A method and apparatus for generating map lane data

Abstract

This invention discloses a method for generating map lane data, relating to the field of autonomous driving technology. The method includes collecting lane layout information and dividing the lanes into ordinary lanes and connecting lanes based on the lane layout information; drawing the edge lines of all ordinary lanes and generating center lines of the ordinary lanes based on the edge lines; performing point spacing uniformization and smoothing processing on the center lines of the ordinary lanes; generating center lines of all connecting lanes according to a connecting line generation algorithm; and generating all missing edge lines of the connecting lanes based on the center lines of the connecting lanes. This invention divides all lanes into ordinary lanes and connecting lanes, then performs a smoothing algorithm on the center lines of the ordinary lanes, and then uses a connecting line generation algorithm to generate the center lines of the connecting lanes, ensuring that the points of all lane center lines have continuous curvature, enabling autonomous vehicles to operate more effectively.

Description

Technical Field

[0001] This invention relates to the field of autonomous driving technology, and in particular to a method and apparatus for generating map lane data. Background Technology

[0002] High-precision maps are an essential component of current autonomous driving technology, and lane data is the core content of high-precision maps. Vehicles typically need to drive automatically along a series of coordinate points provided by the lane centerline data in the high-precision map as a predetermined trajectory. In order for the vehicle's automatic control module to accurately control the vehicle and achieve smooth driving, the sequence of lane centerline coordinate points needs to conform to the vehicle's kinematics, that is, the curvature should be continuous, or the curve should be as smooth as possible.

[0003] Current technologies perform smoothing in segments or locally. This local processing method can ensure that most points on the map are relatively smooth, but it is difficult to guarantee that all points globally have continuous curvature for complex locations such as topological bifurcation points. Existing global smoothing methods cannot handle the curvature continuity problem at bifurcation points well, and are even more difficult to smooth loops. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to overcome the shortcomings of the prior art and provide a method and apparatus for generating map lane data.

[0005] To solve the above technical problems, the technical solution of the present invention is as follows:

[0006] A method for generating map lane data, comprising,

[0007] Collect lane layout information and divide the lanes into ordinary lanes and connecting lanes based on the lane layout information;

[0008] Draw the edge lines of all regular lanes, and generate the center lines of the regular lanes based on the edge lines of the regular lanes;

[0009] The center line of the ordinary lane is subjected to point spacing uniformization and smoothing.

[0010] Generate the center lines of all connecting lanes based on the connector generation algorithm;

[0011] Generate all missing edge lines for the connecting lanes based on the center line of the connecting lanes.

[0012] In a preferred embodiment of the map lane data generation method of the present invention, after drawing the edge lines of all ordinary lanes and generating the center lines of ordinary lanes based on the edge lines of the ordinary lanes, the method further includes:

[0013] Determine whether there is a topological bifurcation point on the center line of the ordinary lane. If there is, set the lane at the topological bifurcation point as a connecting lane.

[0014] In a preferred embodiment of the map lane data generation method of the present invention, the step of drawing the edge lines of all ordinary lanes and generating the center lines of ordinary lanes based on the edge lines of the ordinary lanes includes:

[0015] Obtain the point sequence of the two side lines of the ordinary lane, and obtain the coordinates of each point in the point sequence in sequence along the direction of the side lines;

[0016] Calculate the proportion of each point on both sides of the ordinary lane relative to its current position on the lane.

[0017] Along the direction of the two side lines, starting from the starting point of the two side lines, points are taken sequentially. Each time a set of target points is taken, the midpoint between the two target points is calculated as a point on the center line, and all the points that make up the center line are obtained, thus generating the center line of the ordinary lane.

[0018] In a preferred embodiment of the map lane data generation method of the present invention, after determining whether there is a topological bifurcation point on the centerline of the ordinary lane, and if so, setting the lane at the topological bifurcation point as a connecting lane, the method further includes:

[0019] Determine whether the centerline of the ordinary lane has a circular topology. If it does, then extract an arc-shaped section of the ordinary lane that forms the circular topology as a connecting lane.

[0020] In a preferred embodiment of the map lane data generation method of the present invention, the centerline arc length of the intercepted arc-shaped lane is greater than or equal to 1m.

[0021] As a preferred embodiment of the map lane data generation method of the present invention, the step of uniformizing the point intervals and smoothing the center lines of the ordinary lanes includes,

[0022] Linear interpolation is performed between the points in the center line of the ordinary lane to make the points in the center line of the ordinary lane evenly distributed.

[0023] Treat all the center lines of the aforementioned ordinary lanes that are connected one after the other as a continuous curve and smooth them out.

[0024] The smoothed curve is used as the center line of the ordinary lane.

[0025] As a preferred embodiment of the map lane data generation method of the present invention, the step of generating the center lines of all connecting lanes according to the connector generation algorithm includes,

[0026] Obtain the starting point parameters and ending point parameters of the center line of the connecting lane to be generated. The starting point parameters include the first coordinates, first curvature, and first direction angle of the starting point, and the ending point parameters include the second coordinates, second curvature, and second direction angle of the ending point.

[0027] The first coordinates, first curvature, and first direction angle of the starting point, and the second coordinates, second curvature, and second direction angle of the ending point are used as input parameters and input into the connector generation algorithm.

[0028] The connecting line generation algorithm generates the center line of the connecting lane, and obtains the curve model and curve parameters corresponding to the center line of the connecting lane.

[0029] In a preferred embodiment of the map lane data generation method of the present invention, after generating all missing edge lines of the connecting lanes based on the center lines of the connecting lanes, the method further includes:

[0030] Nonlinear interpolation is performed on all points in the center lines of the ordinary lanes to make the point density in the center lines of the ordinary lanes reach a set value. Then, the points corresponding to the center lines of the connecting lanes are resampled according to the curve model and curve parameters of the connecting lanes to make the point density in the center lines of the connecting lanes reach a set value.

[0031] The present invention also discloses a map lane data generation device, comprising,

[0032] The first acquisition module is used to acquire lane layout information and divide the lanes into ordinary lanes and connecting lanes based on the lane layout information.

[0033] The first generation module is used to generate the edge lines of all the ordinary lanes and generate the center lines of the ordinary lanes based on the edge lines of the ordinary lanes.

[0034] The first processing module is used to perform point homogenization processing and smoothing processing on the center line of the ordinary lane.

[0035] The first judgment module is used to determine whether there is a topological bifurcation point on the center line of the ordinary lane, and if there is, set the lane at the topological bifurcation point as a connecting lane.

[0036] The second judgment module is used to determine whether there is a circular topology in the center line of the ordinary lane, and if there is, to cut out a section of the arc lane in the ordinary lane that forms the circular topology as a connecting lane.

[0037] The second generation module is used by the connector generation algorithm to generate the center lines of all connecting lanes;

[0038] The third generation module is used to generate all the missing edge lines of the connecting lanes based on the center line of the connecting lane;

[0039] The second processing module is used to perform nonlinear interpolation on all the points in the center lines of the ordinary lanes to make the point density in the center lines of the ordinary lanes reach a set value, and to resample the points corresponding to the center lines of the connecting lanes according to the curve model and curve parameters of the connecting lanes to make the point density in the center lines of the connecting lanes reach a set value.

[0040] In a preferred embodiment of the map lane data generation device of the present invention, the centerline arc length of the intercepted arc-shaped lane is greater than or equal to 1m.

[0041] The beneficial effects of this invention are:

[0042] (1) This invention divides all lanes into ordinary lanes and connecting lanes, then performs a smoothing algorithm on the center line of the ordinary lanes, and then uses a connecting line generation algorithm to generate the center line of the connecting lanes, ensuring that the points of the center line of all lanes are curvature continuous, so that autonomous vehicles can operate better.

[0043] (2) The present invention analyzes the map as a whole, divides the lanes into ordinary lanes and connecting lanes, processes them separately, and divides the ordinary lanes at the topological bifurcation points, such as lane splitting and merging, into connecting lanes. In the case of ordinary lanes forming a ring topology, such as roundabouts, any segment of ordinary lanes in the ring road is separated and divided into connecting lanes, which effectively solves the problem of topological bifurcation points and the difficulty in smoothing the center line of the ring.

[0044] (3) The present invention first sets the interval of the lane center line to be relatively large, smooths the lane center line, and regenerates the dense point sequence after smoothing. Finally, all the lane center lines are continuous in curvature and uniformly spaced, and the generation speed is very fast.

[0045] (4) By improving the process and method, the present invention makes the center line generation speed faster, can better ensure that the center line with continuous curvature can be obtained in all positions and situations, and the center line is smoother. At the same time, the virtual edge line is more reasonable and smoother. Finally, the efficiency of producing a large amount of lane data is higher and the data quality is better, so that the autonomous vehicle drives more smoothly and the passengers have a better riding experience. Attached Figure Description

[0046] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0047] Figure 1 A flowchart illustrating the steps of the map lane data generation method provided by this invention;

[0048] Figure 2 A schematic diagram of the specific process of step S105 in the map lane data generation method provided by the present invention;

[0049] Figure 3 A schematic diagram of the specific process of step S106 in the map lane data generation method provided by the present invention;

[0050] Figure 4 A schematic diagram showing the effect after drawing the edge lines of a regular lane;

[0051] Figure 5 This is an auxiliary schematic diagram of the ordinary lane centerline generation algorithm in this embodiment;

[0052] Figure 6 This is a schematic diagram showing the effect after the center line of a regular lane is generated.

[0053] Figure 7 This is a schematic diagram illustrating the first effect of dividing ordinary lanes at a topological bifurcation point into connecting lanes;

[0054] Figure 8 This is a schematic diagram illustrating the second effect of dividing ordinary lanes at a topological bifurcation point into connecting lanes;

[0055] Figure 9 This is a schematic diagram illustrating the effect of dividing a section of a regular lane into a connecting lane in the middle of a regular lane that forms a circular topology.

[0056] Figure 10 A schematic diagram illustrating the effect of generating a centerline for a connecting lane with continuous curvature;

[0057] Figure 11 A diagram illustrating a lane connection that lacks a left or right lane line;

[0058] Figure 12 This is a schematic diagram of the final high-precision lane effect;

[0059] Figure 13 This is a schematic diagram of the map lane data generation device. Detailed Implementation

[0060] To make the content of this invention easier to understand, the invention will be further described in detail below with reference to specific embodiments and accompanying drawings.

[0061] See Figure 1 This embodiment provides a flowchart of a method for generating map lane data. The method includes steps S101 to S108, and the specific steps are described below:

[0062] Step S101: Collect lane layout information and divide the lanes into ordinary lanes and connecting lanes based on the lane layout information.

[0063] Specifically, data collection sensors are installed on a map collection vehicle. As the vehicle moves along the road, these sensors collect lane layout information. After collection, the data is processed to obtain road information and a road base map within the collected area. Based on this road information, lane layout information can be obtained, and lanes are divided into regular lanes and connecting lanes. Regular lanes are generally lanes with clearly defined road markings as their edges, while connecting lanes are typically virtual lanes without clearly defined edges, such as turning lanes within intersections.

[0064] Step S102: Draw the edge lines of all ordinary lanes, and generate the center lines of ordinary lanes based on the edge lines of ordinary lanes;

[0065] Specifically, after lane division is completed, the edge lines of ordinary lanes are drawn on the road base map. Each ordinary lane has two edge lines, one on the left and one on the right, with the lengths of the two edge lines being roughly the same. The drawing of the ordinary lane edge lines is based on the lane layout information collected by the sensors, that is, the lane markings of the ordinary lanes collected by the sensors. The effect of drawing the ordinary lane edge lines can be seen in [the image / description]. Figure 4 .

[0066] It is understandable that the edge lines of ordinary lanes are either solid lines or dashed lines. In this embodiment, the lane lines are segmented at the junction of the solid and dashed lines, and the lanes are numbered sequentially according to the segmentation.

[0067] After drawing the edges of all regular lanes, the center lines of the regular lanes are generated based on the left and right edges. For any regular lane, the left and right edges and lane direction are specified, and then the center line generation algorithm is used to generate the center line of the regular lane. The center line generation algorithm for any regular lane includes the following steps:

[0068] Step S102a: Obtain the point sequence of the two side lines of the ordinary lane, and obtain the coordinates of each point in the point sequence in sequence along the direction of the side lines.

[0069] Specifically, the side lines of a regular lane are represented by a series of points arranged in sequence, forming a broken line. When obtaining the coordinates of each point in the sequence, the direction of the side line needs to be determined—whether it's forward or reverse. If it's reverse, the coordinates of that side line are taken from back to front when generating the center line, as shown in the diagram (L4, L3, L2, L1, L0). If it's forward, the coordinates of that side line are taken from front to back, as shown in the diagram (L0, L1, L2, L3, L4).

[0070] Step S102b: Calculate the proportion of each point on both sides of the ordinary lane to the position on the current lane.

[0071] Specifically, with Figure 5 Taking the left-hand line as an example, first calculate the length of each segment on the broken line. For example, the length between L0 and L1 is d1, and the length between L1 and L2 is d2. Then, add up the lengths of all segments to get the total length d of the broken line. Then, starting from L0, calculate the proportion value of each point's position in turn. For example, the proportion value of L0 is s0 = 0.0, the proportion value of L1 is s1 = d1 / d, the proportion value of L2 is s2 = (d1 + d2) / d, that is, the proportion value s = (cumulative length from the starting point of the broken line to the current point) / the total length of the broken line. The proportion value of the endpoint L4 is s4 = 1.0. In this way, we get all the proportion values ​​of the left-hand line (s0, s1, s2, s3, s4) and all the proportion values ​​of the right-hand line (w0, w1, w2, w3, w4).

[0072] Step S102c: Along the directions of the two sidelines, starting from the starting points of the two sidelines, take points sequentially. For each set of target points, calculate the midpoint between the two target points, and save it as a point on the centerline to obtain all the points constituting the centerline. The specific steps are as follows:

[0073] Step S102c-1: Obtain the target point L on the left line and the target point R on the right line, calculate the midpoint between the two points, and save them.

[0074] Step S102c-2: Take the next point on the left and right lines as target points L and R.

[0075] Step S102c-3: If the proportion of the current point L on the left line is less than the proportion of the current point R on the right line, or if the current point R on the right line is the endpoint, proceed to step S102c-4; otherwise, proceed to step S102c-5.

[0076] Step S102 c-4: Let r = (proportional value of the current point L on the left line - proportional value of the previous point on the right line) / (proportional value of the current point on the right line - proportional value of the previous point on the right line). Calculate the coordinates (x, y) of point Q on the right line segment: x = (x-coordinate of the current point R on the right line - x-coordinate of the previous point on the right line) * r + x-coordinate of the previous point on the right line, y = (y-coordinate of the current point R on the right line - y-coordinate of the previous point on the right line) * r + y-coordinate of the previous point on the right line. Then calculate the midpoint P between the current point L and Q on the left line, i.e., x = (L.x + Qx) / 2, y = (L.y + Qy) / 2. Save the results. Then move the target point L one point to the right and determine whether L and R have reached the endpoint. If yes, proceed to step S102c-6; otherwise, proceed to step S102c-3.

[0077] Step S102 c-5: Let r = (proportional value of the current point R on the right line - proportional value of the previous point on the left line) / (proportional value of the current point L on the left line - proportional value of the previous point on the left line). Calculate the coordinates (x, y) of the point Z on the left line: x = (x-coordinate of the current point L on the left line - x-coordinate of the previous point on the left line) * r + x-coordinate of the previous point on the left line; y = (y-coordinate of the current point L on the left line - y-coordinate of the previous point on the left line) * r + y-coordinate of the previous point on the left line. Then calculate the midpoint P between R and Z and save it to the result. Then move the target point R one point to the right and determine whether L and R have reached the endpoint. If yes, proceed to S102c-6; otherwise, proceed to step S102c-3.

[0078] Step S102c-6: Calculate and save the midpoint between target points L and R, thereby obtaining all the points that constitute the center line.

[0079] Step S102c-7: Check all points on the centerline sequentially, calculate the distance between any two points, set a threshold T, and determine if the distance is less than T. If it is, delete any point. It should be noted that the start and end points of the entire polyline are not deleted; only the midpoints are deleted. This yields the generated centerline of the ordinary lane. In this embodiment, the threshold T is set to 0.00001.

[0080] Repeat the above process until all center lines for ordinary lanes are generated. See the image below for the final result. Figure 6 .

[0081] Step S103: Determine whether there is a topological bifurcation point on the center line of the ordinary lane. If there is, set the lane at the topological bifurcation point as a connecting lane.

[0082] Specifically, the topological relationships of all regular lanes are examined to determine if any topological bifurcation points exist. At locations where topological bifurcations exist, all or some regular lanes are reclassified as connecting lanes. This ensures that lane sequences formed by connecting consecutive regular lanes are topologically separate, and lane sequences are connected only through connecting lanes.

[0083] The lane sequence is determined by the coordinates of the starting and ending points of the centerline of a regular lane. The endpoint coordinates (x, y) serve as the basis for identifying topological nodes. If the endpoint coordinates are equal, they are considered to be the same topological node. All lanes traveling in the same direction and connected to each other at their respective topological nodes are connected together to form a lane sequence, which also has two preceding and following topological nodes. It is essential to ensure that no two lane sequences are topologically connected, i.e., they are separate from each other.

[0084] See Figure 7 and Figure 8 This diagram illustrates the process of dividing ordinary lanes at a topological fork into connecting lanes. If an ordinary lane at the topological fork has actual left and right side lines, and these lines are smoothly connected to the side lines of the preceding and following ordinary lanes, then the center line of this ordinary lane can be smoothed together with the center lines of the preceding and following ordinary lanes. Otherwise, the ordinary lane at the topological fork must be divided into connecting lanes, and the center lines of these connecting lanes will be generated separately later.

[0085] Additionally, it should be noted that for ordinary lanes without topological bifurcation, that is, for a certain ordinary lane segment, if its left and right side lines are obviously not smooth enough, the ordinary lane segment can be extracted and used as a connecting lane. The center line of the connecting lane can then be generated separately. Otherwise, it will lead to the problem that the overall curve of the lane center line is not smooth enough.

[0086] Step S104: Determine whether there is a circular topology in the center line of the ordinary lane. If there is, extract a segment of the arc-shaped lane from the ordinary lane that forms the circular topology as a connecting lane.

[0087] Specifically, examine the topology of all regular lanes. For roundabouts and other locations where a loop is formed topologically, break up a small section of the looping lane and designate it as a connecting lane. See also... Figure 9 .

[0088] The purpose of this approach is that many smoothing algorithms cannot handle circular point sequences, or they simply treat the circular point sequence as a regular polyline for smoothing, treating the start and midpoints of the circle as two different points. The resulting smoothed sequence either separates the start / end point (which is the same point) into two different points, or although the coordinates are equal, the curvature at that point is discontinuous. This embodiment splits the circular ordinary lane centerline into an arc-shaped ordinary lane centerline and an arc-shaped connecting lane centerline. The smoothing algorithm only needs to process the ordinary lane centerline, while the connecting lane centerline is generated separately by a subsequent connecting line generation algorithm. This effectively solves the problem of the difficulty in smoothing circular ordinary lane centerlines.

[0089] In this embodiment, the arc length of the extracted centerline of the curved lane is greater than or equal to 1m, which facilitates the subsequent generation of connecting lane centerlines and makes data quality checks easier. Lane centerlines that are too short are difficult to observe on the visualization interface.

[0090] Step S105: Perform point spacing uniformization and smoothing on the center line of the ordinary lane.

[0091] Specifically, the centerline of the ordinary lane generated by the centerline generation algorithm in step S102 has sparse and uneven points, making it unsuitable as a trajectory for autonomous vehicles. Therefore, it needs to undergo point spacing homogenization and smoothing processing. The specific processing steps are as follows:

[0092] Step S105a: Perform linear interpolation on the points along the center line of the ordinary lane to ensure a uniform distribution of the points. The target interval for linear interpolation is 0.20m to 1.00m. In this embodiment, the target interval is 0.50m. After linear interpolation, the interval between all points along the center line of the ordinary lane is approximately 0.50m. Perfect uniformity is not required; only approximate uniformity is needed. For example, the actual interval between points is 0.40m to 0.60m.

[0093] Step S105b: Treat all consecutive center lines of ordinary lanes as a continuous curve and smooth them. That is, connect all consecutive center lines of ordinary lanes together and smooth them as a whole.

[0094] Step S105c: Use the smoothed curve as the new centerline of the ordinary lane. After smoothing, the number of points will not change and will remain approximately uniform; that is, after smoothing, the point interval will still be approximately 0.50m. At this point, the curvature of all points is continuous, and the positions of all points cannot be moved. Points can only be deleted or added one by one at the two endpoints of the overall ordinary lane centerline. Internal points cannot be changed to ensure the continuity of the curvature of all points.

[0095] It should be noted that the point spacing uniformization and smoothing processes used in this embodiment are existing processing methods, and the specific processing steps will not be described in detail.

[0096] Step S106: Generate the center lines of all connecting lanes according to the connecting line generation algorithm.

[0097] Specifically, connecting lane center lines are added at locations such as turning lanes, intersections, lane splits, and merging points to connect the ordinary lane center lines processed in step S105. See also Figure 10 This is a schematic diagram illustrating the effect of generating a continuously curvature connecting lane centerline. The steps for generating the connecting lane centerline are as follows:

[0098] Step S106a: Obtain the starting point parameters and ending point parameters of the center line of the connecting lane to be generated. The starting point parameters include the first coordinates (x1, y1), the first curvature k1, and the first direction angle a1 of the starting point. The ending point parameters include the second coordinates (x2, y2), the second curvature k2, and the second direction angle a2 of the ending point.

[0099] Step S106c: Input the first coordinates (x1, y1), first curvature k1, first direction angle a1 of the starting point, and the second coordinates (x2, y2), second curvature k2, and second direction angle a2 as input parameters to the connector generation algorithm.

[0100] Step S106d: Generate the center lines of the connecting lanes using the connecting line generation algorithm, and obtain the curve model and curve parameters of the corresponding connecting lane center lines.

[0101] For example, if the curve model is a combination of two spiral curves, then the parameters are (x1, y1), (x2, y2), k1, k2, a1, a2, k3, s1, and s2. Here, (x1, y1), k1, and a1 are the coordinates of the starting point, curvature, and direction angle; (x2, y2), k2, and a2 are the coordinates of the ending point, curvature, and direction angle; k3 is the intermediate curvature; s1 is the length of the first spiral curve segment (from the starting point to the intermediate curvature); and s2 is the length of the second spiral curve segment (from the intermediate curvature to the ending curvature). After saving the curve model type and parameter values, the entire curve can be calculated using the curve equation. Given any point interval s, all points on the curve can be calculated with a step size of s.

[0102] In this embodiment, after initially generating the center line of the connecting lane, a larger point interval, such as 0.5m, is first set to generate the actual center line point sequence of the connecting lane. This sparser point sequence is easier to adjust and maintain. After confirming that there are no errors, a smaller point interval is set to facilitate use by autonomous vehicles.

[0103] Step S107: Generate all missing edge lines of the connecting lanes based on the center line of the connecting lane.

[0104] Specifically, while the center lines of connecting lanes at some intersections, lane splits, and merging points are complete, some connecting lanes lack left and right side lines. (See [link / reference]). Figure 11 Therefore, virtual lane lines need to be added. The method for generating virtual lane lines is to automatically generate the left or right lane line based on the lane centerline using a lane line generation algorithm.

[0105] It should be noted that the edge generation algorithm is also an existing algorithm, and will not be described in detail in this embodiment.

[0106] Step S108: Perform nonlinear interpolation on all points in the center lines of ordinary lanes to make the point density in the center lines of ordinary lanes reach the set value. Resample the points in the corresponding center lines of connecting lanes according to the curve model and curve parameters of the connecting lane center lines to make the point density in the connecting lane center lines reach the set value.

[0107] Specifically, after step S107, the edge lines and center lines of all ordinary lanes and connecting lanes have been generated, and all center line points are curvature continuous with a point spacing of 0.5m. For autonomous vehicles, this point spacing is relatively sparse, and the point density needs to be increased. The processing method is as follows:

[0108] Step S108a: Perform nonlinear interpolation on the points along the center line of the ordinary lane. For each point along the center line, obtain the coordinates of several adjacent points before and after it. According to L'Hôpital's rule, the curvature k, the derivative of curvature dk, and the direction angle α of each point can be calculated relatively accurately. Then, for two adjacent points, based on the point coordinates, curvature k, derivative of curvature dk, and direction angle α, use nonlinear interpolation algorithms such as spline interpolation to interpolate between the two points with a point interval of 0.05m to 0.20m. The resulting point sequence still maintains very high curvature continuity. In this embodiment, the target interval for nonlinear interpolation is 0.10m.

[0109] Step S108b: For the center line of the connecting lane, use the curve model and curve parameters obtained in step S106 to simply resample the points on the center line of the connecting lane, so that the point density of the center line of the connecting lane is increased to 0.10m.

[0110] See the final lane diagram after step S108. Figure 12 .

[0111] Therefore, the above method analyzes the entire map, divides the lanes into ordinary lanes and connecting lanes, processes them separately, and at the locations where ordinary lanes undergo topological bifurcation, such as lane splitting or merging, the ordinary lanes at the topological bifurcation points are classified as connecting lanes. In the case of roundabouts and other situations where ordinary lanes form a circular topology, any segment of ordinary lanes in the circular lane is separated and classified as a connecting lane, effectively solving the problem of topological bifurcation points and the difficulty in smoothing the center line of the circular lane.

[0112] This embodiment also provides a map lane data generation device, see [link to documentation]. Figure 13 This is a schematic diagram of the device. The map lane data generation device includes:

[0113] The first data acquisition module is used to collect lane layout information and divide the lanes into regular lanes and connecting lanes based on the lane layout information. In this embodiment, the first data acquisition module includes a map acquisition vehicle and acquisition sensors mounted on the map acquisition vehicle. The acquisition sensors include, but are not limited to, positioning devices, cameras, and LiDAR.

[0114] The first generation module is used to generate the edge lines of all ordinary lanes and generate the center lines of ordinary lanes based on the edge lines of ordinary lanes.

[0115] The first processing module is used to perform point homogenization and smoothing on the center line of ordinary lanes.

[0116] The first judgment module is used to determine whether there is a topological bifurcation point on the center line of the ordinary lane, and if so, to set the lane at the topological bifurcation point as a connecting lane.

[0117] The second judgment module is used to determine whether a circular topology exists in the centerline of the ordinary lane, and if so, to extract an arc-shaped section of the ordinary lane forming the circular topology as a connecting lane. In this embodiment, the arc length of the centerline of the extracted arc-shaped lane is greater than or equal to 1m.

[0118] The second generation module is used by the connector generation algorithm to generate the center lines of all connecting lanes.

[0119] The third generation module is used to generate all missing edge lines for all connecting lanes based on the center line of the connecting lane.

[0120] The second processing module performs nonlinear interpolation on all points in the center lines of the ordinary lanes to achieve a set point density. It then resamples the points on the corresponding connecting lane center lines using the curve model and curve parameters to achieve the same set point density. In this embodiment, the set point density in the lane center lines is 0.10m.

[0121] In addition to the above embodiments, the present invention may have other implementation methods; all technical solutions formed by equivalent substitution or equivalent transformation fall within the protection scope claimed by the present invention.

Claims

1. A method for generating map lane data, characterized in that: include, Collect lane layout information and divide the lanes into ordinary lanes and connecting lanes based on the lane layout information; Draw the edge lines of all regular lanes, and generate the center lines of the regular lanes based on the edge lines of the regular lanes; Determine whether there is a topological bifurcation point on the center line of the ordinary lane. If there is, set the lane at the topological bifurcation point as a connecting lane. The location of the topological bifurcation point involves lane splitting or merging. Determine whether the centerline of the ordinary lane has a circular topology. If it does, then extract an arc-shaped lane from the ordinary lane that forms the circular topology as a connecting lane. The center line of the ordinary lane is subjected to point spacing uniformization and smoothing. Generate the center lines of all connecting lanes based on the connector generation algorithm; Generate all missing edge lines for the connecting lanes based on the center line of the connecting lanes.

2. The map lane data generation method according to claim 1, characterized in that: The process of drawing the edge lines of all ordinary lanes and generating the center lines of the ordinary lanes based on the edge lines includes: Obtain the point sequence of the two side lines of the ordinary lane, and obtain the coordinates of each point in the point sequence in sequence along the direction of the side lines; Calculate the proportion of each point on both sides of the ordinary lane relative to its current position on the lane. Along the direction of the two side lines, starting from the starting point of the two side lines, points are taken sequentially. Each time a set of target points is taken, the midpoint between the two target points is calculated as a point on the center line, and all the points that make up the center line are obtained, thus generating the center line of the ordinary lane.

3. The map lane data generation method according to claim 1, characterized in that: The arc length of the centerline of the intercepted arc-shaped lane is greater than or equal to 1m.

4. The map lane data generation method according to claim 1, characterized in that: The step of uniformizing the point intervals and smoothing the center line of the ordinary lane includes... Linear interpolation is performed between the points in the center line of the ordinary lane to make the points in the center line of the ordinary lane evenly distributed. Treat all the center lines of the aforementioned ordinary lanes that are connected one after the other as a continuous curve and smooth them out. The smoothed curve is used as the center line of the ordinary lane.

5. The map lane data generation method according to claim 1, characterized in that: The process of generating centerlines for all connecting lanes according to the connector generation algorithm includes... Obtain the starting point parameters and ending point parameters of the center line of the connecting lane to be generated. The starting point parameters include the first coordinates, first curvature, and first direction angle of the starting point, and the ending point parameters include the second coordinates, second curvature, and second direction angle of the ending point. The first coordinates, first curvature, and first direction angle of the starting point, and the second coordinates, second curvature, and second direction angle of the ending point are used as input parameters and input into the connector generation algorithm. The connecting line generation algorithm generates the center line of the connecting lane, and obtains the curve model and curve parameters corresponding to the center line of the connecting lane.

6. The map lane data generation method according to claim 5, characterized in that: After generating all the missing edge lines of the connecting lanes based on the center line of the connecting lanes, the process further includes: Nonlinear interpolation is performed on all points in the center lines of the ordinary lanes to make the point density in the center lines of the ordinary lanes reach a set value. Then, the points corresponding to the center lines of the connecting lanes are resampled according to the curve model and curve parameters of the connecting lanes to make the point density in the center lines of the connecting lanes reach a set value.

7. A map lane data generation device, characterized in that: include, The first acquisition module is used to acquire lane layout information and divide the lanes into ordinary lanes and connecting lanes based on the lane layout information. The first generation module is used to generate the edge lines of all the ordinary lanes and generate the center lines of the ordinary lanes based on the edge lines of the ordinary lanes. The first processing module is used to perform point homogenization processing and smoothing processing on the center line of the ordinary lane. The first judgment module is used to determine whether there is a topological bifurcation point on the center line of the ordinary lane, and if there is, to set the lane at the topological bifurcation point as a connecting lane; the location of the topological bifurcation point is where there is lane splitting or merging. The second judgment module is used to determine whether there is a circular topology in the center line of the ordinary lane, and if there is, to cut out a section of the arc lane in the ordinary lane that forms the circular topology as a connecting lane. The second generation module is used by the connector generation algorithm to generate the center lines of all connecting lanes; The third generation module is used to generate all the missing edge lines of the connecting lanes based on the center line of the connecting lane; The second processing module is used to perform nonlinear interpolation on all the points in the center lines of the ordinary lanes to make the point density in the center lines of the ordinary lanes reach a set value, and to resample the points corresponding to the center lines of the connecting lanes according to the curve model and curve parameters of the connecting lanes to make the point density in the center lines of the connecting lanes reach a set value.

8. The map lane data generation device according to claim 7, characterized in that: The arc length of the centerline of the intercepted arc-shaped lane is greater than or equal to 1m.

Images