corresponding data of all continuous targets in the current first perception segment are obtained.

[0085] Secondly, the relative pose conversion matrix between the two frames is calculated using the pose of the previous frame and the pose of the current frame. For the target point position information of each continuous target in the historical fusion segment, the calculated relative pose conversion matrix is ​​used for coordinate conversion, so as to convert the target point position of each continuous target in the historical fusion segment from the previous frame pose coordinate system to the current frame pose coordinate system. After the conversion is completed, the continuous targets in the historical fusion segment and the continuous targets in the first perception segment are expressed in the coordinate system corresponding to the current pose.

[0086] Then, for each first continuous target, the lateral distance between it and all continuous targets in the historical fusion segment is calculated. The lateral distance can be determined by calculating the absolute value of the coordinate difference in the lateral direction of the vehicle (for example, the Y-axis direction). The continuous target in the historical fusion segment with the smallest lateral distance from the first continuous target is determined as the second continuous target.

[0087] As an example, Figure 4 As shown, the first perceptual segment j consists of sampling point T j and p of all lane lines it perceives sj -p ej The historical fusion segment includes not only all lane lines obtained by historical fusion (such as l i ), including the sampling point T i and the previous series of sampling points. The continuous target l in the first perception segment j j Continuous target l in the history fusion segment i The horizontal distance is the shortest, so l j With l i Form a matching pair.

[0088] Step B2, fusing the first continuous target with the second continuous target to obtain a fused continuous target.

[0089] Specifically, the first continuous target and the second continuous target are fused to obtain a fused continuous target, including: deleting the end point of the second continuous target; taking the starting point of the second continuous target, the starting point and the end point of the first continuous target as target points, and fusing based on the target points to obtain a fused continuous target.

[0090] Obtain the starting point and the ending point of the first continuous target, and the related data of the starting point and the ending point of the second continuous target. In the storage or data structure, mark the related data representing the ending point of the second continuous target as deleted or directly remove it from the data. The starting point of the fused continuous target is the starting point of the second continuous target. The ending point of the fused continuous target is the ending point of the first continuous target. Use the fused starting point, the ending point, and a series of target points in the middle to construct the fused continuous target.

[0091] As an example, Figure 5 As shown, taking the lane line as an example, assuming that the lane line l in the historical fusion segment i and the lane line l in the first perception segment j If a match is formed, delete l i The end point p ei , by p si , p sj , p ej Construct a new lane line, that is, obtain the fused lane line. Finally, all the fused lane lines, the unmatched lane lines in the historical fusion segment, the unmatched lane lines in the first perception segment, and all the sampled poses together constitute the first fusion segment.

[0092] Step B3, constructing a first fusion segment using the fused continuous target, the third continuous target, the fourth continuous target, all sampling points in the historical fusion segment, and the first sampling point, wherein the third continuous target is the continuous target in the historical fusion segment except the second continuous target, and the fourth continuous target is the continuous target in the first perception segment that is not matched to the second continuous target.

[0093] Specifically, first, the third continuous target is determined, that is, the continuous target other than the second continuous target in the historical fusion segment. This part of the target retains a specific part of the historical fusion information. At the same time, the fourth continuous target is determined, which is the continuous target that is not matched to the second continuous target in the first perception segment. By screening out this part of the unmatched target, new information is added to the fusion segment. Then, the fused continuous target, the third continuous target, the fourth continuous target and all the sampling points in the historical fusion segment are combined, and the first sampling point as a key reference is added to jointly construct the first fusion segment. This construction method fully integrates historical information and current perception information, taking into account specific continuous targets in the historical fusion segment, and incorporating unmatched targets in the new perception segment as well as all historical sampling points and key first sampling points, so that the first fusion segment can more comprehensively and accurately reflect the status and change trend of the vehicle's environment at the first sampling point.

[0094] The method provided in the embodiment of the present application can integrate multiple perception information by matching and fusing continuous targets, reduce data inaccuracies caused by limitations or errors of a single perception segment, and make the first fusion segment finally constructed more accurately reflect the actual situation. Secondly, fusing related continuous targets at different times can supplement and improve data, avoid information loss, and provide more comprehensive perception information. And by finding the closest continuous target for fusion, the tracking of continuous targets can be made more continuous and stable, reducing the interruption or loss of continuous targets.

[0095] Step S203, sequentially obtaining first accumulated distances between sampling points after the first sampling point and the first sampling point, until a second sampling point whose first accumulated distance is greater than or equal to a first threshold is obtained.

[0096] In the embodiment of the present application, each pose point output by the INS system is traversed after the first sampling point, the distances between adjacent pose points are accumulated to obtain the first accumulated distance between them and the first sampling point, and the calculated first accumulated distance is compared with the first threshold. If the currently calculated first accumulated distance is less than the first threshold, the next pose point is processed. If the currently calculated first accumulated distance is greater than or equal to the first threshold, the current pose point is marked as the second sampling point, and the loop ends. Finally, the information of the second sampling point is output.

[0097] It should be noted that, although the original ins pose output frequency is very high, such a high frequency pose point is not necessary for the mapping of this application. Therefore, in order to save resources, this application reduces the number of pose points according to the driving distance, for example, pose sampling is performed every 1 meter. The specific operation is to accumulate the distance between the current pose point and the pose point of the next adjacent frame, starting from the current sampling point. When the accumulated distance reaches 1 meter, the pose point at this time is determined as the next sampling point. Then, take this newly determined pose point as the new current sampling point, and reset the first accumulated distance to 0, and repeat the above process of accumulating the first accumulated distance and determining the next sampling point.

[0098] For example, if Figure 6 As shown, from the first sampling point T i Starting from 0, the distance between the pose points of two adjacent frames is accumulated to obtain the first accumulated distance d i , until d i ≥1m, the next sampling point T can be obtained j . After that, T j is the current sampling point, the first accumulated distance is reset to 0, and the above process is repeated. i With T j 、T j With T k If the distances between them are all greater than or equal to 1m, the three can be selected as sampling points.

[0099] Step S204, obtaining a second perception segment obtained by processing the perception information of the current road section when the vehicle passes through the second sampling point, fusing the second perception segment with the first fusion segment to obtain a second fusion segment, and determining whether a preset condition is met when the vehicle passes through the second sampling point. If the preset condition is met, fitting the second fusion segment to obtain a target vector segment.

[0100] In an embodiment of the present application, obtaining a second fused segment obtained by sampling the current road section when the vehicle passes the second sampling point includes: obtaining a second perception segment sampling the current road section when the vehicle passes the second sampling point, and fusing the continuous targets in the second perception segment with the continuous targets in the first fused segment to obtain a second fused segment. The specific fusion process refers to the fusion process of the first fused segment in the above embodiment, which will not be described in detail here.

[0101] In an embodiment of the present application, determining whether a preset condition is satisfied when the vehicle passes through the second sampling point includes: determining whether the preset condition is satisfied when the vehicle passes through the second sampling point based on whether the second accumulated distance is greater than or equal to a second threshold, wherein if the second accumulated distance is greater than or equal to the second threshold, it is determined that the preset condition is satisfied when the vehicle passes through the second sampling point; or, if the comparison result is that the second accumulated distance is less than the second threshold, it is determined that the preset condition is not satisfied when the vehicle passes through the second sampling point.

[0102] Specifically, first, obtain the second accumulated distance when the vehicle passes the second sampling point. This second accumulated distance is calculated by accumulating the distance from the end sampling point of the vehicle passing the previous vector segment. Then, compare the second accumulated distance with the second threshold. If the second accumulated distance is greater than or equal to the second threshold, it is determined that the preset condition is met when the vehicle passes the second sampling point. This means that in this case, the cumulative distance traveled by the vehicle reaches or exceeds a certain standard, indicating that the vehicle has completed a specific mileage and the target vector segment can be output. On the contrary, if the comparison result is that the second accumulated distance is less than the second threshold, then it is determined that the preset condition is not met when the vehicle passes the second sampling point. This means that the cumulative distance traveled by the vehicle has not yet reached the specified standard, and subsequent fusion operations need to be continued.

[0103] In the embodiment of the present application, fitting the second fused segment to obtain the target vector segment includes the following steps C1-C6:

[0104] Step C1: Create a sliding window according to a preset window length for each continuous target in the second fusion segment.

[0105] Specifically, the target point data of each continuous target in the second fusion segment is obtained, and a sliding window with a preset window length of 4 is created.

[0106] Step C2: sliding the sliding window from the starting target point of each continuous target to the ending target point in sequence, and taking the target point of the continuous target in the sliding window as the control point.

[0107] Step C3, performing third-order B-spline fitting using the control points to obtain a fitting curve, wherein the fitting curve includes a plurality of fitting points.

[0108] Specifically, the target point in the sliding window is used as the control point to perform third-order B-spline fitting. The parameter equation of the third-order B-spline curve is as follows:

[0109]

[0110] Wherein, P is the control point matrix, t is a vector composed of powers of parameter t, satisfying 0≤t≤1, and C is the polynomial coefficient of the third-order B-spline curve.

[0111] Then, in each sliding window, t=0.1, 0.2, 0.3...0.9, 1 are taken in turn to obtain 10 fitting point outputs and the fitting curve (such as Figure 7 As shown, the polyline includes the original target points input before fitting, and the curve includes the fitting points output after fitting), and then the curvature at each fitting point is calculated along the fitting curve. For example: For each continuous target in the fusion segment, a sliding window will be created according to the pre-set window size. After the sliding window is created, the window will be slid forward in sequence according to a certain step size (for example, the position of one target point is moved each time). After each slide, the target points contained in the window are used as control points. These control points are used for subsequent calculations and processing. Inside each sliding window, data is selected according to specific rules to generate 10 fitting points.

[0112] Step C4, calculating the curvature of each position based on the fitting points in the fitting curve.

[0113] Specifically, the curvature reflects the degree of curvature of the curve at a certain point. By analyzing the positional relationship of each fitting point on the fitting curve, the curvature of the curve at each fitting point can be calculated using a specific mathematical method.

[0114] Step C5, obtaining a target sampling interval corresponding to the curvature, and downsampling the fitting points using the target sampling interval, and obtaining fitting sampling points by downsampling, wherein the fitting sampling points are used to represent the final state of the continuous target.

[0115] Specifically, obtaining the target sampling interval corresponding to the curvature includes: comparing the curvature with a preset curvature; if the curvature is greater than or equal to the preset curvature, using the first sampling interval as the target sampling interval; or, if the curvature is less than the preset curvature, using the second sampling interval as the target sampling interval, wherein the first sampling interval is less than the second sampling interval.

[0116] It should be noted that if the curvature is greater than or equal to the preset curvature, the smaller first sampling interval (such as 1 meter) is used as the target sampling interval. If the curvature is less than the preset curvature, the larger second sampling interval (such as 3 meters) is used as the target sampling interval. Finally, according to the determined target sampling interval, the fitting points are downsampled to obtain fitting sampling points.

[0117] Step C6: constructing a continuous target within the second threshold using the fitting sampling points, and constructing the target vector segment based on the continuous target within the second threshold and all the sampling points within the second threshold.

[0118] Specifically, the continuous targets within the second threshold are constructed using the fitting sampling points. Then, a target vector segment is constructed based on the continuous targets within the second threshold and all the sampling points within the second threshold. This target vector segment combines the information of the continuous targets constructed by the fitting sampling points and all the sampling points within the second threshold, and can more comprehensively reflect the characteristics and status of the specific area.

[0119] The method provided in the embodiment of the present application first uses a sliding window to select control points for B-spline fitting, which can smooth the target points of the continuous target and output a continuous target shape that conforms to the actual situation. Secondly, different sampling intervals are dynamically selected according to the size of the curvature to further downsample the fitting points. Using a smaller sampling interval in areas with large curvature (such as sharp turns on a curve) can more accurately capture shape changes; using a larger sampling interval in areas with small curvature (such as flat roads) can save resources without losing key information, making the sampling more adaptive and targeted. The fitting points are then downsampled to remove redundant data, highlight key features, and reduce the amount of data that needs to be stored. Ultimately, the adaptive sampling interval based on curvature can more accurately reflect the true characteristics of the road or curve, and improve the accuracy of the target vector fragment.

[0120] In this embodiment, a device for generating a map vector fragment is also provided, and the device is used to implement the above-mentioned embodiments and preferred implementation modes, and the descriptions that have been made are not repeated. As used below, the term "module" can be a combination of software and / or hardware that implements a predetermined function. Although the device described in the following embodiments is preferably implemented in software, the implementation of hardware, or a combination of software and hardware, is also possible and conceivable.

[0121] This embodiment provides a device for generating a map vector segment, such as Figure 8 As shown, including:

[0122] An acquisition module 801 is used to acquire a first perception segment obtained by processing the perception information of the current road section when the vehicle passes through the first sampling point, wherein the first perception segment includes a plurality of first continuous targets;

[0123] A first fusion module 802 is used to obtain a historical fusion segment, and fuse the first continuous target in the first perception segment with the historical continuous target in the historical fusion segment to obtain a first fusion segment, wherein if the previous sampling point of the first sampling point is the starting sampling point of the vector segment, the historical fusion segment obtained at the first sampling point is the perception segment obtained by the vehicle at the previous sampling point;

[0124] A traversal module 803 is used to sequentially obtain a first accumulated distance between the sampling points after the first sampling point and the first sampling point until a second sampling point whose distance is greater than or equal to the first threshold is obtained;

[0125] The second fusion module 804 is used to obtain a second perception segment obtained by processing the perception information of the current road section when the vehicle passes through the second sampling point, fuse the second perception segment with the first fusion segment to obtain a second fusion segment, and determine whether a preset condition is met when the vehicle passes through the second sampling point. If the preset condition is met, the second fusion segment is fitted to obtain a target vector segment.

[0126] In an embodiment of the present application, an acquisition module 801 is used to obtain forward-looking perception data obtained by processing a road section image when a vehicle passes a first sampling point, wherein the road section image is an image of a current road section, and the forward-looking perception data includes multiple continuous targets; using each of the multiple continuous targets to determine a cut-off starting point of the continuous target, and determining a cut-off end point of the continuous target based on the cut-off starting point of the continuous target and a cut-off length of the segment; using the cut-off starting point and the cut-off end point of each continuous target to cut off the corresponding continuous target to obtain a first continuous target, and using the first continuous target and the first sampling point to construct a first perception segment.

[0127] In an embodiment of the present application, the first fusion module 802 is used to match the first continuous target in the first perception segment with all continuous targets in the historical fusion segment to obtain a second continuous target that matches the first continuous target, wherein the second continuous target is the continuous target in the historical fusion segment that is closest to the first continuous target in lateral distance; fuse the first continuous target with the second continuous target to obtain a fused continuous target; and construct a first fused segment using the fused continuous target, the third continuous target, the fourth continuous target, all sampling points in the historical fusion segment, and the first sampling point, wherein the third continuous target is a continuous target in the historical fusion segment other than the second continuous target, and the fourth continuous target is a continuous target in the first perception segment that is not matched with the second continuous target.

[0128] In an embodiment of the present application, the first fusion module 802 is used to delete the end point of the second continuous target; take the starting point of the second continuous target, the starting point and the end point of the first continuous target as target points, and fuse based on the target points to obtain a fused continuous target.

[0129] In an embodiment of the present application, the second fusion module 804 is used to obtain a second accumulated distance when the vehicle passes through a second sampling point, wherein the second accumulated distance is accumulated and calculated based on the termination sampling point of the previous vector fragment; and whether a preset condition is met when the vehicle passes through the second sampling point is determined according to whether the second accumulated distance is greater than or equal to a second threshold, wherein if the second accumulated distance is greater than or equal to the second threshold, it is determined that the preset condition is met when the vehicle passes through the second sampling point; or, if the second accumulated distance is less than the second threshold, it is determined that the preset condition is not met when the vehicle passes through the second sampling point.

[0130] In an embodiment of the present application, the second fusion module 804 is used to create a sliding window according to a preset window length for each continuous target in the second fusion segment; slide the sliding window in sequence, and use the target points of the continuous targets in the sliding window as control points; use the control points to perform third-order B-spline fitting to obtain a fitting curve, wherein the fitting curve includes multiple fitting points; calculate the curvature of each position based on the fitting points in the fitting curve; obtain the target sampling interval corresponding to the curvature, and use the target sampling interval to downsample the fitting points to obtain fitting sampling points, wherein the fitting sampling points are used to represent the final state of the continuous target; use the fitting sampling points to construct a continuous target within a second threshold, and construct a target vector segment based on the continuous target within the second threshold and all sampling points within the second threshold.

[0131] In an embodiment of the present application, the second fusion module 804 is used to compare the curvature with a preset curvature; if the curvature is greater than or equal to the preset curvature, the first sampling interval is used as the target sampling interval; or, if the curvature is less than the preset curvature, the second sampling interval is used as the target sampling interval, wherein the first sampling interval is less than the second sampling interval.

[0132] See also Fig. 9 , Fig. 9 is a schematic diagram of the structure of a computer device provided by an optional embodiment of the present invention, such as Fig. 9 As shown, the computer device includes: one or more processors 10, a memory 20, and interfaces for connecting various components, including high-speed interfaces and low-speed interfaces. Various components are connected to each other using different buses for communication, and can be installed on a common mainboard or installed in other ways as needed. The processor can process the instructions executed in the computer device, including instructions stored in or on the memory to display the graphical information of the GUI on an external input / output device (such as, a display device coupled to the interface). In some optional embodiments, if necessary, multiple processors and / or multiple buses can be used together with multiple memories and multiple memories. Similarly, multiple computer devices can be connected, and each device provides some necessary operations (for example, as a server array, a group of blade servers, or a multi-processor system).

[0133] The processor 10 may be a central processing unit, a network processor or a combination thereof. The processor 10 may further include a hardware chip. The hardware chip may be a dedicated integrated circuit, a programmable logic device or a combination thereof. The programmable logic device may be a complex programmable logic device, a field programmable gate array, a general purpose array logic or any combination thereof.

[0134] The memory 20 stores instructions executable by at least one processor 10, so that the at least one processor 10 executes the method shown in the above embodiment.

[0135] The memory 20 may include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application required for at least one function; the data storage area may store data created by the use of a computer device based on the presentation of a small program landing page, etc. In addition, the memory 20 may include a high-speed random access memory, and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some optional embodiments, the memory 20 may optionally include a memory remotely arranged relative to the processor 10, and these remote memories may be connected to the computer device via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0136] The memory 20 may include a volatile memory, such as a random access memory; the memory may also include a non-volatile memory, such as a flash memory, a hard disk or a solid state drive; the memory 20 may also include a combination of the above types of memory.

[0137] The computer device further comprises a communication interface 30 for the computer device to communicate with other devices or a communication network.

[0138] The embodiment of the present invention also provides a computer-readable storage medium. The method according to the embodiment of the present invention can be implemented in hardware, firmware, or can be implemented as a computer code that can be recorded in a storage medium, or can be implemented as a computer code that is originally stored in a remote storage medium or a non-temporary machine-readable storage medium and will be stored in a local storage medium through a network download, so that the method described herein can be stored in such software processing on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. Among them, the storage medium can be a magnetic disk, an optical disk, a read-only storage memory, a random access memory, a flash memory, a hard disk or a solid-state hard disk, etc.; further, the storage medium can also include a combination of the above types of memories. It can be understood that a computer, a processor, a microprocessor controller, or programmable hardware includes a storage component that can store or receive software or computer code. When the software or computer code is accessed and executed by a computer, a processor, or hardware, the method shown in the above embodiment is implemented.

[0139] Although the embodiments of the present invention have been described in conjunction with the accompanying drawings, those skilled in the art may make various modifications and variations without departing from the spirit and scope of the present invention, and such modifications and variations are all within the scope defined by the appended claims.

Claims

1. A method for generating a map vector segment, characterized in that: The method comprises: Acquire a first perception segment obtained by processing perception information of a current road section when the vehicle passes through a first sampling point, wherein the first perception segment includes a plurality of first continuous targets; Acquire a historical fusion segment, and fuse the first continuous target in the first perception segment with the historical continuous target in the historical fusion segment to obtain a first fusion segment, wherein if the previous sampling point of the first sampling point is the starting sampling point of the vector segment, then the historical fusion segment acquired at the first sampling point is the perception segment obtained by the vehicle at the previous sampling point; Sequentially acquiring first accumulated distances between sampling points after the first sampling point and the first sampling point until a second sampling point whose first accumulated distance is greater than or equal to a first threshold is obtained; A second perception segment is obtained by processing the perception information of the current road section when the vehicle passes through the second sampling point, the second perception segment is fused with the first fusion segment to obtain a second fusion segment, and it is determined whether a preset condition is satisfied when the vehicle passes through the second sampling point. If the preset condition is satisfied, the second fusion segment is fitted to obtain a target vector segment.

2. The method according to claim 1, characterized in that The obtaining of the first perception fragment obtained by processing the perception information of the current road section when the vehicle passes the first sampling point includes: Acquire forward-looking perception data obtained by processing a road section image when the vehicle passes through the first sampling point, wherein the road section image is an image of a current road section, and the forward-looking perception data includes a plurality of continuous targets; Determine a cut-off point of the continuous target by using each of the plurality of continuous targets, and determine a cut-off point of the continuous target based on the cut-off point of the continuous target and the cut-off length of the fragment; The corresponding continuous target is intercepted using the interception starting point and the interception end point of each continuous target to obtain a first continuous target, and the first perception segment is constructed using the first continuous target and the first sampling point.

3. The method according to claim 1, characterized in that The step of fusing the first continuous target in the first perception segment with the historical continuous target in the historical fusion segment to obtain a first fusion segment includes: Matching the first continuous target in the first perception segment with all the continuous targets in the historical fusion segment to obtain a second continuous target matched with the first continuous target, wherein the second continuous target is the continuous target with the shortest lateral distance to the first continuous target in the historical fusion segment; Fusing the first continuous target with the second continuous target to obtain a fused continuous target; A first fusion segment is constructed using the fused continuous target, the third continuous target, the fourth continuous target, all sampling points in the historical fusion segment, and the first sampling point, wherein the third continuous target is a continuous target in the historical fusion segment except the second continuous target, and the fourth continuous target is a continuous target in the first perception segment that is not matched to the second continuous target.

4. The method according to claim 3, characterized in that The step of fusing the first continuous target with the second continuous target to obtain a fused continuous target includes: Deleting the end point of the second continuous target; The starting point of the second continuous target, the starting point and the ending point of the first continuous target are taken as target points, and fusion is performed based on the target points to obtain the fused continuous target.

5. The method according to claim 1, characterized in that The determining whether a preset condition is satisfied when the vehicle passes through the second sampling point includes: Acquire a second accumulated distance when the vehicle passes through a second sampling point, wherein the second accumulated distance is accumulated and calculated based on the end sampling point of the previous vector segment; According to whether the second accumulated distance is greater than or equal to a second threshold, it is determined whether a preset condition is satisfied when the vehicle passes through the second sampling point, wherein if the second accumulated distance is greater than or equal to the second threshold, it is determined that the preset condition is satisfied when the vehicle passes through the second sampling point; or, if the second accumulated distance is less than the second threshold, it is determined that the preset condition is not satisfied when the vehicle passes through the second sampling point.

6. The method according to claim 5, characterized in that The step of fitting the second fused segment to obtain the target vector segment includes: For each continuous target in the second fusion segment, a sliding window is created according to a preset window length; Sliding the sliding window in sequence, and taking the target points of the continuous targets in the sliding window as control points; Performing third-order B-spline fitting using the control points to obtain a fitting curve, wherein the fitting curve includes a plurality of fitting points; Calculating the curvature of each position based on the fitting points in the fitting curve; Obtaining a target sampling interval corresponding to the curvature, and downsampling the fitting points using the target sampling interval to obtain fitting sampling points, wherein the fitting sampling points are used to represent the final state of the continuous target; The continuous target within the second threshold is constructed by using the fitting sampling points, and the target vector segment is constructed based on the continuous target within the second threshold and all the sampling points within the second threshold.

7. The method according to claim 6, characterized in that The obtaining of the target sampling interval corresponding to the curvature includes: comparing the curvature with a preset curvature; If the curvature is greater than or equal to the preset curvature, the first sampling interval is used as the target sampling interval; or, if the curvature is less than the preset curvature, the second sampling interval is used as the target sampling interval, wherein the first sampling interval is less than the second sampling interval.

8. A device for generating a map vector segment, characterized in that: The device comprises: an acquisition module, configured to acquire a first perception segment obtained by processing the perception information of the current road section when the vehicle passes through the first sampling point, wherein the first perception segment includes a plurality of first continuous targets; a fusion module, configured to obtain a historical fusion segment, and fuse the first continuous target in the first perception segment with the historical continuous target in the historical fusion segment to obtain a first fusion segment, wherein if the previous sampling point of the first sampling point is the starting sampling point of the vector segment, the historical fusion segment obtained at the first sampling point is the perception segment obtained by the vehicle at the previous sampling point; A traversal module, used for sequentially acquiring a first accumulated distance between the first sampling point and the sampling points after the first sampling point, until a second sampling point whose distance is greater than or equal to a first threshold is obtained; a processing module, used to obtain a second perception segment obtained by processing the perception information of the current road section when the vehicle passes through the second sampling point, fuse the second perception segment with the first fusion segment to obtain a second fusion segment, and determine whether a preset condition is met when the vehicle passes through the second sampling point; if the preset condition is met, the second fusion segment is fitted to obtain a target vector segment.

9. A computer device, characterized in that: include: A memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the method according to any one of claims 1 to 7 by executing the computer instructions.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a computer to execute the method according to any one of claims 1 to 7.

Images