Map generation method, medium, product and electronic device

Abstract

The present disclosure relates to a map generation method, medium, product and electronic device. The method comprises: obtaining SD map data and road collection data collected when a vehicle is traveling on a road; marking elements in the SD map according to element information in the SD map data to generate a reference map; and correcting the position of the elements marked in the reference map according to the road collection data to obtain a target map. In this way, based on the SD map data, correction is made with reference to the road collection data, thereby efficiently generating a lightweight high-definition map.

Description

Technical Field

[0001] The present disclosure relates to the technical field of electronic map generation, and in particular, to a map generation method, medium, product and electronic device. Background Art

[0002] As an indispensable part of autonomous driving, high-precision maps play a very important role in path planning and vehicle positioning. However, high-precision maps have the characteristics of high element accuracy, high richness, high production cost, long update cycle, and inconsistency with standard definition (SD) road data. In order to reduce the dependence on high-precision maps in autonomous driving, the concept of lightweighting high-precision maps has been proposed. Summary of the invention

[0003] The purpose of the present disclosure is to provide a map generation method, medium, product and electronic device that can efficiently generate lightweight high-definition maps.

[0004] In order to achieve the above object, the present disclosure provides a map generation method, the method comprising:

[0005] Acquire SD map data and road data collected when the vehicle is traveling on the road;

[0006] Marking elements in the SD map according to the element information in the SD map data to generate a reference map;

[0007] The positions of the elements marked in the reference map are corrected according to the road collection data to obtain a target map.

[0008] Optionally, the step of correcting the positions of the elements marked in the reference map according to the road collection data to obtain the target map includes:

[0009] Extracting elements in the reference map as a first type of element set according to a mapping relationship between the road data in the SD map data and the trajectory data in the road collection data;

[0010] Determining a relative positional relationship between elements in the first type of element set as a first positional relationship;

[0011] Using a machine learning method, extracting elements in the reference map according to the road collection data as a second type of element set;

[0012] Determining a relative positional relationship between elements in the second type of element set as a second positional relationship;

[0013] Determine a deviation value of each element in the first type of element set according to the first position relationship and the second position relationship;

[0014] According to the deviation value, the position of each element in the first type of element set is corrected to obtain the target map.

[0015] Optionally, extracting elements in the reference map according to the mapping relationship between the road data in the SD map data and the trajectory data in the road collection data as a first type of element set includes:

[0016] Determine a mapping relationship between the road data in the SD map data and the trajectory data in the road collection data as a first mapping relationship;

[0017] Determining a positional relationship between the road data in the SD map data and the elements marked in the reference map as a second mapping relationship;

[0018] According to the first mapping relationship and the second mapping relationship, elements in the reference map are extracted as the first type of element set.

[0019] Optionally, determining a mapping relationship between the road data in the SD map data and the trajectory data in the road collection data includes:

[0020] For each route in the SD map data, determining the minimum value of the distance between the trajectory point in the road collection data and the center point of the route as the distance corresponding to the route;

[0021] According to the minimum value, projecting the route onto a trajectory formed by the road collection data, and determining a trajectory point in the road collection data corresponding to the route;

[0022] The angle between the vector formed by the starting point and the end point of the trajectory point corresponding to the route in the road collection data and the vector formed by the starting point and the end point of the route is determined as the angle corresponding to the route.

[0023] Optionally, determining the relative positional relationship between the elements in the first-category element set as the first positional relationship includes: for each element in the first-category element set, determining a set of multiple vectors with the element as a starting point and multiple other elements in the first-category element set as end points, as the first positional relationship;

[0024] The determining of the relative positional relationship between the elements in the second category of element set as the second positional relationship includes: for each element in the second category of element set, determining a set of multiple vectors with the element as the starting point and other multiple elements in the second category of element set as the end points, as the second positional relationship.

[0025] Optionally, determining the deviation value of each element in the first type of element set according to the first position relationship and the second position relationship includes:

[0026] For each element in the first element set, searching the second element set for an element whose distance to the element is less than a predetermined distance threshold;

[0027] Moving the element to the position of the found element;

[0028] Matching the first position relationship with the second position relationship;

[0029] If the match is successful, the deviation angle of the element relative to the found element is determined, wherein the deviation value includes a vector formed by the movement of the element and the deviation angle.

[0030] Optionally, matching the first position relationship with the second position relationship includes:

[0031] Match multiple vectors with the first element in the first type of element set as the starting point with multiple vectors with the element in the second type of element set as the starting point, wherein, among the multiple vectors with the first element as the starting point, if the proportion of successfully matched vectors is greater than a first proportion threshold, then the first position relationship and the second position relationship are successfully matched.

[0032] Optionally, if the proportion of the first vector falling within a predetermined range around the second vector is greater than a second proportion threshold, the first vector and the second vector are matched successfully.

[0033] Optionally, determining the deviation angle of the element relative to the found element includes:

[0034] Determine the angle between two successfully matched vectors;

[0035] An average value of angles corresponding to successfully matched vectors among a plurality of vectors starting from the first element is taken as a deviation angle of the first element relative to the found element.

[0036] Optionally, the elements include ground turn arrows and lane lines, and the method further includes:

[0037] A virtual lane at an intersection is generated in the target map according to the ground turn arrows and lane lines in the target map.

[0038] Optionally, generating a virtual lane at an intersection in the target map according to ground turn arrows and lane lines in the target map includes:

[0039] When a vehicle enters an incoming road straight after exiting an exit road, and the number of lanes of the exit road is different from the number of lanes of the incoming road and both are greater than 1, for each exit lane of the exit road, an incoming lane of the incoming road whose overlap with the exit lane is greater than an overlap threshold is connected to the exit lane to generate a virtual lane;

[0040] Among them, the overlap degree is the ratio of the intersection stop line at the starting end of the entry lane that is intercepted by the lane extension line of the exit lane and included in the extension line to the intersection stop line, the exit road is the road that the vehicle enters the intersection after exiting, and the entry road is the road that the vehicle enters after exiting the intersection.

[0041] Optionally, generating a virtual lane at an intersection in the target map according to ground turn arrows and lane lines in the target map includes:

[0042] When a vehicle turns from an exit road to an entry road, and the number of lanes of the exit road is different from the number of lanes of the entry road and both are greater than 1, N consecutive lanes in the lanes of the multi-lane road are divided into a group, and the groups are connected with the lanes in the road with fewer lanes to generate virtual lanes;

[0043] Among them, the multi-lane road is the one with more lanes between the exit road and the entrance road, the few-lane road is the one with fewer lanes between the exit road and the entrance road, the number of lanes in the few-lane road is greater than or equal to N, the exit road is the road that the vehicle enters after exiting the intersection, and the entrance road is the road that the vehicle enters after exiting the intersection.

[0044] The present disclosure also provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the steps of the above method provided by the present disclosure are implemented.

[0045] The present disclosure also provides a computer program product, including a computer program, which implements the steps of the above method provided by the present disclosure when the computer program is executed by a processor.

[0046] The present disclosure also provides an electronic device, the electronic device comprising:

[0047] processor;

[0048] a memory for storing processor-executable instructions;

[0049] Among them, the processor is configured to execute the steps of the above-mentioned map generation method provided by the present disclosure.

[0050] Through the above technical solution, the elements in the SD map are marked according to the element information in the SD map data to generate a reference map. The positions of the elements marked in the reference map are corrected according to the road collection data to obtain the target map. In this way, based on the SD map data, the correction is made with reference to the road collection data, so as to efficiently generate a lightweight high-definition map.

[0051] Other features and advantages of the present disclosure will be described in detail in the following detailed description. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] The accompanying drawings are used to provide a further understanding of the present disclosure and constitute a part of the specification. Together with the following specific embodiments, they are used to explain the present disclosure but do not constitute a limitation of the present disclosure. In the accompanying drawings:

[0053] Figure 1 is a flowchart of a map generating method provided by an exemplary embodiment.

[0054] Figure 2 It is a schematic diagram of roads in SD map data and trajectories in road collection data provided by an exemplary embodiment.

[0055] Figure 3 It is a schematic diagram of multiple vectors starting from one element provided by an exemplary embodiment.

[0056] Figure 4 is a schematic diagram of vector matching provided by an exemplary embodiment.

[0057] Figure 5 It is a schematic diagram of determining lane overlap provided by an exemplary embodiment. DETAILED DESCRIPTION

[0058] The specific implementation of the present disclosure is described in detail below in conjunction with the accompanying drawings. It should be understood that the specific implementation described herein is only used to illustrate and explain the present disclosure, and is not used to limit the present disclosure.

[0059] Figure 1 FIG. 1 is a flowchart of a map generation method provided by an exemplary embodiment. Figure 1 As shown, the method includes the following steps.

[0060] Step 101, obtaining SD map data and road data collected when the vehicle is traveling on the road.

[0061] Step 102, marking elements in the SD map according to the element information in the SD map data to generate a reference map.

[0062] Step 103, correcting the positions of the elements marked in the reference map according to the road collection data to obtain a target map.

[0063] The SD map data may be existing map data, including track data and element information, etc. The element information may include element types, parameters, etc. The element types may include, for example, ground turn arrows, lane lines, intersections, traffic lights, etc. The parameters may include, for example, the position of ground turn arrows, the number of lanes, and the width of lanes, etc.

[0064] Road collection data refers to data collected during the actual driving process of the vehicle, which may include trajectory data, images obtained by the camera, obstacle distances detected by radar, etc.

[0065] The road collection data can be preprocessed. The inertial navigation data, point cloud, and image data collected by the vehicle can be solved by IE (Inertial Explorer), that is, using the principle of post-processing difference, combined with the Global Navigation Satellite System (GNSS) data and Inertial Measurement Unit (IMU) data of the mobile measurement equipment, and the static data of the GNSS base station in the same time period, the corresponding matching processing algorithm is used to improve the trajectory accuracy. The solved data is adjusted to ensure the accuracy of the collected trajectory. The least squares method, Kalman filter and other adjustment algorithms can be used.

[0066] After obtaining the adjustment trajectory, the trajectory can be segmented. For example, the data can be segmented, processed, and merged based on the grids defined by the internal production. The following division principle can be used: the global range (excluding the North and South Poles) is divided into four parts, and the division is repeated multiple times based on the previous step. Each time it is repeated, the grid level is increased by one level to determine the operation area or the area of ​​interest.

[0067] The road data can be denoised in advance. First, the non-operating range of the data is removed at a fixed distance based on the trajectory, and then the dynamic objects are removed. Dynamic object removal can be done by, for example, using the inter-frame difference method, background removal method, motion blur method, etc. In addition, point cloud fusion can also be performed.

[0068] Taking the ground turning arrow as an example, the method of marking the element in the SD map according to the element information in the SD map data may include:

[0069] In the SD map data, the lane-related information in the road table is obtained, and the ground turning arrow information of the current lane is extracted; after obtaining the ground turning arrow information, multiple ground turning arrows are virtually generated according to the national standard at the end of the current road, and the ground turning arrows are arranged according to the order of the associated lanes; the road width and the number of lanes are obtained, and the lanes of the entire road can be divided according to the principle of average distribution, and then the ground turning arrows are moved to the center of the corresponding lane. At this time, the set of element information in the SD map data is A0 = [a1, a2, ..., am], where a1 ~ am, for example, represents the 1st to mth ground turning arrows.

[0070] After the marking is completed, the relative position M1 between the ground turn arrow and the track data in the SD map data may be calculated and recorded. The road data in the SD map data may include the positions of multiple road segments (links).

[0071] Through the above technical solution, the elements in the SD map are marked according to the element information in the SD map data to generate a reference map. The positions of the elements marked in the reference map are corrected according to the road collection data to obtain the target map. In this way, based on the SD map data, the correction is made with reference to the road collection data, so as to efficiently generate a lightweight high-definition map.

[0072] In yet another embodiment, the position of the element marked in the reference map is corrected according to the road collection data to obtain the target map, including:

[0073] According to the mapping relationship between the road data in the SD map data and the trajectory data in the road collection data, the elements in the reference map are extracted as the first type of element set;

[0074] Determine the relative positional relationship between the elements in the first type of element set as the first positional relationship;

[0075] Using machine learning methods, the elements in the reference map are extracted according to the road collection data as the second type of feature set;

[0076] Determine the relative positional relationship between the elements in the second type of element set as the second positional relationship;

[0077] Determine a deviation value of each element in the first type element set according to the first position relationship and the second position relationship;

[0078] According to the deviation value, the position of each element in the first type of element set is corrected to obtain the target map.

[0079] The element type in the embodiment of this article is described by taking a ground turning arrow as an example.

[0080] The first and second feature sets are both collections of features in the reference map. The first and second feature sets use different methods to extract features (feature locations). The features can be roughly extracted using machine learning + model recognition to obtain the second feature set A2. For example, model training (open source algorithm) plus hidden Markov model algorithm, Bayesian classification algorithm, etc. can be used.

[0081] The relative position relationship includes the distance and orientation relationship between elements. The position of elements obtained by different methods may be different, and thus the position relationship between elements may also be different. Compared with the deviation value obtained by comparing the elements individually, the deviation value of each element in the first type of element set is determined based on the difference in the position relationship between the two element sets, which has better integrity and more accurate comparison results.

[0082] In another embodiment, according to the mapping relationship between the road data in the SD map data and the trajectory data in the road collection data, the elements in the reference map are extracted as the first type of element set, including:

[0083] Determine a mapping relationship between the road data in the SD map data and the trajectory data in the road collection data as a first mapping relationship;

[0084] Determining a positional relationship between road data in the SD map data and elements marked in the reference map as a second mapping relationship;

[0085] According to the first mapping relationship and the second mapping relationship, elements in the reference map are extracted as a first type of element set.

[0086] Among them, the mapping relationship between the road data in the SD map data and the trajectory data in the road collection data can be the corresponding relationship between each point on the road in the SD map data and each point on the trajectory in the road collection data.

[0087] The positional relationship between the road data in the SD map data and the elements marked in the reference map is the relative position between the elements marked in the reference map and the links in the SD map.

[0088] By combining the first mapping relationship with the second mapping relationship, that is, taking the road data in the SD map data as the intermediate quantity, the relative position between the elements marked in the reference map and the trajectory data in the road collection data can be obtained, thereby extracting the first type of element set.

[0089] In this embodiment, the road data in the SD map data is used as the intermediate quantity, and elements are extracted according to the trajectory data in the road collection data to obtain a first type of element set.

[0090] In yet another embodiment, determining a mapping relationship between road data in the SD map data and trajectory data in the road collection data includes:

[0091] For each route in the SD map data, determine the minimum value of the distance between the trajectory point in the road collection data and the center point of the route as the distance corresponding to the route;

[0092] According to the minimum value, the route is projected onto the trajectory formed by the road collection data, and the trajectory point corresponding to the route in the road collection data is determined;

[0093] The angle between the vector formed by the starting point and the end point of the trajectory point corresponding to the route in the road collection data and the vector formed by the starting point and the end point of the route is determined as the angle corresponding to the route.

[0094] The SD map data includes multiple routes, that is, multiple links. Figure 2 FIG. 1 is a schematic diagram of a road in SD map data and a track in road collection data provided by an exemplary embodiment. Figure 2 As shown, the track in the road collection data is shown as E1, and the road in the SD map data is shown as E2. E2 includes the first segment route L1, the second segment route L2 and the third segment route L3 connected. The starting point of the second segment route L2 is c1 and the end point is c2.

[0095] Taking the second section of the route L2 as an example, the distance between each trajectory point in the trajectory E1 in the road collection data and the center point c3 of the second section of the route L2 is calculated respectively. The minimum value d0 of these distances is the distance corresponding to the second section of the route L2, that is, the distance between the second section of the route L2 and the corresponding part of the trajectory E1 in the road collection data.

[0096] The method of projecting the second section of route L2 onto the track E1 in the road collection data may be, for example, to find the track point d3 in the track E1 that is at the minimum distance d0 from the center point c3. Extend from point c1 along the direction from c3 to d3, and the intersection with E1 is d1. Extend from point c2 along the direction from c3 to d3, and the intersection with E1 is d2. Then the part of the second section of route L2 projected onto the track E1 in the road collection data is the track point from d1 to d2 in E1. The track point in the road collection data corresponding to the second section of route L2 is the track point from d1 to d2 in E1.

[0097] The angle between the vector formed by the starting point d1 and the end point d2 of the trajectory point corresponding to the second segment route L2 in E1 and the vector formed by the starting point c1 and the end point c2 of the second segment route L2 is determined by translating one of the vectors. Figure 2In , the vector formed by d1 and d2 is translated, and the angle between the two vectors is α, which is the angle corresponding to the second route L2.

[0098] If α is greater than an angle threshold (for example, 45°), it can be considered that E1 and E2 are traveling in opposite directions, which can be considered a matching error and the route segment needs to be remapped.

[0099] The parameters in the mapping relationship (matching relationship) between the road data in the SD map data and the trajectory data in the road collection data may include the distance, trajectory point and angle corresponding to each route segment in the SD map data. In this embodiment, the mapping relationship is established by fewer parameters, the data processing amount is small, and the accuracy is high.

[0100] In yet another embodiment, determining the relative positional relationship between elements in the first type element set as the first positional relationship includes: for each element in the first type element set, determining a set of multiple vectors having the element as a starting point and other multiple elements in the first type element set as end points as the first positional relationship;

[0101] Determining the relative positional relationship between the elements in the second category element set as the second positional relationship includes: for each element in the second category element set, determining a set of multiple vectors with the element as a starting point and other multiple elements in the second category element set as end points as the second positional relationship.

[0102] Taking the ground turning arrow as an example, the center point of the circumscribed rectangle of a ground turning arrow is used as the center of the circle to query the same or different types of elements within a predetermined radius (for example, 50m). After the query is completed, the positional relationship between the coordinates of the elements is recorded, which can be expressed in the form of vectors, that is, the vector from the center of the circle to the queried point. Then repeat the above steps for the next ground turning arrow, traverse all the elements within the processing range, and form the final vectorized grid.

[0103] Figure 3 It is a schematic diagram of multiple vectors starting from one element provided by an exemplary embodiment. Figure 3 In the , graphics of the same shape represent the same type of features. For example, a diamond represents a ground turn arrow, an ellipse represents a traffic light, and a triangle represents a sidewalk. A ground turn arrow is used as the starting point, pointing to the six surrounding features, forming six vectors. Figure 3 In the case where one element is used as the starting point, each element in the first type of element set is traversed to obtain the vectorized grid of the first type of element set, that is, the first position relationship. Each element in the second type of element set is traversed to obtain the vectorized grid of the second type of element set, that is, the second position relationship.

[0104] Since the first positional relationship and the second positional relationship need to be matched later, in order to reduce the matching error, other elements can be queried in various directions around as much as possible when the vectorized grid is generated. For example, the area around the element is evenly divided into sectors with the same angle, and the same number of elements are queried in each sector. In some simple scenarios (such as highways), the driving direction is relatively certain, and the vectors in the longitudinal and lateral directions of the vehicle can be searched during matching.

[0105] In this embodiment, the positional relationship is implemented using a vector grid, which makes it easier to determine the deviation value.

[0106] In yet another embodiment, determining the deviation value of each element in the first type element set according to the first position relationship and the second position relationship includes:

[0107] For each feature in the first feature set, searching for a feature in the second feature set whose distance to the feature is less than a predetermined distance threshold (e.g., 5 m);

[0108] Move the feature to the location of the found feature;

[0109] matching the first position relationship with the second position relationship;

[0110] If the match is successful, the deviation angle of the feature relative to the found feature is determined, wherein the deviation value includes the vector formed by the movement of the feature and the deviation angle.

[0111] Among them, searching for elements whose distance is less than a predetermined distance threshold in the second category of element set is a preliminary screening of elements in the second category of element set that match the current elements in the first category of element set. Elements of the same type as the current elements in the first category of element set can be screened in the second category of element set.

[0112] Move the current element in the first element set to the position of the element in the second element set found, try to match the first position relationship and the second position relationship, if the match is successful, it can be considered that the two elements are corresponding elements. In addition, the deviation value of an element in the first element set indicates how it can be obtained based on the element in the second element set that has been matched successfully. The vector formed by the element movement indicates that the element in the first element set needs to be moved to reach the end position of the vector. The deviation angle indicates that the element in the first element set also needs to be rotated by the deviation angle.

[0113] In this embodiment, the two parameters of the vector and the deviation angle formed by the movement of the element can determine the deviation value of each element in the first type of element set, so that the position of each element in the first type of element set can be corrected according to the deviation value with high accuracy.

[0114] In yet another embodiment, matching the first position relationship and the second position relationship includes:

[0115] Multiple vectors with the first element in the first category of element set as the starting point are matched with multiple vectors with the element in the found second category of element set as the starting point. Among the multiple vectors with the first element as the starting point, if the proportion of successfully matched vectors is greater than a first proportion threshold, the first position relationship and the second position relationship are matched successfully.

[0116] The first element can be any element in the first type of element set. Figure 3 In the embodiment of the invention, the first element in the first type of element set is used as the starting point, and six vectors are formed around it. The first proportion threshold is 0.6. If four of the six vectors are successfully matched, the proportion of the successfully matched vectors is 0.66, which is greater than the first proportion threshold 0.6. Figure 3 The first positional relationship of the first element in the first element set and the second positional relationship of the corresponding element in the second element set are matched successfully. Next, the elements in the second element set that are matched successfully can be used to determine the deviation value of the first element in the first element set.

[0117] In this embodiment, whether the first position relationship and the second position relationship are matched successfully is determined based on the proportion of the number of successfully matched vectors. When the match is successful, the deviation value is determined, providing a method for determining whether the match is successful, which is simple, easy to implement and highly reliable.

[0118] In yet another embodiment, if the proportion of the first vector falling within a predetermined range around the second vector is greater than a second proportion threshold, the first vector and the second vector are matched successfully.

[0119] The first vector is a vector to be matched in one of the first element set and the second element set, and the second vector is a vector to be matched in the other of the first element set and the second element set. A predetermined range can be defined around the second vector, for example, an area where the distance to the line segment of the vector is less than a predetermined distance. The predetermined range is a rectangle.

[0120] Figure 4 FIG. 1 is a schematic diagram of vector matching provided by an exemplary embodiment. Figure 4 As shown, the dotted rectangular box is a predetermined range around the second vector. The starting point of the first vector is O and the end point is P. The proportion of the first vector falling into the dotted rectangular box is the ratio of the length of OP1 to the length of OP. If the ratio is greater than the second proportion threshold, it can be considered that the first vector and the second vector are matched successfully.

[0121] In this embodiment, the similarity of two vectors is determined according to the proportion of the vectors falling within a predetermined range around the other party, and then whether the match is successful is determined. The method is simple and the processing speed is fast.

[0122] In yet another embodiment, determining the deviation angle of the element relative to the found element includes:

[0123] Determine the angle between two successfully matched vectors;

[0124] The average value of the included angles corresponding to the successfully matched vectors among the multiple vectors with the first element as the starting point is taken as the deviation angle of the first element relative to the found element.

[0125] There are multiple vectors starting from the first element. The successfully matched vectors are selected, and the average value of the angles corresponding to the successfully matched vectors is calculated as the deviation angle that the first element needs to rotate. For example, among the six vectors starting from an element in the first category of element set, four vectors are successfully matched, and the angles between these four vectors and the corresponding successfully matched elements are 10°, 15°, 12° and 20° respectively. Among them, clockwise or counterclockwise is used as the positive direction of the angle. The average value of these four angles (10+15+12+20) / 4=14.25° is used as the deviation angle of the element in the first category of element set relative to the found element.

[0126] In this embodiment, the deviation angle is determined according to the angle averaging method, and the angle errors are averaged. When the deviation angle obtained in this way is used for correction, the correction result is more accurate.

[0127] After the ground turn arrow is corrected according to the above embodiment, the accurate position of the ground turn arrow is obtained. Since the ground turn arrow is located in the middle of the lane, the road width and the number of lanes in the SD road collection data can be batch processed to correct the marked lane line position, thereby obtaining a more accurate lane line position.

[0128] Through the above methods, we can further build a corresponding relationship diagram between different scenarios (such as lane type, road type, etc.) and deviation values ​​to support the generation of high-precision maps in future scenarios without data (no road collection data). For example, establish a corresponding relationship between deflection values ​​and road types and lane types, and then support the generation of high-precision data without collected data.

[0129] In yet another embodiment, the elements include ground turn arrows and lane markings. The method further includes:

[0130] According to the ground turn arrows and lane lines in the target map, a virtual lane at the intersection is generated in the target map.

[0131] That is, after marking the lane lines and ground turn arrows on the map, the lane lines can be connected at the intersection to generate a virtual lane.

[0132] Determine the number of lanes of the exit road and the entrance road; when the number of exit lanes is equal to the number of entrance lanes, a smooth connection can be directly made based on the lane endpoints to make the exit road and the entrance road correspond one to one; if the number of either the exit lane or the entrance lane is equal to 1, all lanes of the other are connected to the lane with the number equal to 1.

[0133] In yet another embodiment, generating a virtual lane at an intersection in a target map according to ground turn arrows and lane lines in a target map includes:

[0134] When a vehicle exits from an exit road and goes straight into an incoming road, and the number of lanes on the exit road is different from the number of lanes on the incoming road and both are greater than 1, for each exit lane on the exit road, an incoming lane whose overlap with the exit lane is greater than an overlap threshold is connected to the exit lane to generate a virtual lane;

[0135] Among them, the overlap is the ratio of the part of the intersection stop line at the starting end of the entry lane that is intercepted by the lane extension line of the exit lane and included in the extension line to the intersection stop line. The exit road is the road that the vehicle enters the intersection after exiting, and the entry road is the road that the vehicle enters after exiting the intersection.

[0136] If the angle between the exit road and the entrance road is less than a predetermined angle threshold (eg, 45°), it is considered that the vehicle exits the exit road and then goes straight into the entrance road.

[0137] Figure 5 FIG. 1 is a schematic diagram of determining lane overlap provided by an exemplary embodiment. Figure 5 As shown, the current lane is the exit lane F1, which is on the exit road, and the dotted line is the lane extension line of the exit lane F1. F2 is the entry lane opposite the intersection. Q1Q2 is the intersection stop line on the entry road. Q1Q3 is the intersection stop line at the starting end of the entry lane, which is intercepted by the lane extension line of the exit lane F1 into two sections, Q1Q4 and Q4Q3, where Q1Q4 is included in the extension line. The ratio of Q1Q4 to the intersection stop line Q1Q3 is the overlap between the exit lane F1 and the entry lane F2. If the overlap is greater than the overlap threshold (for example, 0.6), the exit lane F1 and the entry lane F2 are connected.

[0138] In this embodiment, when a vehicle passes through an intersection in a straight line, lane connection is performed according to the overlap between the exit lane and the entrance lane.

[0139] In yet another embodiment, generating a virtual lane at an intersection in a target map according to ground turn arrows and lane lines in a target map includes:

[0140] When a vehicle turns from an exit road to an entry road, and the number of lanes on the exit road is different from that on the entry road and both are greater than 1, the N consecutive lanes in the lanes on the multi-lane road are divided into a group, and the groups are connected with the lanes on the road with fewer lanes to generate virtual lanes;

[0141] Among them, the multi-lane road is the one with more lanes between the exit road and the entrance road, the few-lane road is the one with fewer lanes between the exit road and the entrance road, the number of lanes in the few-lane road is greater than or equal to N, the exit road is the road that the vehicle enters after exiting the intersection, and the entrance road is the road that the vehicle enters after exiting the intersection.

[0142] If the angle between the exit road and the entrance road is greater than or equal to a predetermined angle threshold (eg, 45°), it is considered that the vehicle exits the exit road and then turns into the entrance road.

[0143] For example, the number of lanes on the exit road is 5, and the number of lanes on the entry road is 2. The exit lanes are connected from the left to the right or from the right to the left in groups of 2 lanes. That is, the exit lanes are connected based on 4 lane groups, namely, lane 1 and lane 2, lane 2 and lane 3, lane 3 and lane 4, lane 4 and lane 5. The 4 lane groups are connected to 2 entry lanes respectively. That is, lane 1 and lane 2 of the exit lane are connected to lane 1 and lane 2 of the entry lane respectively; lane 2 and lane 3 of the exit lane are connected to lane 1 and lane 2 of the entry lane respectively; lane 3 and lane 4 of the exit lane are connected to lane 1 and lane 2 of the entry lane respectively; lane 4 and lane 5 of the exit lane are connected to lane 1 and lane 2 of the entry lane respectively.

[0144] In this embodiment, N consecutive lanes on one side of the intersection with a larger number of lanes are divided into a group, and the groups are connected to the lanes on the other side of the intersection with a smaller number of lanes, so that they are planned separately and have high driving efficiency.

[0145] Based on the same inventive concept, the present disclosure also provides a map generating device, which includes an acquiring module, a marking module and a correcting module.

[0146] The acquisition module is used to acquire SD map data and road data collected when the vehicle is traveling on the road.

[0147] The marking module is used to mark features in the SD map according to feature information in the SD map data and generate a reference map.

[0148] The correction module is used to correct the positions of the elements marked in the reference map according to the road collection data to obtain the target map.

[0149] Optionally, the correction module is used to:

[0150] According to the mapping relationship between the road data in the SD map data and the trajectory data in the road collection data, the elements in the reference map are extracted as the first type of element set;

[0151] Determine the relative positional relationship between the elements in the first type of element set as the first positional relationship;

[0152] Using machine learning methods, the elements in the reference map are extracted according to the road collection data as the second type of feature set;

[0153] Determine the relative positional relationship between the elements in the second type of element set as the second positional relationship;

[0154] Determine a deviation value of each element in the first type element set according to the first position relationship and the second position relationship;

[0155] According to the deviation value, the position of each element in the first type of element set is corrected to obtain the target map.

[0156] Optionally, the correction module is used to:

[0157] Determine a mapping relationship between the road data in the SD map data and the trajectory data in the road collection data as a first mapping relationship;

[0158] Determining a positional relationship between road data in the SD map data and elements marked in the reference map as a second mapping relationship;

[0159] According to the first mapping relationship and the second mapping relationship, elements in the reference map are extracted as a first type of element set.

[0160] Optionally, the correction module is used to:

[0161] For each route in the SD map data, determine the minimum value of the distance between the trajectory point in the road collection data and the center point of the route as the distance corresponding to the route;

[0162] According to the minimum value, the route is projected onto the trajectory formed by the road collection data, and the trajectory point corresponding to the route in the road collection data is determined;

[0163] The angle between the vector formed by the starting point and the end point of the trajectory point corresponding to the route in the road collection data and the vector formed by the starting point and the end point of the route is determined as the angle corresponding to the route.

[0164] Optionally, the correction module is used to:

[0165] For each element in the first type of element set, determine a set of multiple vectors with the element as a starting point and multiple other elements in the first type of element set as end points, as a first positional relationship;

[0166] For each element in the second type element set, a set of multiple vectors with the element as a starting point and other multiple elements in the second type element set as end points is determined as the second positional relationship.

[0167] Optionally, the correction module is used to determine the deviation value of each element in the first type of element set according to the first position relationship and the second position relationship, including:

[0168] For each feature in the first category feature set, searching for a feature in the second category feature set whose distance to the feature is less than a predetermined distance threshold;

[0169] Move the feature to the location of the found feature;

[0170] matching the first position relationship with the second position relationship;

[0171] If the match is successful, the deviation angle of the feature relative to the found feature is determined, wherein the deviation value includes a vector formed by the movement of the feature and the deviation angle.

[0172] Optionally, the correction module is used to:

[0173] Matching the first position relationship and the second position relationship includes:

[0174] Multiple vectors with the first element in the first category of element set as the starting point are matched with multiple vectors with the element in the found second category of element set as the starting point. Among the multiple vectors with the first element as the starting point, if the proportion of successfully matched vectors is greater than a first proportion threshold, the first position relationship and the second position relationship are matched successfully.

[0175] Optionally, the correction module is used for: if the proportion of the first vector falling within a predetermined range around the second vector is greater than a second proportion threshold, the first vector and the second vector are matched successfully.

[0176] Optionally, the correction module is used to:

[0177] Determine the angle between two successfully matched vectors;

[0178] The average value of the included angles corresponding to the successfully matched vectors among the multiple vectors with the first element as the starting point is taken as the deviation angle of the first element relative to the found element.

[0179] Optionally, the elements include ground turn arrows and lane lines. The device also includes a generation module.

[0180] The generation module is used to generate virtual lanes at intersections in the target map according to ground turn arrows and lane lines in the target map.

[0181] Optionally, a generation module is used to:

[0182] When a vehicle exits from an exit road and goes straight into an incoming road, and the number of lanes on the exit road is different from the number of lanes on the incoming road and both are greater than 1, for each exit lane on the exit road, an incoming lane whose overlap with the exit lane is greater than an overlap threshold is connected to the exit lane to generate a virtual lane;

[0183] Among them, the overlap is the ratio of the part of the intersection stop line at the starting end of the entry lane that is intercepted by the lane extension line of the exit lane and included in the extension line to the intersection stop line. The exit road is the road that the vehicle enters the intersection after exiting, and the entry road is the road that the vehicle enters after exiting the intersection.

[0184] Optionally, a generation module is used to:

[0185] When a vehicle turns from an exit road to an entry road, and the number of lanes on the exit road is different from that on the entry road and both are greater than 1, the N consecutive lanes in the lanes on the multi-lane road are divided into a group, and the groups are connected with the lanes on the road with fewer lanes to generate virtual lanes;

[0186] Among them, the multi-lane road is the one with more lanes between the exit road and the entrance road, the few-lane road is the one with fewer lanes between the exit road and the entrance road, the number of lanes in the few-lane road is greater than or equal to N, the exit road is the road that the vehicle enters after exiting the intersection, and the entrance road is the road that the vehicle enters after exiting the intersection.

[0187] The present disclosure also provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the steps of the above method provided by the present disclosure are implemented.

[0188] The present disclosure also provides a computer program product, including a computer program, which implements the steps of the above method provided by the present disclosure when the computer program is executed by a processor.

[0189] The present disclosure also provides an electronic device, the electronic device comprising:

[0190] processor;

[0191] a memory for storing processor-executable instructions;

[0192] Among them, the processor is configured to execute the steps of the above-mentioned map generation method provided by the present disclosure.

[0193] Regarding the device in the above embodiment, the specific manner in which each module performs operations has been described in detail in the embodiment of the method, and will not be elaborated here.

[0194] Through the above technical solution, the elements in the SD map are marked according to the element information in the SD map data to generate a reference map. The positions of the elements marked in the reference map are corrected according to the road collection data to obtain the target map. In this way, based on the SD map data, the correction is made with reference to the road collection data, so as to efficiently generate a lightweight high-definition map.

[0195] The preferred embodiments of the present disclosure are described in detail above in conjunction with the accompanying drawings; however, the present disclosure is not limited to the specific details in the above embodiments. Within the technical concept of the present disclosure, a variety of simple modifications can be made to the technical solution of the present disclosure, and these simple modifications all fall within the protection scope of the present disclosure.

[0196] It should also be noted that the various specific technical features described in the above specific embodiments can be combined in any suitable manner without contradiction. In order to avoid unnecessary repetition, the present disclosure will not further describe various possible combinations.

[0197] In addition, various embodiments of the present disclosure may be arbitrarily combined, and as long as they do not violate the concept of the present disclosure, they should also be regarded as the contents disclosed by the present disclosure.

Claims

1. A map generation method, characterized in that: The method comprises: Acquire SD map data and road data collected when the vehicle is traveling on the road; Marking elements in the SD map according to the element information in the SD map data to generate a reference map; Correcting the positions of the elements marked in the reference map according to the road collection data to obtain a target map; The step of correcting the positions of the elements marked in the reference map according to the road collection data to obtain the target map includes: Extracting elements in the reference map as a first type of element set according to a mapping relationship between the road data in the SD map data and the trajectory data in the road collection data; Determining a relative positional relationship between elements in the first type of element set as a first positional relationship; Using a machine learning method, extracting elements in the reference map according to the road collection data as a second type of element set; Determining a relative positional relationship between elements in the second type of element set as a second positional relationship; Determine a deviation value of each element in the first type of element set according to the first position relationship and the second position relationship; According to the deviation value, the position of each element in the first type of element set is corrected to obtain the target map.

2. The method according to claim 1, characterized in that The extracting of elements in the reference map according to the mapping relationship between the road data in the SD map data and the trajectory data in the road collection data as a first type of element set includes: Determine a mapping relationship between the road data in the SD map data and the trajectory data in the road collection data as a first mapping relationship; Determining a positional relationship between the road data in the SD map data and the elements marked in the reference map as a second mapping relationship; According to the first mapping relationship and the second mapping relationship, elements in the reference map are extracted as the first type of element set.

3. The method according to claim 2, characterized in that The determining of the mapping relationship between the road data in the SD map data and the trajectory data in the road collection data includes: For each route in the SD map data, determining the minimum value of the distance between the trajectory point in the road collection data and the center point of the route as the distance corresponding to the route; According to the minimum value, projecting the route onto a trajectory formed by the road collection data, and determining a trajectory point in the road collection data corresponding to the route; The angle between the vector formed by the starting point and the end point of the trajectory point corresponding to the route in the road collection data and the vector formed by the starting point and the end point of the route is determined as the angle corresponding to the route.

4. The method according to claim 1, characterized in that: The determining the relative positional relationship between the elements in the first type of element set as the first positional relationship includes: for each element in the first type of element set, determining a set of multiple vectors with the element as a starting point and multiple other elements in the first type of element set as end points as the first positional relationship; The determining of the relative positional relationship between the elements in the second category of element set as the second positional relationship includes: for each element in the second category of element set, determining a set of multiple vectors with the element as the starting point and other multiple elements in the second category of element set as the end points, as the second positional relationship.

5. The method according to claim 4, characterized in that The step of determining the deviation value of each element in the first type of element set according to the first position relationship and the second position relationship includes: For each element in the first element set, searching the second element set for an element whose distance to the element is less than a predetermined distance threshold; Moving the element to the position of the found element; Matching the first position relationship with the second position relationship; If the match is successful, the deviation angle of the element relative to the found element is determined, wherein the deviation value includes a vector formed by the movement of the element and the deviation angle.

6. The method according to claim 5, characterized in that The matching the first position relationship and the second position relationship includes: Match multiple vectors with the first element in the first type of element set as the starting point with multiple vectors with the element in the second type of element set as the starting point, wherein, among the multiple vectors with the first element as the starting point, if the proportion of successfully matched vectors is greater than a first proportion threshold, then the first position relationship and the second position relationship are successfully matched.

7. The method according to claim 6, characterized in that If the proportion of the first vector falling within the predetermined range around the second vector is greater than a second proportion threshold, the first vector and the second vector are matched successfully.

8. The method according to claim 7, characterized in that Determining the deviation angle of the element relative to the found element includes: Determine the angle between two successfully matched vectors; An average value of angles corresponding to successfully matched vectors among a plurality of vectors starting from the first element is taken as a deviation angle of the first element relative to the found element.

9. The method according to claim 1, characterized in that: The elements include ground turn arrows and lane lines, and the method further includes: A virtual lane at an intersection is generated in the target map according to the ground turn arrows and lane lines in the target map.

10. The method according to claim 9, characterized in that The step of generating a virtual lane at an intersection in the target map according to the ground turn arrows and lane lines in the target map comprises: When a vehicle enters an incoming road straight after exiting an exit road, and the number of lanes of the exit road is different from the number of lanes of the incoming road and both are greater than 1, for each exit lane of the exit road, an incoming lane of the incoming road whose overlap with the exit lane is greater than an overlap threshold is connected to the exit lane to generate a virtual lane; Among them, the overlap degree is the ratio of the intersection stop line at the starting end of the entry lane that is intercepted by the lane extension line of the exit lane and included in the extension line to the intersection stop line, the exit road is the road that the vehicle enters the intersection after exiting, and the entry road is the road that the vehicle enters after exiting the intersection.

11. The method according to claim 9, characterized in that The step of generating a virtual lane at an intersection in the target map according to the ground turn arrows and lane lines in the target map comprises: When a vehicle turns from an exit road to an entry road, and the number of lanes of the exit road is different from the number of lanes of the entry road and both are greater than 1, N consecutive lanes in the lanes of the multi-lane road are divided into a group, and the groups are connected with the lanes in the road with fewer lanes to generate virtual lanes; Among them, the multi-lane road is the one with more lanes between the exit road and the entrance road, the few-lane road is the one with fewer lanes between the exit road and the entrance road, the number of lanes in the few-lane road is greater than or equal to N, the exit road is the road that the vehicle enters after exiting the intersection, and the entrance road is the road that the vehicle enters after exiting the intersection.

12. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method described in any one of claims 1 to 11 are implemented.

13. A computer program product, characterized in that The invention comprises a computer program which, when executed by a processor, implements the steps of the method according to any one of claims 1 to 11.

14. An electronic device, characterized in that: The electronic device comprises: processor; a memory for storing processor-executable instructions; The processor is configured to execute the steps of the map generation method according to any one of claims 1 to 11.

Images