Virtual lane line construction method, electronic device, and computer program product

Abstract

This disclosure provides a method for constructing virtual lane lines, an electronic device, and a computer program product. The method for constructing virtual lane lines includes: acquiring trajectory data of at least two candidate vehicles surrounding the vehicle within a target time period for a target road; determining the trajectory consistency between any two candidate vehicles based on the trajectory data of each candidate vehicle; analyzing the trajectory consistency between any two candidate vehicles and selecting target vehicles that meet the trajectory consistency screening threshold from the candidate vehicles; performing centerline fitting on the trajectory data of the target vehicles based on the road curvature of the target road to determine the lane centerline; determining the lane width and lane boundary line by analyzing the lateral envelope of the target vehicle's travel path based on the spatial distribution relationship between the target vehicle's trajectory data and the lane centerline; and constructing a virtual lane line based on the lane centerline, lane width, and lane boundary line.

Description

Technical Field

[0001] This disclosure relates to a method for constructing virtual lane lines, electronic devices, storage media, and computer program products. Background Technology

[0002] In complex road scenarios such as rural roads, old urban areas, or construction zones, physical lane markings are often blurred, incomplete, or even completely missing, rendering traditional visual lane marking recognition methods ineffective. Existing technologies attempt to build high-precision maps offline using GPS trajectory big data, but this method is costly and difficult to update in real time; other solutions rely on the driving trajectory of a single vehicle ahead as a following benchmark, but when the target vehicle changes lanes or is driving unstable, it can easily lead to misjudgment and danger for the vehicle itself. Summary of the Invention

[0003] This disclosure provides a method for constructing virtual lane lines, an electronic device, and a computer program product.

[0004] According to one aspect of this disclosure, a method for constructing virtual lane lines is provided, comprising: acquiring trajectory data of at least two candidate vehicles surrounding a vehicle within a target time period for a target road; determining trajectory consistency between any two candidate vehicles based on the trajectory data of each candidate vehicle, the trajectory consistency including lateral position consistency and heading angle consistency between the two candidate vehicles; analyzing the trajectory consistency between any two candidate vehicles and selecting target vehicles that meet the trajectory consistency screening threshold from the candidate vehicles; performing centerline fitting on the trajectory data of the target vehicles based on the road curvature of the target road to determine a lane centerline; determining lane width and lane boundary lines by analyzing the lateral envelope range of the target vehicle's driving path based on the spatial distribution relationship between the trajectory data of the target vehicles and the lane centerline; and constructing virtual lane lines based on the lane centerline, the lane width, and the lane boundary lines.

[0005] According to the virtual lane line construction method disclosed herein, trajectory data of at least two candidate vehicles around the vehicle are dynamically acquired and analyzed under the target road. The consistency of the lateral position and heading angle of the candidate vehicles is comprehensively evaluated, enabling accurate identification and elimination of abnormal vehicles such as those with unstable driving or intentional lane changes. By establishing a centerline fitting mechanism that adapts to the road curvature and combining the distribution characteristics of the target vehicle trajectory relative to the centerline to determine the lane boundary and width, a high-precision and highly stable virtual lane line is constructed. This effectively overcomes the shortcomings of single vehicle trajectories being susceptible to interference and traditional methods being unable to dynamically adapt to complex road conditions. It significantly improves the reliability of path planning, real-time response capability, and driving safety of the intelligent driving system in scenarios without lane markings, ensuring the safety and continuity of lateral control.

[0006] According to at least one embodiment of the virtual lane line construction method of this disclosure, before determining the trajectory consistency between every two candidate vehicles based on the trajectory data of each candidate vehicle, the method includes: calculating the difference between the lateral positions of the candidate vehicles at adjacent time points based on the trajectory data of the candidate vehicles, and determining the lateral position difference of each candidate vehicle; and filtering each candidate vehicle based on the lateral position difference of each candidate vehicle to determine the candidate vehicles used to determine trajectory consistency.

[0007] According to at least one embodiment of the virtual lane line construction method of this disclosure, before selecting target vehicles that meet the trajectory consistency screening threshold from the candidate vehicles based on trajectory consistency analysis between every two candidate vehicles, the method includes: calculating the heading angle change, lateral velocity, and lateral displacement change of the candidate vehicles within a sampling period based on the trajectory data of the candidate vehicles; and screening each candidate vehicle based on the heading angle change, lateral velocity, and lateral displacement change of the candidate vehicles to determine the candidate vehicles used for screening through the trajectory consistency.

[0008] According to at least one embodiment of the virtual lane line construction method of this disclosure, the trajectory consistency between every two candidate vehicles is analyzed, and target vehicles that meet the trajectory consistency screening threshold are selected from the candidate vehicles, including: when the number of candidate vehicles is greater than or equal to a target number, the candidate vehicles whose trajectory consistency is greater than or equal to a first threshold are selected as target vehicles; when the number of candidate vehicles is less than the target number, the candidate vehicles whose trajectory consistency is greater than or equal to a second threshold are selected as target vehicles; the first threshold is greater than the second threshold.

[0009] According to at least one embodiment of the virtual lane line construction method of this disclosure, based on the road curvature of the target road, centerline fitting is performed on the trajectory data of the target vehicle to determine the lane centerline, including: when the road curvature of the target road is less than or equal to the target curvature, global fitting is performed on the trajectory data of the target vehicle to determine the lane centerline; when the road curvature of the target road is greater than the target curvature, segmented fitting is performed on the trajectory data of the target vehicle to determine the lane centerline.

[0010] According to at least one embodiment of the virtual lane line construction method of this disclosure, based on the trajectory data of the target vehicle and the spatial distribution relationship of the lane centerline, the lane width is determined by analyzing the lateral envelope range of the target vehicle's driving path, including: for the longitudinal range under the target road, performing median statistics on the trajectory data of the lane centerline and the target vehicles located on the left and right sides of the vehicle to determine the lane width.

[0011] According to at least one embodiment of the virtual lane line construction method of this disclosure, based on the trajectory data of the target vehicle and the spatial distribution relationship of the lane centerline, the lane boundary line is determined by analyzing the lateral envelope range of the target vehicle's driving path, including: calculating the lateral distances between the target vehicle located to the left of the vehicle and the target vehicle located to the right of the vehicle and the lane centerline, respectively, based on the trajectory data of the target vehicle, and determining the left boundary line and the right boundary line; and smoothing the left boundary line and the right boundary line to determine the lane boundary line.

[0012] According to at least one embodiment of the virtual lane line construction method of this disclosure, based on the trajectory data of the target vehicle and the spatial distribution relationship of the lane centerline, the lane boundary line is determined by analyzing the lateral envelope range of the target vehicle's driving path, including: identifying the lateral position of the road edge of the target road; and determining the lane boundary line based on the lateral position of the road edge and the lane width.

[0013] The virtual lane line construction method according to at least one embodiment of the present disclosure further includes: when the number of target vehicles or the trajectory consistency between candidate vehicles changes, acquiring the trajectory data of the candidate vehicles to reconstruct the virtual lane line.

[0014] According to at least one embodiment of the virtual lane line construction method of this disclosure, before acquiring trajectory data of at least two candidate vehicles surrounding the vehicle within a target time period for a target road, the method includes: acquiring image data of the road; identifying physical lane lines in the image data and calculating a confidence score for the physical lane lines; and determining the road as the target road when the confidence score of the physical lane lines is less than a confidence threshold.

[0015] According to at least one embodiment of the virtual lane line construction method of this disclosure, determining the trajectory consistency between every two candidate vehicles based on the trajectory data of each candidate vehicle includes: This indicates the trajectory consistency between candidate vehicle i and candidate vehicle j. and Indicates the weighting coefficient. The Pearson correlation coefficient represents the historical lateral position sequence between candidate vehicle i and candidate vehicle j. The heading angle of the candidate vehicle i is represented. The heading angle of the candidate vehicle j is represented. This represents the maximum heading angle difference.

[0016] According to another aspect of this disclosure, an electronic device is provided, comprising: a memory storing execution instructions; and a processor executing the execution instructions stored in the memory, such that the processor performs a virtual lane line construction method according to any embodiment of this disclosure.

[0017] According to another aspect of this disclosure, a readable storage medium is provided, wherein executable instructions are stored therein, which, when executed by a processor, are used to implement the virtual lane line construction method of any embodiment of this disclosure.

[0018] According to another aspect of this disclosure, a computer program product is provided, including a computer program that, when executed by a processor, implements a virtual lane line construction method according to any embodiment of this disclosure. Attached Figure Description

[0019] The accompanying drawings illustrate exemplary embodiments of the present disclosure and, together with the description thereof, serve to explain the principles of the present disclosure. These drawings are included to provide a further understanding of the present disclosure and are incorporated in and constitute a part of this specification.

[0020] Figure 1 This is a schematic diagram of the overall process of a virtual lane line construction method according to one embodiment of the present disclosure.

[0021] Figure 2 This is a flowchart illustrating the process of determining a target road in a virtual lane line construction method according to one embodiment of the present disclosure.

[0022] Figure 3 This is a flowchart illustrating the process of determining candidate vehicles for trajectory consistency in a virtual lane line construction method according to one embodiment of the present disclosure.

[0023] Figure 4 This is a schematic diagram of the process for determining candidate vehicles for screening by trajectory consistency in a virtual lane line construction method according to one embodiment of the present disclosure.

[0024] Figure 5 This is a flowchart illustrating the process of determining lane boundary lines in a virtual lane line construction method according to one embodiment of the present disclosure.

[0025] Figure 6 This is a flowchart illustrating the process of determining lane boundary lines in a virtual lane line construction method according to one embodiment of the present disclosure.

[0026] Figure 7 This is a schematic structural block diagram of a virtual lane line construction device according to one embodiment of the present disclosure.

[0027] Figure 8This is a schematic structural block diagram of an electronic device according to one embodiment of the present disclosure. Detailed Implementation

[0028] The present disclosure will now be described in further detail with reference to the accompanying drawings and examples. It should be understood that the specific examples described herein are for illustrative purposes only and are not intended to limit the scope of the disclosure. Furthermore, it should be noted that, for ease of description, only the parts relevant to the present disclosure are shown in the accompanying drawings.

[0029] It should be noted that, where there is no conflict, the embodiments and features described in this disclosure can be combined with each other. The technical solutions of this disclosure will now be described in detail with reference to the accompanying drawings and embodiments.

[0030] In complex traffic scenarios such as rural sidewalks, old urban roads, and temporary construction sections where physical lane lines are severely lacking or completely invisible, intelligent driving systems face the risk of lateral control inaccuracy due to the lack of a stable driving benchmark. Relying solely on the real-time trajectory of the vehicle itself can easily cause lane lines to drift as the vehicle moves laterally. Failure to fully consider the coordination and consistency of the historical trajectories of multiple vehicles can cause lane-changing vehicles or unstable vehicles to interfere with the generation of virtual lane lines.

[0031] To address this issue, this disclosure proposes a virtual lane line construction method. By acquiring trajectory data of candidate vehicles surrounding the vehicle, the method evaluates the degree of trajectory coordination among candidate vehicles from two dimensions: lateral position consistency and heading angle consistency. This effectively identifies and eliminates abnormal vehicles caused by lane changes or unstable driving, selecting target vehicles with consistent direction and stable driving as reliable benchmarks. Furthermore, a lane centerline is adaptively generated based on road curvature, and the lane width and boundary lines are dynamically calculated based on the target vehicle trajectory data. Ultimately, a virtual lane line that highly matches the geometry of the real road is constructed. This provides a stable and accurate driving benchmark for the intelligent driving system in scenarios without physical lane lines, effectively suppressing single-vehicle trajectory drift and interference from abnormal vehicles, significantly improving the accuracy of lateral control and driving safety.

[0032] The virtual lane line construction method disclosed herein can be directly embedded into the domain controller of passenger vehicles, commercial vehicles, or unmanned delivery vehicles with L2 or higher autonomous driving functions. Utilizing onboard millimeter-wave radar and cameras, it perceives the trajectories of surrounding vehicles in real time and independently completes the dynamic construction of virtual lane lines, providing a critical path reference for the vehicle's adaptive cruise control (ACC) and lane keeping assist (LKA). Alternatively, massive amounts of anonymized multi-vehicle trajectory data can be uploaded to the cloud. The server uses the virtual lane line construction method disclosed herein to perform big data analysis on the historical traffic flow of specific unmarked lane segments, generating and continuously iterating reference virtual lane line models for that segment, which are then deployed to the vehicle as prior knowledge. Furthermore, in intelligent traffic management scenarios, roadside units (RSUs) can deploy the virtual lane line construction method disclosed herein to analyze intersection traffic flow trajectories in real time, providing data support for traffic light timing optimization or broadcasting temporary virtual guide lines to connected vehicles.

[0033] Figure 1 A schematic diagram illustrating the overall flow of a virtual lane line construction method according to one embodiment of this disclosure is shown. Figure 1 The method M100 shown includes steps S110 to S160. This method can be implemented by an on-board controller.

[0034] In step S110, for the target road, the trajectory data of at least two candidate vehicles around the vehicle are obtained within the target time period.

[0035] A multi-sensor system consisting of front radar, corner radar, and front / side-view cameras continuously scans the area in front of the vehicle and adjacent lanes to identify and track candidate vehicles within the field of view. For each tracked candidate vehicle, its position coordinates (i.e., longitudinal distance and lateral offset), speed, heading angle, and corresponding timestamp are recorded relative to the vehicle at a target sampling period. These continuously collected trajectory data points are then used to construct a historical trajectory data sequence that includes a time dimension.

[0036] The target road is a road scenario in which the physical lane lines cannot be effectively identified by the vehicle-mounted sensors due to wear, coverage, or absence, and there are other motor vehicles that can be tracked.

[0037] Preferably, the target time period is 5 seconds before the current time, so as to obtain the complete trajectory data of each candidate vehicle within the effective time window.

[0038] In step S120, based on the trajectory data of each candidate vehicle, the trajectory consistency between every two candidate vehicles is determined. The trajectory consistency includes the lateral position consistency and heading angle consistency between every two candidate vehicles.

[0039] In one specific embodiment, the trajectory consistency between any two candidate vehicles is determined based on the trajectory data of each candidate vehicle, wherein the trajectory consistency between candidate vehicles is calculated using the following formula: This indicates the trajectory consistency between candidate vehicle i and candidate vehicle j. and Indicates the weighting coefficient. The Pearson correlation coefficient represents the historical lateral position sequence between candidate vehicle i and candidate vehicle j. Let represent the heading angle of candidate vehicle i. Let represent the heading angle of candidate vehicle j. This represents the maximum heading angle difference.

[0040] In step S130, the trajectory consistency between every two candidate vehicles is analyzed, and target vehicles that meet the trajectory consistency screening threshold are selected from the candidate vehicles.

[0041] Preferably, when the number of candidate vehicles is greater than or equal to the target number, candidate vehicles with trajectory consistency greater than or equal to a first threshold are selected as target vehicles; when the number of candidate vehicles is less than the target number, candidate vehicles with trajectory consistency greater than or equal to a second threshold are selected as target vehicles; the first threshold is greater than the second threshold. By setting dynamic filtering thresholds, adaptive selection of target vehicles under different traffic densities is achieved. When there are sufficient vehicles around the vehicle, a stricter first threshold is used, thereby ensuring high quality of data sources and high accuracy of virtual lane lines. When there are few vehicles around the vehicle, the standard is automatically relaxed, and a lower second threshold is used to retain more referenceable trajectory data, improving adaptability to complex and changing traffic environments.

[0042] Specifically, when the number of candidate vehicles is greater than or equal to 3, that is, there are two groups of candidate vehicles, and the trajectory consistency score of at least two groups of candidate vehicles is greater than or equal to 0.7, then the two groups of candidate vehicles with the highest trajectory consistency score are selected as the target vehicles; when the number of candidate vehicles is less than 3, and there is a group of candidate vehicles with a trajectory consistency score greater than or equal to 0.6, then that group of candidate vehicles is selected as the target vehicles.

[0043] In one specific embodiment, when the number of candidate vehicles is 1 and the candidate vehicle is located in front of the vehicle, the candidate vehicle is selected as the target vehicle if the candidate vehicle changes its heading angle by less than or equal to 5°, its speed fluctuation by less than or equal to 5 km / h, and its trajectory straightness deviation by less than or equal to 0.3m within 5 seconds.

[0044] In step S140, the centerline of the target vehicle's trajectory data is fitted based on the road curvature of the target road to determine the lane centerline.

[0045] Based on the complexity of the current road geometry, the optimal mathematical modeling strategy is dynamically selected, and the trajectories of the selected target vehicle group are merged into a reference path that can accurately describe the road direction and has good smoothness, which serves as the lane centerline of the virtual lane.

[0046] In one specific embodiment, when the curvature of the target road is less than or equal to the target curvature, the trajectory data of the target vehicle is globally fitted to determine the lane centerline; when the curvature of the target road is greater than the target curvature, the trajectory data of the target vehicle is segmented and fitted to determine the lane centerline.

[0047] Preferably, the trajectory data of the target vehicle is globally fitted using cubic polynomial fitting to determine the lane centerline.

[0048] Preferably, the trajectory data of the target vehicle is segmented and fitted using third-order B-spline fitting to determine the lane centerline.

[0049] Preferably, trajectory data with high historical trajectory data point density and high trajectory consistency score are used for fitting.

[0050] In step S150, based on the trajectory data of the target vehicle and the spatial distribution relationship of the lane centerline, the lane width and lane boundary line are determined by analyzing the lateral envelope range of the target vehicle's driving path.

[0051] The lateral envelope of the aforementioned target vehicle's travel path is the maximum spatial boundary covered by the travel trajectories of all target vehicles in the lateral (left-right) direction of the road.

[0052] Using the merged lane centerline as a reference, the lateral offset distances of historical trajectory points of target vehicles located to the left and right of the current vehicle relative to the lane centerline are statistically analyzed. Statistical methods are then used to determine the safe passage boundaries on both sides, and the lane width is calculated accordingly. Based on the lane width and the trajectory data of the target vehicles, the lane boundary lines are determined.

[0053] Preferably, after generating the lane boundary line, the distance between the lane boundary line and the road edge is checked to see if it is greater than or equal to the edge threshold. If not, the lane width is adjusted.

[0054] Preferably, when the target vehicle consists of only one vehicle on the side relative to the vehicle, or when the number of target vehicles is 1, the lane width is 3.75m by default.

[0055] In one specific embodiment, the lane width is determined by statistically analyzing the median of the trajectory data of the lane centerline and the target vehicles located on the left and right sides of the vehicle within the longitudinal range of the target road.

[0056] In step S160, a virtual lane line is constructed based on the lane center line, lane width, and lane boundary line.

[0057] The calculated three core curves—the lane centerline, left boundary line, and right boundary line—along with the lane width value used as reference information, are encapsulated and formatted to form a virtual lane line with a unified spatiotemporal reference.

[0058] By acquiring dynamic trajectory data of surrounding candidate vehicles through multi-sensor fusion in target road scenarios where physical lane lines are missing, and introducing trajectory consistency scores based on both lateral position and heading angle, this method achieves accurate identification and elimination of abnormal vehicles such as those exhibiting unstable driving or making temporary lane changes, thereby filtering out target vehicles with reliable behavior representing the mainstream traffic flow. A high-precision lane centerline is generated by adaptively selecting a global or segmented fitting strategy based on road curvature, and the lane width and boundary lines are dynamically determined based on the statistical distribution of target vehicle trajectories relative to the centerline, ultimately constructing a complete virtual lane. This method effectively overcomes the interference of single-vehicle trajectory drift and mixed noise in multi-vehicle data, significantly improving the accuracy and real-time performance of virtual lane line construction in complex scenarios such as rural sidewalks and old urban areas.

[0059] Regarding step S110, before acquiring trajectory data of at least two candidate vehicles surrounding the vehicle within the target time period for the target road, some embodiments of this disclosure may include, for example... Figure 2 Steps S210 to S230 are shown.

[0060] In step S210, image data of the road is acquired.

[0061] In step S220, physical lane lines in the image data are identified, and a confidence score for the physical lane lines is calculated.

[0062] In one specific embodiment, the acquired road image data is input into a pre-trained deep learning semantic segmentation model to classify each pixel in the image and output a probability map containing lane line categories. The probability map is then post-processed, using thresholding and curve fitting algorithms to extract one or more potential physical lane line geometric parameters (such as polynomial coefficients). A confidence score for the physical lane lines is then calculated, weighted by multiple dimensions, including the original average probability value output by the model, the continuity and smoothness of the fitted curve, the consistency between the current frame result and the trajectory of short-term historical frames, and the effective length occupied by the lane line in the image. Finally, a comprehensive confidence score is obtained to quantify the reliability of lane line visibility in the current environment.

[0063] In step S230, when the confidence score of the physical lane line is less than the confidence threshold, the road is determined as the target road.

[0064] Preferably, the confidence threshold is 0.3.

[0065] By acquiring road images and identifying physical lane lines, a multi-dimensional weighted confidence scoring mechanism is introduced, which can comprehensively quantify and evaluate the reliability of lane line detection results. This effectively avoids decision-making errors caused by misjudgment due to a single indicator, and ensures the functional safety and continuity of the autonomous driving system in complex scenarios.

[0066] Regarding step S120, before determining the trajectory consistency between any two candidate vehicles based on the trajectory data of each candidate vehicle, some embodiments of this disclosure may include, for example... Figure 3 Steps S310 to S320 are shown.

[0067] In step S310, based on the trajectory data of the candidate vehicles, the difference between the lateral positions of the candidate vehicles at adjacent time points is calculated to determine the lateral position difference of each candidate vehicle.

[0068] In step S320, each candidate vehicle is screened based on the lateral position difference of each candidate vehicle to determine the candidate vehicle used to determine trajectory consistency.

[0069] In one specific embodiment, for adjacent time points, if the lateral position difference of each candidate vehicle is greater than 0.5m or the longitudinal deviation is greater than 2m, the trajectory data of that candidate vehicle at the current moment is determined to be a trajectory jump point, and the trajectory jump point is removed. For the trajectory data of candidate vehicles, if the proportion of trajectory jump points of any candidate vehicle is greater than 10%, the candidate vehicle is determined to be a detected erroneous vehicle, and the entire candidate vehicle is excluded.

[0070] Preferably, for the trajectory data of each candidate vehicle, when there is no valid trajectory data at a certain time point, linear interpolation of adjacent valid points is used to complete the trajectory to ensure the continuity of the trajectory.

[0071] By calculating the lateral and longitudinal position differences of candidate vehicles at adjacent time points, accurate identification and purification of sensor detection anomalies are achieved. This effectively detects and eliminates trajectory jumps caused by radar or camera target matching errors. By setting a threshold for the proportion of jump points, vehicles with consistently low perception quality are excluded from the data source. Simultaneously, a linear interpolation strategy is used to fill in temporary data gaps, ensuring the continuity and integrity of effective trajectories.

[0072] Regarding step S130, before filtering every two candidate vehicles based on the trajectory consistency between candidate vehicles to determine the target vehicle, in some embodiments of this disclosure, it may include, as follows: Figure 4 Steps S410 to S420 are shown.

[0073] In step S410, based on the trajectory data of the candidate vehicles, the changes in heading angle, lateral velocity, and lateral displacement of the candidate vehicles within the sampling period are calculated.

[0074] In step S420, each candidate vehicle is screened based on the changes in heading angle, lateral velocity, and lateral displacement of the candidate vehicles to determine the candidate vehicles to be screened through trajectory consistency.

[0075] Preferably, if a candidate vehicle experiences a sudden change in heading angle greater than 10° / s within a sampling period, or if the cumulative change in heading angle exceeds 8° within multiple consecutive sampling periods, the candidate vehicle is directly determined to be an abnormal vehicle and is excluded.

[0076] Preferably, if a candidate vehicle's lateral speed is consistently greater than 1 m / s and its lateral displacement changes by more than 0.8 m within 500 ms, and this causes its trajectory consistency score with other candidate vehicles to drop sharply by more than 30%, then the candidate vehicle is determined to be a lane-changing vehicle and is immediately excluded.

[0077] Preferably, when the number of candidate vehicles is greater than or equal to 3, if the trajectory consistency score of a candidate vehicle with the other 2 or more candidate vehicles is less than 0.6, then the candidate vehicle is determined to be an abnormal vehicle and is excluded.

[0078] In one specific embodiment, after excluding vehicles that temporarily change lanes, monitoring continues for 2 seconds. If, after 2 seconds, the average difference between the heading angle of the temporary lane-changing vehicle and other candidate vehicles is less than or equal to 3° and its lateral position is stable, the trajectory consistency score of the temporary lane-changing vehicle is recalculated. If the trajectory consistency score meets the standard, the temporary lane-changing vehicle is retained.

[0079] By comprehensively analyzing the dynamic characteristics of candidate vehicles, such as changes in heading angle, lateral velocity, and displacement, and combining them with their trajectory consistency scores with other candidate vehicles, the system effectively identifies and eliminates unstable vehicles, vehicles that temporarily change lanes, and abnormal vehicles whose trajectories deviate from those of other vehicles. This process removes unreliable data sources that may introduce interference before trajectory consistency calculation, ensuring that candidate vehicles entering the subsequent consistency evaluation have stable driving states and reliable trajectory quality. This improves the accuracy of virtual lane line construction and reduces the risk of lane line drift caused by abnormal behavior of a single vehicle or instantaneous lane changes.

[0080] Regarding step S150, based on the trajectory data of the target vehicle and the spatial distribution relationship of the lane centerline, the lane boundary line is determined by analyzing the lateral envelope range of the target vehicle's driving path. In some embodiments of this disclosure, this may include, for example... Figure 5 Steps S510 to S520 are shown.

[0081] In step S510, based on the trajectory data of the target vehicle, the lateral distances between the target vehicle located to the left of the vehicle and the target vehicle located to the right of the vehicle and the center line of the lane are calculated, and the left boundary line and the right boundary line are determined.

[0082] The effective calculation interval for the longitudinal distance range is determined, and all historical trajectory data points of target vehicles located to the left and right of the current vehicle within the effective calculation interval are selected. For target vehicles on the left and right, the lateral distance from each trajectory data point to the generated lane centerline is calculated, and the distance values ​​are statistically analyzed to determine the left and right boundary lines of the virtual lane.

[0083] Preferably, for the historical trajectory data points of the target vehicle on the left or right, the 75th percentile is taken as the offset of the left or right boundary line relative to the lane center line, and the left and right boundary lines are calculated based on the offset.

[0084] In step S520, the left and right boundary lines are smoothed to determine the lane boundary lines.

[0085] Preferably, a Gaussian filter with a standard deviation of 0.3m is used to smooth the left and right boundary lines.

[0086] By statistically analyzing the lateral distribution characteristics of target vehicle trajectory points on both sides of the vehicle relative to the lane centerline, and using the 75th percentile method to determine the boundary lines, a virtual lane boundary that balances mainstream traffic habits with safety redundancy is effectively constructed. This avoids incorrect compression of the boundary due to individual vehicles crossing the line, ensuring the rationality of the lane space. Furthermore, by applying Gaussian filtering and other smoothing processes to the left / right boundary lines, minor jitters caused by data sampling noise are eliminated, generating continuous and smooth lane boundary lines.

[0087] Regarding step S150, calculating the lane boundary line based on the target vehicle's trajectory data and the lane centerline, in some other embodiments of this disclosure, may include, for example: Figure 6 Steps S610 to S620 are shown.

[0088] In step S610, the lateral position of the road edge of the target road is identified.

[0089] Images of both sides of the road are acquired using vehicle-mounted front-view and side-view cameras to identify potential road edge features such as curbs, guardrails, vegetation boundaries, or road surface-shoulder boundaries. Millimeter-wave radar point cloud data is fused to detect static obstacle clusters with high reflectivity and a continuous linear distribution on both sides of the road. Visual recognition results and radar detection results are jointly calibrated and matched spatiotemporally to filter out edge points with high confidence. Lateral position clustering is performed on these high-confidence edge points, and continuous road edge reference lines are generated through curve fitting. The average lateral offset of the road edge reference lines in the vehicle's local coordinate system represents the identified lateral position of the road edge.

[0090] In step S620, the lane boundary line is determined based on the lateral position of the road edge and the lane width.

[0091] It effectively overcomes the limitations of relying solely on traffic flow (which fails when traffic is sparse) or solely on road edges (which fails when there are no clear road edges), providing redundant and complementary boundary judgment criteria in complex scenarios such as rural dirt roads and construction access roads. Through the dual constraints of physical boundaries and dynamic traffic space, it improves the accuracy and environmental adaptability of virtual lane boundary lines.

[0092] In one specific embodiment, when the number of target vehicles or the trajectory consistency between candidate vehicles changes, the trajectory data of the candidate vehicles is acquired to reconstruct the virtual lane lines.

[0093] Specifically, when the number of candidate vehicles participating in the construction of virtual lane lines increases or decreases (such as the preceding vehicle leaving or a new vehicle entering a stable following state), or when the trajectory consistency score between any two candidate vehicles drops beyond a preset threshold (such as a sudden drop of more than 30%), the latest trajectory data of the candidate vehicles will be obtained in real time, and the virtual lane lines will be reconstructed to ensure a rapid response to sudden events in traffic flow (such as a neighboring vehicle suddenly changing lanes or cutting in).

[0094] Preferably, when a new candidate vehicle is added, and the trajectory consistency score between the candidate vehicle and the existing candidate vehicles is greater than or equal to a first threshold, the virtual lane line is reconstructed.

[0095] In one specific embodiment, the virtual lane lines are updated according to a dynamic period of 50~100ms (100ms is used when the vehicle speed is less than or equal to 40km / h, and 50ms is used when the vehicle speed is greater than 40km / h), real-time trajectory points of candidate vehicles are added, historical trajectory data points of candidate vehicles 5 seconds ago are removed, and the virtual lane lines are reconstructed.

[0096] In one specific embodiment, when there is no valid target vehicle (e.g., no candidate vehicle or all candidate vehicles are abnormal vehicles), an audio-visual prompt is triggered to activate the low-speed cruise mode of the autonomous driving system at speeds less than or equal to 40 km / h.

[0097] The technical solution of this disclosure will be further explained below with specific implementation and application examples.

[0098] On a rural road without physical lane markings (lane marking recognition confidence level of 0.3), there are vehicles in front (f), to the left (l), and to the right (r) around the vehicle. Multiple sensors collect 5 seconds of historical trajectory data.

[0099] Specifically, the trajectory of the preceding vehicle: longitudinal distance range x∈[10~100m], lateral position range y∈[0.1~0.3m], heading angle range θ∈[0.4~0.6°], with no abrupt changes; Right-side vehicle trajectory: longitudinal distance range x∈[10~100m], lateral position range y∈[-1.8~-2.0m], heading angle range θ∈[0.5~0.7°], no jump points; The trajectory of the vehicle on the left: longitudinal distance range x∈[10~100m], lateral position range y∈[1.9~2.1m], heading angle range θ∈[5.2~6.0°], there are 3 jump points (accounting for 6%), and the vehicle on the left is retained after preprocessing; Based on this, it was determined that none of the three vehicles were detected as erroneous vehicles, and the process entered the consistency adjustment phase.

[0100] For any two vehicles with 5 seconds of historical trajectory data after time alignment, calculate the consistency score S (S∈[0,1]): in, , The average difference in heading angle between two vehicles reflecting the consistency of their driving directions. Exceeding At that time, this item was counted as 0.

[0101] Therefore, the trajectory consistency score between the vehicle in front and the vehicle to the right is... Scoring for the consistency of the trajectories of the vehicle in front and the vehicle to the left. ， Count as 0）The trajectory consistency score between the car on the right and the car on the left is S=0.42. Among them, the trajectory consistency score of the vehicle to the left with the other two vehicles was less than 0.6, and the average difference in heading angle between the vehicle and the other vehicles was greater than 5°. Therefore, the vehicle to the left was determined to be an abnormal vehicle and was excluded. The vehicle in front and the vehicle to the right were identified as the target vehicles.

[0102] Based on the trajectory data of the vehicle in front and the vehicle to the right, a cubic polynomial is used to obtain... The fitting error is 0.15m; This represents the center line of the virtual lane.

[0103] Within the range of x∈[20~80m], the median lateral distance between the vehicle on the right and the vehicle in front is 3.8m, therefore the lane width W is 3.8m; Distance between the car on the right and the center line The 75th percentile is 1.9m, therefore the right boundary line... Distance between the left boundary and the center line based on the default width It is 1.9m, therefore After Gaussian filtering and smoothing, the lateral deviation of the lane boundary line is 0.08m.

[0104] In addition, when the vehicle speed is 50km / h, an update cycle of 50ms is used, and a new trajectory point is added and refitted every 50ms. The centerline fitting error is stable at 0.15~0.2m.

[0105] When the lateral speed of the vehicle to the right suddenly increases to 1.3 m / s, and its lateral displacement changes by 0.9 m within 500 ms, it is determined to be a temporary lane-changing vehicle. The trajectory consistency score drops sharply to 0.35, triggering an emergency update and excluding the vehicle to the right. The vehicle in front is then re-selected as the target vehicle, with a default lane width W = 3.75 m and an update delay of 28 ms. After 2 seconds, the average difference between the heading angle of the vehicle to the right and that of the vehicle in front decreases to 2.8°, and its lateral position stabilizes. The trajectory consistency score of the vehicle to the right is recalculated to 0.72, and it is re-selected as the target vehicle, updating the virtual lane lines.

[0106] Based on any of the above embodiments, this disclosure also provides a virtual lane line construction device.

[0107] Figure 7 This is a schematic block diagram of a virtual lane line construction device according to one embodiment of the present disclosure.

[0108] like Figure 7 As shown, the virtual lane line construction device includes: The trajectory acquisition module 7002 acquires trajectory data of at least two candidate vehicles surrounding the vehicle within a target time period for the target road. The trajectory determination module 7004 determines the trajectory consistency between any two candidate vehicles based on the trajectory data of each candidate vehicle. The trajectory consistency includes the lateral position consistency and heading angle consistency between any two candidate vehicles. The vehicle screening module 7006 analyzes the trajectory consistency between every two candidate vehicles and selects target vehicles that meet the trajectory consistency screening threshold from the candidate vehicles. The centerline fitting module 7008 performs centerline fitting on the trajectory data of the target vehicle based on the road curvature of the target road to determine the lane centerline. The width and boundary calculation module 7010 determines the lane width and lane boundary line by analyzing the lateral envelope range of the target vehicle's driving path, based on the trajectory data of the target vehicle and the spatial distribution relationship of the lane centerline.

[0109] The virtual lane line construction module 7012 constructs virtual lane lines based on the lane center line, lane width, and lane boundary lines.

[0110] The aforementioned virtual lane line construction device can be in the form of computer software, and each module of the aforementioned virtual lane line construction device can be implemented through computer software modules.

[0111] The specific implementation process of the functions and roles of each module in the above-mentioned virtual lane line construction device can be found in the implementation process of the corresponding steps in the above method, and will not be repeated here.

[0112] This disclosure also provides an electronic device. Figure 8 A schematic diagram of the hardware implementation using the processing system is shown.

[0113] The hardware structure of the electronic device / apparatus 1000 can be implemented using a bus architecture. The bus architecture can include any number of interconnect buses and bridges, depending on the specific application and overall design constraints of the hardware. Bus 1100 connects various circuits including one or more processors 1200, memory 1300, and / or hardware modules. Bus 1100 can also connect various other circuits 1400 such as peripherals, voltage regulators, power management circuits, external antennas, etc. Bus 1100 can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Component (EISA) bus, etc. Buses can be categorized as address buses, data buses, control buses, etc. For ease of illustration, only one connection line is used in this figure, but this does not indicate that there is only one bus or one type of bus.

[0114] For ease of explanation, certain steps of the above method are described in relation to modules. It should be understood that the corresponding module performing one or more steps of the above method may be one or more hardware modules specifically configured to perform the corresponding step, or implemented by a processor configured to perform the corresponding step, or stored in a computer-readable medium for implementation by a processor, or implemented by some combination thereof.

[0115] This disclosure also provides a readable storage medium storing a computer program that, when executed by a processor, is used to implement the methods described above. A "readable storage medium" can be any means capable of containing, storing, communicating, propagating, or transmitting a program for use by or in conjunction with an instruction execution system, apparatus, or device. More specific examples of a readable storage medium include: an electrical connection with one or more wires (electronic device), a portable computer disk drive (magnetic device), random access memory (RAM), read-only memory (ROM), erasable and programmable read-only memory (EPROM or flash memory), fiber optic devices, and portable read-only memory (CDROM), etc.

[0116] This disclosure also provides a computer program product, the methods of which can be implemented wholly or partially through software, hardware, firmware, or any combination thereof. When implemented in software, it can be implemented wholly or partially as a computer program product. The computer program product includes one or more computer programs or instructions. When the computer program or instructions are loaded and executed, all or part of the processes or functions of this disclosure are performed.

[0117] Computer programs or instructions can be stored in a readable storage medium or transferred from one readable storage medium to another. For example, the computer program or instructions can be transferred from one website, computer, server, or data center to another website, computer, server, or data center via wired or wireless means. The readable storage medium can be any available medium capable of access, or a data storage device such as a server or data center that integrates one or more available media. The available medium can be a magnetic medium, such as a floppy disk, hard disk, or magnetic tape; an optical medium, such as a digital video optical disc; or a semiconductor medium, such as a solid-state drive. The computer-readable storage medium can be a volatile or non-volatile storage medium, or it can include both volatile and non-volatile types of storage media.

[0118] Those skilled in the art will understand that embodiments of this disclosure can be provided as methods, systems, or computer program products. Therefore, this disclosure can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this disclosure can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0119] This disclosure is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to this disclosure. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0120] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0121] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0122] In the description of this specification, the references to terms such as "one embodiment / mode," "some embodiments / modes," "example," "specific example," or "some examples," etc., refer to specific features, structures, or characteristics described in connection with that embodiment / mode or example, which are included in at least one embodiment / mode or example of this disclosure. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment / mode or example. Moreover, the specific features, structures, or characteristics described may be combined in any suitable manner in one or more embodiments / modes or examples. Furthermore, without contradiction, those skilled in the art can combine and integrate the different embodiments / modes or examples described in this specification, as well as the features of different embodiments / modes or examples.

[0123] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this disclosure, "a plurality of" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0124] Those skilled in the art should understand that the above embodiments are merely for illustrating the present disclosure and are not intended to limit the scope of the disclosure. Those skilled in the art can make other changes or modifications based on the above disclosure, and these changes or modifications still fall within the scope of the present disclosure.

Claims

1. A method for constructing virtual lane lines, characterized in that, include: For the target road, acquire the trajectory data of at least two candidate vehicles located around the vehicle within the target time period; Based on the trajectory data of each candidate vehicle, the trajectory consistency between any two candidate vehicles is determined. The trajectory consistency includes the lateral position consistency and heading angle consistency between any two candidate vehicles. The trajectory consistency between any two candidate vehicles is analyzed, and target vehicles that meet the trajectory consistency screening threshold are selected from the candidate vehicles. Based on the road curvature of the target road, centerline fitting is performed on the trajectory data of the target vehicle to determine the lane centerline; Based on the trajectory data of the target vehicle and the spatial distribution relationship of the lane centerline, the lane width and lane boundary line are determined by analyzing the lateral envelope of the target vehicle's driving path. as well as A virtual lane line is constructed based on the lane centerline, the lane width, and the lane boundary line.

2. The virtual lane line construction method as described in claim 1, characterized in that, Before determining the trajectory consistency between any two candidate vehicles based on the trajectory data of each candidate vehicle, the following steps are included: Based on the trajectory data of the candidate vehicles, the difference between the lateral positions of the candidate vehicles at adjacent time points is calculated to determine the lateral position difference of each candidate vehicle. Based on the lateral position difference of each candidate vehicle, each candidate vehicle is screened to determine the candidate vehicles used to determine trajectory consistency.

3. The virtual lane line construction method as described in claim 1, characterized in that, Before analyzing the trajectory consistency between any two candidate vehicles and selecting target vehicles that meet the trajectory consistency screening threshold from the candidate vehicles, the following steps are included: Based on the trajectory data of the candidate vehicles, the changes in heading angle, lateral velocity, and lateral displacement of the candidate vehicles within the sampling period are calculated. Based on the changes in heading angle, lateral velocity, and lateral displacement of the candidate vehicles, each candidate vehicle is screened to determine the candidate vehicles to be screened through the trajectory consistency.

4. The virtual lane line construction method as described in claim 1, characterized in that, The trajectory consistency between any two candidate vehicles is analyzed, and target vehicles that meet the trajectory consistency screening threshold are selected from the candidate vehicles, including: When the number of candidate vehicles is greater than or equal to the target number, the candidate vehicle whose trajectory consistency is greater than or equal to the first threshold is taken as the target vehicle; when the number of candidate vehicles is less than the target number, the candidate vehicle whose trajectory consistency is greater than or equal to the second threshold is taken as the target vehicle; the first threshold is greater than the second threshold.

5. The virtual lane line construction method as described in claim 1, characterized in that, Based on the road curvature of the target road, centerline fitting is performed on the trajectory data of the target vehicle to determine the lane centerline, including: When the curvature of the target road is less than or equal to the target curvature, the trajectory data of the target vehicle is globally fitted to determine the lane centerline; when the curvature of the target road is greater than the target curvature, the trajectory data of the target vehicle is segmented and fitted to determine the lane centerline.

6. The virtual lane line construction method as described in claim 1, characterized in that, Based on the trajectory data of the target vehicle and the spatial distribution relationship of the lane centerline, the lane width is determined by analyzing the lateral envelope of the target vehicle's travel path, including: Within the longitudinal range of the target road, the median of the trajectory data of the lane centerline and the target vehicles located on the left and right sides of the vehicle is statistically analyzed to determine the lane width.

7. The virtual lane line construction method as described in claim 1, characterized in that, Based on the trajectory data of the target vehicle and the spatial distribution relationship of the lane centerline, the lane boundary line is determined by analyzing the lateral envelope of the target vehicle's travel path, including: Based on the trajectory data of the target vehicle, calculate the lateral distances between the target vehicle located to the left of the vehicle and the target vehicle located to the right of the vehicle and the center line of the lane, and determine the left boundary line and the right boundary line. The left and right boundary lines are smoothed to determine the lane boundary lines.

8. The virtual lane line construction method as described in claim 1, characterized in that, Based on the trajectory data of the target vehicle and the spatial distribution relationship of the lane centerline, the lane boundary line is determined by analyzing the lateral envelope of the target vehicle's travel path, including: Identify the lateral position of the road edge of the target road; The lane boundary line is determined based on the lateral position of the road edge and the lane width; Optionally, it also includes: when the number of target vehicles or the trajectory consistency among candidate vehicles changes, acquiring the trajectory data of the candidate vehicles to reconstruct the virtual lane lines; Optionally, before acquiring trajectory data of at least two candidate vehicles surrounding the vehicle within a target time period for the target road, the following steps are included: Collect image data of the road; The physical lane lines in the image data are identified, and a confidence score for the physical lane lines is calculated. When the confidence score of the physical lane line is less than the confidence threshold, the road is determined to be the target road; Optionally, based on the trajectory data of each candidate vehicle, the trajectory consistency between any two candidate vehicles is determined, including: This indicates the trajectory consistency between candidate vehicle i and candidate vehicle j. and Indicates the weighting coefficient. The Pearson correlation coefficient represents the historical lateral position sequence between candidate vehicle i and candidate vehicle j. The heading angle of the candidate vehicle i is represented. The heading angle of the candidate vehicle j is represented. This represents the maximum heading angle difference.

9. An electronic device, characterized in that, include: The memory stores execution instructions; as well as A processor that executes execution instructions stored in the memory, causing the processor to perform the virtual lane line construction method according to any one of claims 1 to 8.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the virtual lane line construction method according to any one of claims 1 to 8.

Images