Methods for generating contextualized map data, lane-level navigation methods and devices

Abstract

This application provides a method for generating scenario-based map data, a lane-level navigation method, and an apparatus. The method for generating scenario-based map data includes: determining a scenario trajectory corresponding to a road traffic scenario based on the vehicle's driving trajectory; extracting driving habit features under multiple traffic modes based on the scenario trajectory; and determining scenario-based map data for the road traffic scenario under multiple traffic modes based on the driving habit features under multiple traffic modes. The scenario-based map data is used to support lane-level navigation for the road traffic scenario under multiple traffic modes. This method aims to improve the accuracy of lane-level navigation.

Description

Technical Field

[0001] This application relates to the field of intelligent driving, and in particular to a method for generating scenario-based map data, a lane-level navigation method, and a device. Background Technology

[0002] Lane-level navigation provides driving guidance for vehicles and is a core capability of intelligent driving; in order to improve the performance of lane-level navigation, a high degree of accuracy in route planning is required.

[0003] In related technologies, lane-level navigation relies on static geometric lane lines provided by static maps, and then uses path planning algorithms to plan paths to achieve lane-level navigation. The paths planned in this way are geometrically optimal or rule-optimal in a static environment, but the dynamic adaptability is poor. In dynamically changing and complex traffic conditions, the planned paths have problems such as abrupt turns and unreasonable lane change timing, resulting in low accuracy of lane-level navigation. Summary of the Invention

[0004] This application provides a method for generating scenario-based map data, a lane-level navigation method, and an apparatus to improve the accuracy of lane-level navigation.

[0005] In a first aspect, embodiments of this application provide a method for generating contextualized map data, including:

[0006] Based on the vehicle's driving trajectory, determine the scene trajectory corresponding to the road traffic scenario;

[0007] Based on scene trajectories, driving habit features are extracted under various traffic modes;

[0008] Based on driving habit characteristics under various traffic modes, scenario-based map data for road traffic scenarios under various traffic modes is determined. The scenario-based map data is used to support lane-level navigation for road traffic scenarios under various traffic modes.

[0009] Secondly, embodiments of this application provide a lane-level navigation method, including:

[0010] Obtain the vehicle's current road traffic scenario and traffic mode;

[0011] Based on the vehicle's current road traffic scenario and traffic mode, scenario recommendation elements are extracted from the scenario-based map data, which is generated based on the scenario-based map data generation method.

[0012] Lane-level navigation is based on static navigation elements and scene recommendation elements corresponding to road traffic scenarios.

[0013] Thirdly, the contextual map data generation apparatus provided in the embodiments of this application includes:

[0014] The trajectory processing module is used to determine the scene trajectory corresponding to the road traffic scenario based on the vehicle's driving trajectory.

[0015] The feature extraction module is used to extract driving habit features under various traffic modes based on scene trajectories;

[0016] The generation module is used to determine the contextualized map data of road traffic scenarios under various traffic modes based on driving habit characteristics under multiple traffic modes. The contextualized map data is used to support lane-level navigation of road traffic scenarios under various traffic modes.

[0017] Fourthly, the lane-level navigation device provided in the embodiments of this application includes:

[0018] The acquisition module is used to acquire the current road traffic scenario and traffic mode of the vehicle;

[0019] The contextualized map data processing module is used to extract scenario recommendation elements from the contextualized map data based on the vehicle's current road traffic scenario and traffic mode. The contextualized map data is generated based on the method in the first aspect.

[0020] The lane-level navigation module is used for lane-level navigation based on static navigation elements and scene recommendation elements corresponding to road traffic scenarios.

[0021] Fifthly, embodiments of this application provide an electronic device, including: a memory and a processor; the memory stores computer-executable instructions; the processor executes the computer-executable instructions stored in the memory, causing the processor to perform the implementation methods of the first and / or second aspects described above.

[0022] Fourthly, embodiments of this application provide a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, are used to implement the embodiments of the first and / or second aspects described above.

[0023] Fifthly, embodiments of this application provide a computer program product, including a computer program that, when executed by a processor, implements the implementation methods of the first and / or second aspects described above.

[0024] The scenario-based map data generation method, lane-level navigation method, and apparatus provided in this application determine the scenario trajectory corresponding to the road traffic scenario in the vehicle's driving trajectory, and extract driving habit features under multiple traffic modes based on the scenario trajectory. That is, for the same road traffic scenario, driving habit features are extracted according to multiple traffic modes, realizing the subdivision of driving habits under different dynamic changes in the same road traffic scenario, improving the accuracy of the extracted driving habit features, and converting driving habit features into scenario-based map data, realizing the transformation of human historical driving experience into quantifiable and reusable map data. Since scenario-based map data can reflect the group driving habits of a large number of drivers under different traffic modes in road traffic scenarios, applying scenario-based map data to lane-level navigation improves the anthropomorphism of lane-level navigation, improves the rigidity of planned paths, enhances the dynamic adaptability of path planning, and thus improves the accuracy of lane-level navigation. Attached Figure Description

[0025] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0026] Figure 1 Flowchart of the method for generating contextualized map data provided in this application Figure 1 ;

[0027] Figure 2 Flowchart of the method for generating contextualized map data provided in this application Figure 2 ;

[0028] Figure 3 A flowchart illustrating the lane-level navigation provided in this application;

[0029] Figure 4 A schematic diagram of the lane-level navigation system provided in this application;

[0030] Figure 5 A schematic diagram of the structure of the device for generating scenario-based map data provided in this application;

[0031] Figure 6 A schematic diagram of the lane-level navigation device provided in this application;

[0032] Figure 7 A schematic diagram of the structure of the electronic device provided in this application.

[0033] The accompanying drawings illustrate specific embodiments of this application, which will be described in more detail below. These drawings and descriptions are not intended to limit the scope of the concept in any way, but rather to illustrate the concept of this application to those skilled in the art through reference to particular embodiments. Detailed Implementation

[0034] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application as detailed in the appended claims.

[0035] Application scenarios of this application: Lane-level navigation is used to provide driving guidance for vehicles and is a core capability of intelligent driving; the goal of lane-level navigation is to provide vehicles with accurate, comfortable and human-like driving guidance. In order to improve the performance of lane-level navigation, the accuracy of route planning is required to be high.

[0036] In related technologies, lane-level navigation relies on static geometric information provided by static maps (such as static high-precision maps), including lane lines, topological connections, traffic signs, etc., and then uses path planning algorithms, such as A* algorithm and Dijkstra algorithm, to find a collision-free and traffic-compliant path from the starting point to the destination in the lane network provided by the map. The vehicle control module then follows this path to achieve lane-level navigation. This method is a navigation method based on an "ideal geometric model". In order to improve comfort, some solutions will introduce dynamic trajectory smoothing algorithms. The path planned by this method is essentially still based on a fixed static map and is the geometrically optimal or rule-optimal path in a static environment. In dynamic and complex traffic conditions, the planned path has problems such as abrupt turning and unreasonable lane change timing, resulting in poor dynamic adaptability and low accuracy of path planning.

[0037] The technical solution of this application and how the technical solution of this application solves the above-mentioned technical problems are described in detail below with specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments. The embodiments of this application will now be described with reference to the accompanying drawings.

[0038] Figure 1 This is a flowchart illustrating the method for generating contextualized map data provided in this application. This method can be applied to electronic devices, which can be cloud-based, such as… Figure 1 As shown, the methods for generating contextualized map data include:

[0039] S101. Determine the scene trajectory corresponding to the road traffic scenario based on the vehicle's driving trajectory.

[0040] Among them, the vehicle's driving trajectory is the trajectory generated by the actual driving of the vehicle.

[0041] Optionally, vehicles equipped with high-precision positioning systems (such as test vehicles and mass-produced vehicles) collect GPS trajectory points in real time during driving, obtain the trajectory data corresponding to the trip at the end of the trip, perform anonymization and compliance processing on the trajectory data to obtain the original trajectory, and upload the original trajectory to the cloud; for example, when the vehicle's control actuator performs a new driving behavior, trajectory points are collected through on-board sensors; the trajectory data is anonymized and processed for compliance by the vehicle's compliance engine to obtain the original trajectory.

[0042] The cloud can acquire a large number of raw trajectories, which come from different vehicles and different journeys. The cloud processes the raw trajectories to obtain the vehicle's driving trajectory. For example, the cloud cleans the raw trajectories (such as removing duplicate trajectory points and abnormal trajectory points) and smooths them to obtain preprocessed trajectories. The preprocessed trajectories are then matched with high-precision maps to associate the discrete trajectories included in the preprocessed trajectories with the road, thus obtaining the vehicle's driving trajectory.

[0043] In one alternative approach, a road traffic scenario refers to a scenario in which a vehicle travels within a road facility area, which includes, but is not limited to, intersection areas, ramp areas, and toll station areas. Correspondingly, the scenario trajectory corresponding to a road traffic scenario is a segment of the travel trajectory that is located within the road facility area.

[0044] Specifically, a large number of driving trajectories are acquired, the first driving trajectory passing through the road facility area is selected from the driving trajectories, and the second driving trajectory within the road facility area is selected from the first driving trajectory to obtain the scene trajectory corresponding to the road traffic scenario.

[0045] Optionally, by dividing the driving trajectory according to different road facility areas, scene trajectories under different road traffic scenarios can be obtained.

[0046] In another alternative approach, the road traffic scenario refers to a scenario in which traffic occurs within a road facility area during a preset time period. For example, the preset time period can be a weekday or a holiday, or it can be a weekday peak period, a weekday off-peak period, a holiday peak period, or a holiday off-peak period. Accordingly, the scenario trajectory corresponding to the road traffic scenario is a segment of the travel trajectory that is within the road facility area during the preset time period.

[0047] Specifically, a large number of driving trajectories are acquired, a first driving trajectory passing through a road facility area is selected from the driving trajectories, a second driving trajectory within the road facility area is selected from the first driving trajectory, and a third driving trajectory corresponding to a preset time period is determined from the second driving trajectory, thus obtaining the scene trajectory corresponding to the road traffic scenario.

[0048] Optionally, by dividing the driving trajectory according to different road facility areas and different preset time periods, scene trajectories under different road traffic scenarios can be obtained.

[0049] In another alternative approach, the road traffic scenario is a driving scenario in which a vehicle passes through a road facility area in a preset time period and a preset direction of travel. A first driving trajectory passing through the road facility area is selected from the driving trajectory, and a second driving trajectory within the road facility area is selected from the first driving trajectory. Based on the preset time period and preset direction of travel, the scenario trajectory under the road traffic scenario is determined from the second driving trajectory.

[0050] Optionally, the driving trajectory can be divided according to different road facility areas, different preset time periods, and different preset traffic directions to obtain scene trajectories under different road traffic scenarios.

[0051] S102. Based on scene trajectory, extract driving habit features under various traffic modes.

[0052] Among them, driving habit characteristics are used to reflect the group driving habits of a large number of drivers in road traffic scenarios; for example, driving habit characteristics include at least one of center trajectory, habitual lane change range, habitual parking position or habitual speed.

[0053] It should be noted that different drivers have different driving habits. However, in the same road traffic scenario, in order to drive more safely or more efficiently in that road traffic scenario, different drivers make driving decisions that have commonalities, and thus the resulting scenario trajectories have commonalities. By performing data mining on the scenario trajectories, we can obtain driving habit characteristics that are tied to that road traffic scenario and reflect human driving habits.

[0054] Among them, the traffic mode is used to reflect the road traffic status; the traffic mode is preset, and there are multiple traffic modes. The number of traffic modes is used as the preset number, and the scene trajectory is clustered according to the preset number to obtain the preset number of clusters. That is, the number of multiple clusters is equal to the number of multiple traffic modes, and the multiple clusters correspond one-to-one with the multiple traffic modes.

[0055] Specifically, the scene trajectories are clustered according to a preset number of multiple access modes to obtain multiple clusters; for each cluster, the access mode corresponding to the cluster is determined based on the scene trajectories included in the cluster.

[0056] For example, there are three traffic modes: smooth traffic, slow traffic, and congestion mode, with a preset number of three. The scene trajectory is clustered according to the preset number to obtain three clusters. Since the core difference between the three traffic modes is vehicle speed, the traffic mode corresponding to each of the three clusters is determined according to the vehicle speed information of the scene trajectory included in the three clusters. For example, the traffic mode of cluster A is smooth traffic, the traffic mode of cluster B is slow traffic, and the traffic mode of cluster C is congestion.

[0057] After determining the traffic patterns corresponding to multiple clusters, for each traffic pattern, the driving habit characteristics under that traffic pattern are determined based on the scene trajectories included in the clusters corresponding to that traffic pattern.

[0058] For example, driving habit characteristics include a center trajectory and a habitual lane-changing interval. Based on the location information of the trajectory points of the scene trajectory in cluster A, the center trajectory in the smooth flow mode is determined. Based on the location information of the trajectory points of the scene trajectory in cluster A and the lane information, the habitual lane-changing interval in the smooth flow mode is determined. Similarly, based on the scene trajectory in cluster B, the center trajectory and habitual lane-changing interval in the slow flow mode can be determined. Based on the scene trajectory in cluster C, the center trajectory and habitual lane-changing interval in the congested mode can be determined.

[0059] S103. Based on the driving habit characteristics under various traffic modes, determine the scenario-based map data for road traffic scenarios under various traffic modes. The scenario-based map data is used to support lane-level navigation for road traffic scenarios under various traffic modes.

[0060] Specifically, for each traffic mode in a road traffic scenario, the driving habit characteristics under that traffic mode are converted into a preset format to obtain scenario-based map data under that traffic mode. In other words, the same road traffic scenario has scenario-based map data under multiple traffic modes; where the preset format can be a structured data format.

[0061] In some embodiments, the method determines the contextualized map data of a road traffic scenario under multiple traffic modes based on driving habit characteristics under multiple traffic modes, including: performing a structured transformation based on driving habit characteristics under multiple traffic modes to obtain the contextualized map data of the road traffic scenario under multiple traffic modes; the method further includes: obtaining the road facility area corresponding to the road traffic scenario; and associating the contextualized map data of the road traffic scenario under multiple traffic modes with static map data based on the road facility area.

[0062] Specifically, the driving habit characteristics under various traffic modes are structurally transformed, that is, converted into a structured data format to obtain scenario-based map data under various traffic modes; the road facility areas corresponding to the road traffic scenarios are obtained, and the scenario-based map data of the road traffic scenarios under various traffic modes are associated with the road facility areas in the static map data to form scenario-based map data bound to specific road facility areas in the static map data.

[0063] In practical applications, the relevant data of road facility areas in static map data are lane layer or road layer data. By associating the scenario-based map data of road traffic scenarios under various traffic modes with static map data, the scenario-based map data and the lane layer or road layer of static map data can be linked.

[0064] In the above embodiments, the driving habit characteristics of road traffic scenarios under various traffic modes are converted into scenario-based map data, and associated with road facility areas to form scenario-based map data bound to specific road facility areas in static map data. This facilitates the rapid acquisition of scenario-based map data in lane-level navigation and improves lane-level navigation efficiency.

[0065] It should be noted that contextualized map data is used to support lane-level navigation in various traffic modes. This can be achieved by: in addition to obtaining static navigation elements corresponding to the vehicle's current road traffic scenario from static map data, determining matching contextualized map data based on the vehicle's current road traffic scenario and traffic mode; determining recommended scenario elements through the contextualized map data; and processing the static navigation elements and recommended scenario elements through a path planning algorithm to achieve lane-level navigation. Since contextualized map data is based on a large number of driving trajectories, it can reflect the collective driving intelligence. Adding contextualized map data to lane-level navigation improves the anthropomorphic level of lane-level navigation.

[0066] The method for generating scenario-based map data provided in this application determines the scenario trajectory corresponding to the road traffic scenario in the vehicle's driving trajectory. Based on a large number of scenario trajectories of the same road traffic scenario, driving habit features are extracted according to multiple traffic modes. This achieves the extraction of detailed driving habit features under different traffic conditions for the same road traffic scenario, improving the accuracy of the extracted driving habit features. The driving habit features are converted into scenario-based map data, realizing the transformation of a large number of typical driving experiences of human drivers into quantifiable and reusable scenario-based map data. Since scenario-based map data can reflect the group driving habits of a large number of human drivers under different traffic modes in road traffic scenarios, applying scenario-based map data to lane-level navigation improves the anthropomorphism of lane-level navigation, improves the rigidity of planned paths, enhances the dynamic adaptability of path planning, and thus improves the accuracy of lane-level navigation.

[0067] In some embodiments, a road traffic scenario is a driving scenario in which a vehicle passes through a road facility area in a preset direction during a preset time period. Determining the scenario trajectory corresponding to the road traffic scenario based on the vehicle's driving trajectory includes: determining the scenario trajectory corresponding to the road traffic scenario based on the vehicle's driving trajectory, including: slicing the vehicle's driving trajectory based on the road facility area to obtain candidate trajectories; and determining the scenario trajectory in which the vehicle passes through the road facility area in a preset direction during a preset time period based on the candidate trajectories.

[0068] The preset time period can be one of the following: weekday peak hours, weekday off-peak hours, weekday night hours, holiday peak hours, and holiday off-peak hours. The number of vehicles and pedestrians traveling in different preset time periods is different, so different preset time periods correspond to different road conditions.

[0069] The preset travel direction indicates the direction in which vehicles travel within the infrastructure area; different infrastructure areas may correspond to different preset travel directions.

[0070] For example, the infrastructure area is an intersection area, and the preset traffic direction can be a north-south straight direction, a south-to-west turning direction, etc.

[0071] For example, the infrastructure area is a ramp area, and the preset traffic direction can be a southbound circular turn, a northbound circular turn, a direction to enter the highway, a direction to exit the highway, etc.

[0072] For example, the infrastructure area is the toll station area, and the preset traffic direction can be the entrance ETC lane direction, the entrance manual lane direction, the exit ETC lane direction, the exit manual lane direction, etc.

[0073] Among them, the candidate trajectory is the trajectory segment within the road facility area of ​​the driving trajectory.

[0074] Specifically, for each road traffic scenario, the corresponding road facility area, preset time period, and preset traffic direction are determined;

[0075] For each driving trajectory, the trajectory is sliced ​​based on the boundary of the road facility area and the location information of the trajectory points in the trajectory, resulting in trajectory segments within and outside the road facility area. The trajectory segments within the road facility area are selected as candidate trajectories.

[0076] For example, the driving trajectory can be sliced ​​based on the boundary of the intersection area to cut out trajectory segments inside and outside the intersection area, and the trajectory segments inside the intersection area can be used as candidate trajectories.

[0077] Based on the timestamps of the trajectory points in the candidate trajectory, the travel time of the candidate trajectory is determined. For example, the average of the timestamps of the trajectory points in the candidate trajectory is used as the travel time of the candidate trajectory, or the earliest timestamp of the trajectory point in the candidate trajectory is used as the travel time of the candidate trajectory. The first candidate trajectory whose travel time belongs to a preset time period is selected from the candidate trajectories. For example, if the preset time period is weekdays, the candidate trajectory whose travel time belongs to weekdays is selected as the first candidate trajectory.

[0078] Based on the timestamps of the trajectory points in the first candidate trajectory, determine the entry trajectory point (the trajectory point with the earliest timestamp) and the exit trajectory point (the trajectory point with the latest timestamp); based on the direction of the road to the entry trajectory point and the direction of the road to the exit trajectory point, determine the travel direction of the first candidate trajectory; for example, if the direction of the road (link) to the entry trajectory point is north-south, and the direction of the road to the exit trajectory point is also north-south, then the travel direction is a straight north-south direction.

[0079] Select a second candidate trajectory from the first candidate trajectory, with the travel direction being the preset travel direction, and use the second candidate trajectory as the scene trajectory corresponding to the road travel scenario.

[0080] It should be noted that different road facility areas, different preset time periods, and different traffic directions can be combined to obtain various road traffic scenarios. For example, road traffic scenario C1 corresponds to the weekday peak hours, north-south straight direction, and intersection area; road traffic scenario C2 corresponds to the holiday off-peak hours, south-to-west turning direction, and intersection area; road traffic scenario C3 corresponds to the weekday night hours, direction of entering the highway, and ramp area; road traffic scenario C4 corresponds to the weekday night hours, direction of entering the ETC lane, and toll station area. For each road traffic scenario, the above process of this embodiment is adopted to obtain the scene trajectory corresponding to each road traffic scenario.

[0081] In the above embodiments, since the scene trajectories of road facility areas in the same time period and in the same direction of traffic have the same trajectory characteristics, the scene trajectory corresponding to the road traffic scenario is selected from the vehicle's driving trajectory according to the road facility area, preset time period and preset direction of traffic. This realizes the scene-based subdivision of vehicle trajectory, and subsequently determines driving habit characteristics based on scene trajectory, thereby improving the accuracy of driving habit characteristics.

[0082] In some embodiments, based on scene trajectories, driving habit features under multiple traffic modes are extracted, including: performing clustering processing on the scene trajectories to obtain multiple clusters; determining the traffic modes corresponding to the multiple clusters respectively; and extracting driving habit features under the traffic modes corresponding to the multiple clusters based on the target trajectories contained in the multiple clusters.

[0083] In one implementation, the scene trajectory is clustered to obtain multiple clusters, including: clustering based on the location information of the trajectory points contained in the scene trajectory according to a preset number to obtain multiple clusters.

[0084] The location information of the trajectory point can be the latitude and longitude of the trajectory point, or the geodetic coordinates of the trajectory point.

[0085] Specifically, there are multiple scene trajectories. Based on the positional information of the trajectory points contained in any two scene trajectories, the positional similarity between any two scene trajectories is determined. Based on the positional similarity, multiple scene trajectories are clustered according to a preset number to obtain multiple clusters. Clustering is performed by positional similarity so that the target trajectory in each cluster is a trajectory with similar position and shape.

[0086] In another implementation, the scene trajectory is clustered to obtain multiple clusters, including: clustering based on the location information, time information and vehicle speed information of the trajectory points contained in the scene trajectory according to a preset number to obtain multiple clusters.

[0087] Among them, the time information of the trajectory point is the timestamp of the trajectory point.

[0088] Specifically, based on the vehicle speed information of the trajectory points in the scene trajectory, the habitual vehicle speed vector of the scene trajectory is determined; for example, based on the vehicle speed information of the trajectory points in the scene trajectory, the average vehicle speed, standard deviation of vehicle speed, and maximum vehicle speed of the scene trajectory are determined, and the habitual vehicle speed vector of the scene trajectory includes the average vehicle speed, standard deviation of vehicle speed, and maximum vehicle speed.

[0089] Based on the vehicle speed information of the trajectory points in the scene trajectory, determine the parking points included in the scene trajectory; for example, trajectory points with vehicle speeds less than the preset vehicle speed are designated as parking points.

[0090] There are multiple scene trajectories. Based on the location information of the trajectory points contained in any two scene trajectories, the positional similarity between any two scene trajectories is determined; based on the habitual vehicle speed vectors of any two scene trajectories, the vehicle speed similarity between any two scene trajectories is determined; based on the parking points included in any two scene trajectories, the parking position similarity between any two trajectories is determined.

[0091] The weighted sum of positional similarity, vehicle speed similarity, and parking position similarity between any two scene trajectories is used to obtain the comprehensive similarity between the two scene trajectories. Based on the comprehensive similarity between any two scene trajectories, multiple scene trajectories are clustered according to a preset number to obtain multiple clusters. Clustering by comprehensive similarity ensures that the target trajectories in each cluster are similar in position and shape, vehicle speed, and parking position.

[0092] Optionally, the positional similarity between any two scene trajectories is determined based on the positional information of the trajectory points contained in any two scene trajectories. This includes: any two scene trajectories include a first scene trajectory and a second scene trajectory; based on the positional information of the trajectory points in the first scene trajectory and the second scene trajectory, a distance matrix is ​​constructed between the first scene trajectory and the second scene trajectory; from the upper left corner to the lower right corner of the distance matrix, a path with the minimum total distance is determined; the distance of this path is used as the dynamic time warping distance; and the positional similarity between the first scene trajectory and the second scene trajectory is determined based on the dynamic time warping distance. In practical applications, this positional similarity is also called DTW (Dynamic Time Warping) similarity, which can reflect the similarity between the shape and position of the two trajectories.

[0093] Optionally, the speed similarity between any two scene trajectories is determined based on the habitual speed vectors of any two scene trajectories, including: using the cosine distance between the first habitual speed vector and the second habitual speed vector as the speed similarity between the first scene trajectory and the second scene trajectory.

[0094] Optionally, based on the parking points included in any two scene trajectories, the parking location similarity between any two trajectories is determined, including: if the first scene trajectory does not include the first parking point and the second scene trajectory does not include the second parking point, then the parking similarity is the first parking similarity.

[0095] If the first scene trajectory includes the first parking point and the second scene trajectory includes the second parking point, the number of matches is determined based on the first parking point and the second parking point. The ratio between this number of matches and the maximum number of the first parking point and the second parking point is used as the second parking similarity.

[0096] The method of determining the number of matches based on the first and second parking points can be achieved by matching the closest first and second parking points in pairs based on the distance between them, obtaining parking point pairs, and counting the number of parking point pairs whose distance is less than a preset distance to obtain the number of matches.

[0097] If the first scene trajectory includes the first parking point, and the second scene trajectory does not include the second parking point, then the parking similarity is the third parking similarity.

[0098] The first parking space has a higher similarity than the second parking space, and the second parking space has a higher similarity than the third parking space.

[0099] Optionally, the position similarity, vehicle speed similarity, and parking position similarity between any two scene trajectories are weighted and summed to obtain the comprehensive similarity between any two scene trajectories. This includes: weighting and summing the position similarity, vehicle speed similarity, and parking position similarity between any two scene trajectories according to position similarity weight, vehicle speed similarity weight, and parking point similarity weight to obtain the comprehensive similarity. The position similarity weight, vehicle speed similarity weight, and parking point similarity weight can be set according to requirements. For example, the position similarity weight, vehicle speed similarity weight, and parking similarity weight can be set to be equal, or the position similarity weight can be set to 0.2, the vehicle speed similarity weight to 0.6, and the parking similarity weight to 0.2.

[0100] In one implementation, determining the traffic patterns corresponding to multiple clusters includes: determining the average speed of multiple clusters based on the vehicle speed information of the target trajectories contained in each cluster, and determining the traffic patterns corresponding to multiple clusters based on the average speed of multiple clusters.

[0101] Specifically, there are three traffic modes: smooth flow, slow flow, and congestion. Correspondingly, there are three clusters. For each cluster, the average vehicle speed information of the trajectory points in the target trajectory included in the cluster is calculated to obtain the cluster average vehicle speed. Among the three clusters, the cluster with the largest average vehicle speed corresponds to the smooth flow mode, the cluster with the second largest average vehicle speed corresponds to the slow flow mode, and the cluster with the smallest average vehicle speed corresponds to the congestion mode.

[0102] In another implementation, when the road facility area in the road traffic scenario includes a turning area and the preset traffic direction is a turning direction, the traffic mode corresponding to multiple clusters is determined, including: determining the parking event type of multiple clusters based on the parking points of the target trajectories contained in each cluster, and determining the traffic mode corresponding to multiple clusters based on the parking event type of multiple clusters.

[0103] Specifically, there are two traffic modes: a waiting mode for turning over and a non-waiting mode. Correspondingly, the number of multiple clusters is two. For each cluster, among the parking points in the target trajectory included in the cluster, a clustered parking point is determined, the location information of the turning over area is obtained, and the target distance between the clustered parking point and the turning over area is determined based on the location information of the clustered parking point and the turning over area. Among the two clusters, the cluster with the smaller target distance is assigned to the waiting mode for turning over, and the cluster with the larger target distance is assigned to the non-waiting mode.

[0104] In one implementation, driving habit features under the traffic patterns corresponding to multiple clusters are extracted based on the target trajectories contained in multiple clusters. This includes: for each cluster, extracting driving habit features of the cluster based on the location information, lane information, time information, and vehicle speed information of the trajectory points included in the target trajectory of the cluster; wherein, the driving habit features include: center trajectory, habitual lane change interval, habitual parking position, and habitual vehicle speed.

[0105] In the above embodiments, clustering processing is used to subdivide scene trajectories according to traffic patterns. By using the target trajectories included in the clusters, driving habit characteristics under different traffic patterns in the same road traffic scenario are determined, thereby improving the accuracy of driving habit characteristics.

[0106] In some embodiments, driving habit features include at least one of: center trajectory, habitual lane change interval, habitual parking position, or habitual speed; based on the target trajectories contained in multiple clusters, driving habit features in the traffic patterns corresponding to multiple clusters are extracted, including:

[0107] When driving habit features include a center trajectory, the center trajectory of each cluster is determined based on the location information of the trajectory points contained in the target trajectory within each cluster. When driving habit features include habitual lane-changing intervals, the habitual lane-changing intervals of each cluster are determined based on the lane information of the trajectory points contained in the target trajectory within each cluster. When driving habit features include habitual parking positions, the habitual parking positions of each cluster are determined based on the location information and vehicle speed information of the trajectory points contained in the target trajectory within each cluster. When driving habit features include habitual vehicle speeds, the habitual speeds of each cluster are determined based on the vehicle speed information of the trajectory points contained in the target trajectory within each cluster.

[0108] Specifically, for each cluster, when the driving habit features include the center trajectory, multiple target trajectories are resampled according to the position information of the trajectory points included in the target trajectory in the cluster, with the same number of points and spacing, to obtain multiple resampled trajectories; based on the multiple resampled trajectories, multiple sets of sampling points are determined, and the average of each set of sampling points is calculated to obtain the center sampling point; the multiple center sampling points are smoothed to obtain the center trajectory of the corresponding traffic mode of the cluster.

[0109] It should be noted that the center trajectory is determined based on multiple target trajectories in the cluster, representing the optimal trajectory of that cluster. The center trajectory is usually smoother than the map's geometric lane lines and is more in line with vehicle dynamics and human driving habits.

[0110] For each cluster, when driving habit features include habitual lane-change intervals, the lane-change start and end points of the target trajectory are determined based on the location information and lane information of the trajectory points included in the target trajectory within the cluster. Multiple lane-change start points of target trajectories are clustered to obtain multiple start-point clusters, with the cluster having the highest density being designated as the target start-point cluster. Similarly, multiple lane-change end points of target trajectories are clustered to obtain multiple end-point clusters, with the cluster having the highest density being designated as the target end-point cluster. The longitudinal range is determined based on the center coordinates of the target start-point and end-point clusters. The lateral range is determined based on the lane information of the target start-point and end-point clusters. The longitudinal and lateral ranges are then corrected based on the lane lines and stop lines of the road facility area to obtain the habitual lane-change intervals. The habitual lane-change intervals include the longitudinal intervals corresponding to the corrected longitudinal range and the lateral intervals corresponding to the corrected lateral range.

[0111] It should be noted that the habitual lane change interval is determined by analyzing the distribution of the starting and ending positions of vehicles initiating lane changes to enter the target lane. In lane-level navigation, it can serve as the optimal lane change suggestion interval, guiding vehicles to complete lane changes within a comfortable and safe range.

[0112] For each cluster, if driving habit features include habitual parking locations, parking points in the target trajectory are determined based on vehicle speed information of trajectory points included in the target trajectory within that cluster. The distribution of vehicle parking locations is then determined based on the location information of parking points across multiple target trajectories, thus obtaining the habitual parking locations. For example, based on the location information of parking points in multiple target trajectories, multiple parking points are clustered to obtain parking point clusters, and the center coordinates of these clusters are used as the habitual parking locations.

[0113] Optionally, a parking spot heat map can be generated based on the location information of parking spots in multiple target trajectories. The parking spot heat map can intuitively reflect parking events of vehicles passing through the road facility area.

[0114] It should be noted that in actual driving, when a vehicle stops due to traffic lights or to give way to other road users, the habitual parking location can reflect the parking situation of the vehicle when passing through the road facility area, providing lane-level navigation with a more accurate parking location that is more in line with human driving habits.

[0115] For each cluster, if the driving habit features include habitual vehicle speed, optionally, the maximum vehicle speed of the target trajectory is determined based on the vehicle speed information of the trajectory points included in the target trajectory within the cluster. The average of the maximum vehicle speeds of multiple target trajectories is taken as the habitual vehicle speed, which reflects the average maximum vehicle speed of a large number of vehicles traveling within the road facility area. Alternatively, based on the vehicle speed information of the trajectory points included in the target trajectory within the cluster, the vehicle speed change sequence of the target trajectory is extracted. The average of the vehicle speed change sequences of multiple target trajectories is then calculated to obtain the habitual vehicle speed, which reflects the speed changes of a large number of vehicles traveling within the road facility area. In practical applications, the habitual vehicle speed, which reflects the speed changes, can provide a typical speed change curve as a reference for comfortable cruising.

[0116] In some embodiments, driving habit characteristics include: center trajectory, habitual lane change range, habitual parking position, and habitual speed.

[0117] In related technologies, massive amounts of trajectory data from human drivers are collected, and machine learning models are used to learn driving patterns, enabling the generated trajectories to mimic human driving styles. However, this approach suffers from poor interpretability: the trajectory decision-making process generated by deep learning models is opaque, making it difficult to verify its safety and compliance, which is detrimental to functional safety certification. Furthermore, this approach lacks structured map constraints, and purely data-driven models may learn shortcuts or bad driving habits that violate traffic regulations or geographical constraints. It is also difficult to accurately align and integrate with existing high-precision map lane networks, and the complex neural network inference requires significant computing power, which may pose challenges for real-time operation on the vehicle.

[0118] In the above embodiments, the lane-level road network of the high-precision map is used as the basic topological constraint to ensure that all generated paths are physically and rule-wise feasible. Based on the target trajectory in the cluster corresponding to the traffic mode, the center trajectory, habitual lane change interval, habitual parking position and habitual speed under various traffic modes are fully explored, which improves the accuracy and comprehensiveness of driving habit features. In addition, the above driving habit features are used to support lane-level navigation services. The process is clear and interpretable. During real-time navigation on the vehicle, efficient querying and matching are performed through contextualized map data. The computational load is controllable, which improves the efficiency of lane-level navigation.

[0119] In a specific example, such as Figure 2 As shown, the method for generating contextualized map data includes:

[0120] S201. Slice the vehicle's driving trajectory based on the road facility area to obtain candidate trajectories. Based on the candidate trajectories, determine the scene trajectory of passing through the road facility area in a preset time period and in a preset direction.

[0121] S202. Perform clustering processing on the scene trajectory to obtain multiple clusters, and determine the traffic patterns corresponding to the multiple clusters respectively;

[0122] S203. Based on the location information of the trajectory points contained in the target trajectory in each cluster, determine the center trajectory of the traffic pattern corresponding to each cluster.

[0123] S204. Based on the lane information of the trajectory points contained in the target trajectory in each cluster, determine the habitual lane-changing intervals in the traffic mode corresponding to each cluster.

[0124] S205. Based on the location information and vehicle speed information of the trajectory points contained in the target trajectory in each cluster, determine the habitual parking position in the traffic mode corresponding to each cluster.

[0125] S206. Based on the vehicle speed information of the trajectory points contained in the target trajectory in each cluster, determine the habitual vehicle speed in the traffic mode corresponding to each cluster.

[0126] S207. Driving habit characteristics include: center trajectory, habitual lane change range, habitual parking position and habitual speed. Based on the driving habit characteristics under multiple traffic modes, a structured transformation is performed to obtain scenario-based map data of road traffic scenarios under multiple traffic modes.

[0127] S208. Obtain the road facility area corresponding to the road traffic scenario, and associate the scenario-based map data of the road traffic scenario under various traffic modes with the static map data based on the road facility area.

[0128] The method for generating scenario-based map data provided in this application utilizes massive amounts of real driving trajectories. Through data mining and fusion, it extracts driving habit features reflecting human driving habits and converts these features into scenario-based map data bound to specific road traffic scenarios. This transforms a large amount of typical driving experience from human drivers into quantifiable and reusable scenario-based map data, providing vehicles with more comfortable and human-like lane-level navigation services. Applying scenario-based map data to lane-level navigation improves the rigidity of planned routes, enhances the dynamic adaptability of route planning, and provides vehicles with more comfortable and human-like lane-level navigation services.

[0129] It should be understood that although the steps in the flowcharts of the above embodiments are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the above embodiments may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0130] Figure 3 This is a flowchart illustrating the lane-level navigation method provided in this application. The method for generating contextualized map data can be applied to electronic devices, which can be cloud-based. Figure 3 As shown, it includes:

[0131] Step 301: Obtain the vehicle's current road traffic scenario and traffic mode;

[0132] Step 302: Determine the contextualized map data based on the vehicle's current road traffic scenario and traffic mode, and extract scenario recommendation elements from the contextualized map data. The contextualized map data is generated using the contextualized map data generation method.

[0133] Step 303: Perform lane-level navigation based on static navigation elements and scene recommendation elements corresponding to the road traffic scenario.

[0134] Among them, the static navigation elements corresponding to the road traffic scenario are obtained from static map data.

[0135] Specifically, during the vehicle's journey, when the vehicle requests lane-level navigation to a destination, the current road traffic scenario and traffic mode are determined by the vehicle. The vehicle determines the traffic mode based on the current traffic conditions, for example, congestion mode, slow-moving mode, or smooth-flowing mode.

[0136] Optionally, the vehicle determines the road traffic scenario based on the current time, road facility area, and direction of travel, and determines the traffic mode based on the current traffic conditions; for example, if the vehicle determines that the current time is during the morning rush hour, the road facility area is an intersection, and the direction of travel is a turn from south to west, the road traffic scenario represents a scenario of turning left from south to west at the intersection during the morning rush hour.

[0137] The cloud obtains the vehicle's current road traffic scenario and traffic mode from the vehicle, searches for the corresponding scenario-based map data, and extracts scenario recommendation elements from the scenario-based map data. Scenario recommendation elements include at least one of the following: center trajectory, habitual lane-changing intervals, habitual parking positions, or habitual vehicle speeds.

[0138] The cloud determines the static navigation elements corresponding to the road traffic scenario based on static map data. Static navigation elements include: static geometric lane lines, lane type, lane-level speed limit, lane-level traffic rules, etc.

[0139] The route planning engine generates a target route based on static navigation elements and scene recommendation elements corresponding to the road traffic scenario, and sends the target route to the vehicle for execution to achieve lane-level navigation.

[0140] Optionally, lane-level navigation is performed based on static navigation elements and scene recommendation elements corresponding to the road traffic scenario, including: path planning based on static navigation elements corresponding to the road traffic scenario to obtain a static path; if the scene recommendation elements include a center trajectory, the static path and the center trajectory are merged; if the scene recommendation elements include a habitual lane-changing section, the lane-changing section in the static path is adjusted according to the habitual lane-changing section; if the scene recommendation elements include a habitual parking location, the static path is adjusted according to the habitual parking location so that the adjusted path includes the habitual parking location; if the scene recommendation elements include a habitual vehicle speed, the lane-level speed limit of the static path is adjusted to the habitual vehicle speed; the target path is obtained after the above adjustments.

[0141] In some embodiments, based on the vehicle's current road traffic scenario and traffic mode, scene recommendation elements are extracted from the scene-based map data, including: determining the road facility area based on the vehicle's current road traffic scenario; determining the target layer associated with the road facility area in the static map data; obtaining the scene-based map data of the road traffic scenario under the traffic mode in the target layer; and extracting scene recommendation elements from the scene-based map data, wherein the scene recommendation elements include at least one of: center trajectory, habitual lane change interval, habitual parking position, or habitual vehicle speed.

[0142] Specifically, since static map data and scenario-based map data are linked through road facility areas, a target layer associated with the road facility areas is determined in the static map data. This target layer includes multiple scenario-based map data points related to the road facility areas. Scenario-based map data for road traffic scenarios and traffic patterns are then obtained from this target layer. Scenario recommendation elements are then extracted from the scenario-based map data. These elements include at least one of the following: center trajectory, habitual lane-changing intervals, habitual parking locations, or habitual vehicle speeds.

[0143] In the above embodiments, the scene-based map can be quickly located by using road traffic scenarios and traffic modes, which improves the efficiency of lane-level navigation.

[0144] In related technologies, lane-level navigation is achieved based on static map data (such as static high-precision maps). However, the production and updating of static high-precision maps are costly and time-consuming, and cannot reflect changes in the actual road usage characteristics caused by construction, ground marking updates, etc. In contrast, the embodiments of this application analyze driving trajectories within a shorter period of time, mine and generate customized driving plans for each intersection, and abstract them into scenario-based map data, which is provided to the system as a navigation reference, making the planned navigation path more in line with the on-site traffic characteristics.

[0145] Furthermore, static map data cannot provide collective driving experience. Specifically, different drivers, or even the same driver in different scenarios (such as sparse or congested traffic flow, or day and night differences), will exhibit diverse trajectory choices, lane-changing timings, speed curves, and stopping points. For example, at large, unprotected left-turn intersections, human drivers often spontaneously choose a smoother, more observable turning trajectory based on oncoming traffic flow, rather than the rigid right-angle turning lane lines on the map. When choosing a lane before an intersection, drivers will adjust their lane changes earlier or later based on the path ahead, traffic speed, and their personal habits. Static high-precision maps and general algorithms cannot capture and reuse this collective driving experience, resulting in overly rigid path planning for autonomous vehicles. This leads to problems such as abrupt steering, unreasonable lane-changing timing, and inaccurate stopping positions, thereby reducing ride comfort, affecting traffic efficiency, and potentially posing safety risks in extreme cases due to behavioral patterns that do not conform to human expectations.

[0146] The lane-level navigation method provided in this application obtains the current road traffic scenario and traffic mode of the vehicle; based on the current road traffic scenario and traffic mode, it extracts scene recommendation elements from the scene-based map data, and performs lane-level navigation based on the static navigation elements corresponding to the road traffic scenario and the scene recommendation elements; since scene-based map data can reflect the collective driving habits of a large number of human drivers under different traffic modes in road traffic scenarios, applying scene-based map data to lane-level navigation and fusing it with static map data can generate a target path and speed plan that is not only passable but also more comfortable and more human-like, improving the dynamic adaptability of path planning and enhancing the human-likeness and accuracy of lane-level navigation.

[0147] In a specific example, the lane-level navigation method is applied to Figure 4 The lane-level navigation system shown includes a cloud and a vehicle. The cloud includes a cloud processing center and a service application layer, while the vehicle includes a vehicle application layer and a data acquisition layer.

[0148] The vehicle application layer sends a navigation request to the service application layer. The service application layer queries the cloud processing center for contextual map data through the map service API to obtain the scene recommendation elements, and also queries static map data to obtain static navigation elements (the query of static map data is not shown in the figure). The query results include scene recommendation elements and static navigation elements. The route planning engine of the service application layer generates a planned route based on the query results and sends it to the vehicle application layer.

[0149] The navigation controller receives the planned route from the service application layer and sends the corresponding instructions to the control actuator. The control actuator executes the new driving behavior. The vehicle's original trajectory is obtained by the on-board sensors of the data acquisition layer and uploaded to the cloud processing center after being processed by the vehicle compliance engine.

[0150] The cloud processing center receives a large number of anonymized raw trajectories, cleans and smooths them to obtain preprocessed trajectories (i.e. cleaned data), and matches them with high-precision maps to associate the discrete trajectories included in the preprocessed trajectories with the roads to obtain the vehicle's driving trajectory.

[0151] The behavior map mining engine performs contextual feature mining based on vehicle driving trajectories to obtain driving habit features (i.e., contextual features), and determines contextual map data based on these features. Driving habit features include:

[0152] Optionally, the data acquisition layer also includes a real-time lightweight clustering module. When a new road traffic scenario is detected by the real-time lightweight clustering module (such as the detection of a new detour scenario) or a high-confidence trajectory is detected, it is then processed by the vehicle-side compliance engine and uploaded to the cloud. This eliminates the need to upload the full trajectory, reducing data upload traffic and cloud storage and computing pressure, and improving data quality.

[0153] Optionally, the data acquisition layer can initiate trajectory recording and reporting when a discrepancy is detected between the trajectory and static map data (such as the detection of new obstacles or lane line changes), or when the comfort level is determined to be below a threshold. This changes the real-time continuous trajectory reporting to event-driven trajectory reporting, greatly reducing the amount of routine data and allowing system resources to be focused on handling problem scenarios, thereby improving the efficiency of lane-level navigation services.

[0154] like Figure 5 As shown, the scene-based map data generation apparatus 50 provided in this application embodiment includes:

[0155] The trajectory processing module 501 is used to determine the scene trajectory corresponding to the road traffic scenario based on the vehicle's driving trajectory.

[0156] The feature extraction module 502 is used to extract driving habit features under various traffic modes based on scene trajectories;

[0157] The generation module 503 is used to determine the scenario-based map data of the road traffic scene under various traffic modes based on the driving habit characteristics under various traffic modes. The scenario-based map data is used to support lane-level navigation of the road traffic scene under various traffic modes.

[0158] In some embodiments, the feature extraction module is used to perform clustering processing on the scene trajectory to obtain multiple clusters; determine the traffic patterns corresponding to the multiple clusters respectively; and extract driving habit features in the traffic patterns corresponding to the multiple clusters based on the target trajectories contained in the multiple clusters.

[0159] In some embodiments, the feature extraction module is configured to: determine the center trajectory of each cluster in the traffic pattern corresponding to each cluster based on the position information of the trajectory points contained in the target trajectory in each cluster when the driving habit features include a center trajectory; determine the habitual lane-changing interval in the traffic pattern corresponding to each cluster based on the lane information of the trajectory points contained in the target trajectory in each cluster when the driving habit features include a habitual parking position ...

[0160] In some embodiments, determining the scene trajectory corresponding to the road traffic scenario based on the vehicle's driving trajectory includes: slicing the vehicle's driving trajectory based on the road facility area to obtain candidate trajectories; and determining the scene trajectory of passing through the road facility area in a preset time period and a preset traffic direction based on the candidate trajectories.

[0161] In some embodiments, determining the scenario-based map data of the road traffic scene under multiple traffic modes based on the driving habit characteristics under multiple traffic modes includes: performing a structured transformation based on the driving habit characteristics under multiple traffic modes to obtain the scenario-based map data of the road traffic scene under multiple traffic modes.

[0162] Fourthly, such as Figure 6 As shown, the lane-level navigation device 60 provided in this application embodiment includes:

[0163] The acquisition module 601 is used to acquire the current road traffic scenario and traffic mode of the vehicle;

[0164] The scenario-based map data processing module 602 is used to extract scenario recommendation elements from the scenario-based map data based on the current road traffic scenario and traffic mode of the vehicle. The scenario-based map data is generated by the scenario-based map data generation method.

[0165] Lane-level navigation module 603 is used for lane-level navigation based on static navigation elements and scene recommendation elements corresponding to road traffic scenarios.

[0166] In some embodiments, the scenario-based map data processing module is used to determine the road facility area based on the current road traffic scenario of the vehicle; determine the target layer associated with the road facility area in the static map data; obtain the scenario-based map data of the road traffic scenario in the traffic mode in the target layer; and extract scenario recommendation elements from the scenario-based map data, wherein the scenario recommendation elements include at least one of: center trajectory, habitual lane change interval, habitual parking position or habitual speed.

[0167] Figure 7 A schematic diagram of the structure of the electronic device provided in this application. Figure 7 As shown, the electronic device 70 provided in this embodiment includes at least one processor 701 and a memory 702. Optionally, the device 70 further includes a communication component 703. The processor 701, memory 702, and communication component 703 are connected via a bus.

[0168] In a specific implementation, at least one processor 701 executes computer execution instructions stored in memory 702, causing at least one processor 701 to perform the above-described method.

[0169] The specific implementation process of processor 701 can be found in the above method embodiments, and its implementation principle and technical effect are similar. It will not be repeated here.

[0170] In the above embodiments, it should be understood that the processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), etc. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the method disclosed in this invention can be directly implemented by a hardware processor, or implemented by a combination of hardware and software modules within the processor.

[0171] The memory may include random access memory (RAM) and may also include non-volatile memory (NVM), such as at least one disk storage device.

[0172] The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. Buses can be categorized as address buses, data buses, control buses, etc. For ease of illustration, the buses shown in the accompanying drawings are not limited to a single bus or a single type of bus.

[0173] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the above-described method.

[0174] This application also provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the above-described method.

[0175] The aforementioned readable storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. The readable storage medium can be any available medium accessible to a general-purpose or special-purpose computer.

[0176] An exemplary readable storage medium is coupled to a processor, enabling the processor to read information from and write information to the readable storage medium. Of course, the readable storage medium can also be a component of the processor. The processor and the readable storage medium can reside in an Application Specific Integrated Circuit (ASIC). Alternatively, the processor and the readable storage medium can exist as discrete components in the device.

[0177] The division of units is merely a logical functional division; in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be indirect coupling or communication connection through some interfaces, devices, or units, and may be electrical, mechanical, or other forms.

[0178] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0179] In addition, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0180] If a function is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0181] Those skilled in the art will understand that all or part of the steps of the above-described method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When executed, the program performs the steps of the above-described method embodiments; and the aforementioned storage medium includes various media capable of storing program code, such as ROM, RAM, magnetic disks, or optical disks.

[0182] Finally, it should be noted that other embodiments of the invention will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This invention is intended to cover any variations, uses, or adaptations of the invention that follow the general principles of the invention and include common knowledge or customary techniques in the art not disclosed herein, and is not limited to the precise structures described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of the invention is limited only by the appended claims.

Claims

1. A method for generating contextualized map data, characterized in that, include: Based on the vehicle's driving trajectory, determine the scene trajectory corresponding to the road traffic scenario; Based on the scene trajectory, driving habit features under various traffic modes are extracted; Based on driving habit characteristics under various traffic modes, scenario-based map data for the road traffic scenario under various traffic modes is determined. The scenario-based map data is used to support lane-level navigation for the road traffic scenario under various traffic modes.

2. The method according to claim 1, characterized in that, Based on the scene trajectory, driving habit features under various traffic modes are extracted, including: The scene trajectories are clustered to obtain multiple clusters; Determine the traffic patterns corresponding to the multiple clusters; Based on the target trajectories contained in the multiple clusters, driving habit features in the traffic patterns corresponding to the multiple clusters are extracted.

3. The method according to claim 2, characterized in that, The driving habit characteristics include at least one of the following: center trajectory, habitual lane change range, habitual parking position, or habitual speed; The step of extracting driving habit features in the traffic patterns corresponding to the multiple clusters based on the target trajectories contained in the multiple clusters includes: When the driving habit features include a center trajectory, the center trajectory of each cluster is determined according to the position information of the trajectory points contained in the target trajectory in each cluster. When the driving habit features include habitual lane change intervals, the habitual lane change intervals in the traffic mode corresponding to each cluster are determined based on the lane information of the trajectory points contained in the target trajectory in each cluster. When the driving habit features include a habitual parking location, the habitual parking location in the traffic pattern corresponding to each cluster is determined based on the location information and vehicle speed information of the trajectory points contained in the target trajectory in each cluster. When the driving habit features include habitual vehicle speed, the habitual vehicle speed in the traffic mode corresponding to each cluster is determined based on the vehicle speed information of the trajectory points contained in the target trajectory in each cluster.

4. The method according to any one of claims 1 to 3, characterized in that, The road traffic scenario is a driving scenario in which a road facility area is passed through during a preset time period and in a preset direction of travel. The step of determining the scene trajectory corresponding to the road traffic scenario based on the vehicle's driving trajectory includes: Candidate trajectories are obtained by slicing the vehicle's trajectory based on the road facility area; Based on candidate trajectories, a scenario trajectory is determined that passes through the road facility area in a preset time period and in a preset direction.

5. The method according to any one of claims 1 to 3, characterized in that, The process of determining the scenario-based map data for the road traffic scene under multiple traffic modes based on driving habit characteristics under various traffic modes includes: Based on the driving habit characteristics under various traffic modes, a structured transformation is performed to obtain the scenario-based map data of the road traffic scenario under various traffic modes; The method further includes: Obtain the road facility area corresponding to the road traffic scenario; Based on the road facility area, the scenario-based map data of the road traffic scene under various traffic modes is associated with the static map data.

6. A lane-level navigation method, characterized in that, include: Obtain the vehicle's current road traffic scenario and traffic mode; Based on the vehicle's current road traffic scenario and traffic mode, scenario recommendation elements are extracted from the scenario-based map data, wherein the scenario-based map data is generated based on the method of any one of claims 1 to 5; Lane-level navigation is performed based on the static navigation elements corresponding to the road traffic scenario and the scenario recommendation elements.

7. The method according to claim 6, characterized in that, The extraction of scene recommendation elements from the scene-based map data based on the vehicle's current road traffic scenario and traffic pattern includes: Determine the area of ​​road facilities based on the current road traffic scenario of the vehicle; Determine the target layer associated with the road facility area in the static map data; Obtain the scene-based map data of the road traffic scene in the traffic mode from the target layer; Scene recommendation elements are extracted from contextualized map data. These elements include at least one of the following: center trajectory, habitual lane-changing intervals, habitual parking locations, or habitual vehicle speeds.

8. A device for generating scenario-based map data, characterized in that, The device includes: The trajectory processing module is used to determine the scene trajectory corresponding to the road traffic scenario based on the vehicle's driving trajectory. The feature extraction module is used to extract driving habit features under various traffic modes based on the scene trajectory. The generation module is used to determine the scenario-based map data of the road traffic scene under multiple traffic modes based on the driving habit characteristics under multiple traffic modes. The scenario-based map data is used to support lane-level navigation of the road traffic scene under multiple traffic modes.

9. A lane-level navigation device, characterized in that, The device includes: The acquisition module is used to acquire the current road traffic scenario and traffic mode of the vehicle; A scenario-based map data processing module is used to extract scenario recommendation elements from scenario-based map data based on the current road traffic scenario and traffic mode of the vehicle, wherein the scenario-based map data is generated based on the method of any one of claims 1 to 5; The lane-level navigation module is used to perform lane-level navigation based on the static navigation elements corresponding to the road traffic scenario and the scenario recommendation elements.

10. An electronic device / computer-readable storage medium / computer program product, characterized in that, The electronic device includes: a processor, and a memory communicatively connected to the processor; the memory stores computer-executable instructions; the processor executes the computer-executable instructions stored in the memory to implement the method as described in any one of claims 1 to 7; The computer-readable storage medium stores computer-executable instructions, which, when executed by a processor, are used to implement the method as described in any one of claims 1 to 7; The computer program product includes computer execution instructions that, when executed by a processor, implement the method as described in any one of claims 1 to 7.

Images