Split mouth lane line anomaly detection and compensation method, device, equipment and medium

Abstract

This invention discloses a method, apparatus, device, and medium for detecting and compensating for lane line anomalies at a merging intersection, relating to the field of vehicle-assisted driving technology. The method includes acquiring left and right lane line parameters in front of the vehicle; determining the lane width change and the change in the difference in the heading angle coefficients of the left and right lane lines based on the left and right lane line parameters; if both the lane width change and the change in the difference in the heading angle coefficients of the left and right lane lines exceed preset thresholds, the current scenario is determined to be an abnormal merging intersection scenario; in response to the determination of an abnormal merging intersection scenario, the right lane line parameters are compensated and reconstructed based on the left lane line parameters. This invention solves the technical problem in existing technologies where only the historical average lane width is used for mirror compensation, without considering the gradual change characteristics of lane width itself in merging intersection scenarios, resulting in insufficient compensation accuracy at the initial stage of a fork.

Description

Technical Field

[0001] This application relates to the field of vehicle assisted driving technology, specifically to methods, devices, equipment, and media for detecting and compensating for lane line anomalies at merging points. Background Technology

[0002] With the popularization of intelligent driving technology, lane keeping assist systems have become standard equipment in mass-produced vehicles. In scenarios such as highway merging or ramp exits, the visual perception of the right lane line is often lost or abruptly changed due to road topology bifurcation, and an effective compensation mechanism is urgently needed to maintain system stability.

[0003] In existing technologies, a mirror compensation scheme based on the left lane line is typically used to address the loss of the right lane line. Specifically, when an anomaly in the right lane line is detected, a virtual right lane line is generated using the geometric parameters of the left lane line combined with a preset lane width.

[0004] However, existing solutions only use historical average lane widths for mirror compensation, failing to consider the gradual changes in lane width at merging points. Since lane widths at merging points often maintain a standard width initially and then gradually increase, fixed or historically averaged width values ​​cannot match actual road geometry changes. This results in insufficient compensation accuracy at the initial merging stage, leading to significant deviations between virtual lane lines and the actual path, thus affecting the control accuracy and safety of the lane keeping system. Summary of the Invention

[0005] This application provides a method, apparatus, device, and computer-readable storage medium for detecting and compensating for lane line anomalies at a merging point. It can solve the technical problem in the prior art that only the historical average value of lane width is used for mirror compensation, without considering the gradual change characteristics of lane width itself in the merging point scenario, resulting in insufficient compensation accuracy at the initial stage of the fork.

[0006] To achieve the above objectives, the technical solution adopted by the present invention is as follows: In a first aspect, embodiments of this application provide a method for detecting and compensating for lane line anomalies at a merging point, including: Obtain the parameters of the left and right lane lines in front of the vehicle; Based on the left lane line parameters and the right lane line parameters, determine the change in lane width and the change in the difference in the heading angle coefficients of the left and right lane lines. If the change in lane width and the change in the difference in the heading angle coefficients of the left and right lanes both exceed the preset threshold, the current scenario is determined to be an abnormal scenario at the diversion point. In response to the determination of an abnormal diversion scenario, the parameters of the right lane line are compensated and reconstructed based on the parameters of the left lane line.

[0007] Based on the above technical solution, determining the change in lane width and the change in the difference in the heading angle coefficients of the left and right lanes includes: Calculate the lane width and the difference in heading angle coefficients between the left and right lanes at multiple longitudinal distance sampling points; The first longitudinal distance sampling point is selected as the near-end reference point, and the second longitudinal distance sampling point is selected as the far-end reference point. The lane width change is determined based on the lane width difference between the far reference point and the near reference point; The change in the difference of the heading angle coefficients of the left and right lane lines is determined based on the difference between the heading angle coefficients of the left and right lane lines at the far reference point and the near reference point.

[0008] Based on the above technical solution, the compensation and reconstruction of the right lane line parameters based on the left lane line parameters includes: Before compensation and reconstruction, the normality of the left lane markings is checked; When the scenario is first identified as an abnormal divergence, the current lane width is saved as the lane width before the divergence. During the compensation and reconstruction process, the lateral offset coefficient of the right lane line is reconstructed based on the width of the lane before the fork.

[0009] Based on the above technical solution, the step of reconstructing the lateral offset coefficient of the right lane line based on the width of the lane before the fork includes: The width of the lane before the fork is limited to a preset lane width range; Using the constrained pre-fork lane width and the lateral offset coefficient of the left lane line, the lateral offset coefficient of the right lane line is calculated, such that the lateral distance between the reconstructed right lane line and the left lane line is equal to the constrained pre-fork lane width.

[0010] Based on the above technical solution, the left lane line parameters and right lane line parameters are represented by a cubic polynomial model, including the lateral offset coefficient C0, the heading angle coefficient C1, the curvature coefficient C2 and the rate of change of curvature coefficient C3. The compensation and reconstruction of the right lane line parameters based on the left lane line parameters also includes: The C1, C2, C3 and the visible range of the left lane line are respectively assigned the corresponding parameters of the right lane line.

[0011] Based on the above technical solution, the compensation and reconstruction are performed only within a preset compensation time window, and the method further includes: Obtain the quality assessment results of the left lane line; If the quality assessment result of the left lane line is good, then the duration threshold of the compensation time window is extended from the first time threshold to the second time threshold. If the quality assessment result of the left lane line is a quality decline, then the duration threshold of the compensation time window is the first time threshold.

[0012] Based on the above technical solution, after the compensation time window expires, it is detected whether the abnormal scenario at the diversion port has been eliminated; If the abnormal scenario at the diversion port has been eliminated, then the abnormal state is reset; If the abnormal scenario at the diversion port is not eliminated, the compensation and reconstruction shall be stopped.

[0013] Secondly, embodiments of this application provide a lane line anomaly detection and compensation device for a merging intersection, comprising: The parameter acquisition module is used to acquire the parameters of the left lane line and the right lane line in front of the vehicle; The change determination module is used to determine the change in lane width and the change in the difference in the heading angle coefficients of the left and right lane lines based on the left lane line parameters and the right lane line parameters. The change in the difference in the heading angle coefficients of the left and right lane lines is used to characterize the change in the degree of divergence of the left and right lane lines along the driving direction. The scenario determination module is used to determine that the current scenario is an abnormal diversion scenario when the change in lane width and the change in the difference between the heading angle coefficients of the left and right lanes both exceed a preset threshold. The compensation and reconstruction module is used to compensate and reconstruct the right lane line parameters based on the left lane line parameters in response to a scenario where the diversion point is deemed to be abnormal.

[0014] Thirdly, embodiments of this application provide a lane line anomaly detection and compensation device for a merging intersection. The merging intersection lane line anomaly detection and compensation device includes a processor, a memory, and a lane line anomaly detection and compensation program stored in the memory and executable by the processor. When the lane line anomaly detection and compensation program is executed by the processor, it implements the steps of the lane line anomaly detection and compensation method for a merging intersection as described in the first aspect.

[0015] Fourthly, embodiments of this application provide a computer-readable storage medium storing a lane line anomaly detection and compensation program for a merging point, wherein when the lane line anomaly detection and compensation program for a merging point is executed by a processor, it implements the steps of the lane line anomaly detection and compensation method for a merging point as described in the first aspect.

[0016] Compared with the prior art, the advantages of the present invention are as follows: This invention relates to a method, apparatus, device, and medium for detecting and compensating for lane line anomalies at merging points. By acquiring the left and right lane line parameters in front of the vehicle, it determines the change in lane width and the change in the difference in the heading angle coefficients of the left and right lane lines. A dual-feature joint judgment mechanism is used to identify abnormal merging point scenarios. In response to the judgment result, the right lane line parameters are reconstructed and compensated based on the left lane line parameters. This achieves accurate scene recognition and lane line parameter reconstruction for specific geometric changes at merging points. This method effectively improves the accuracy of lane line anomaly detection and the timeliness of compensation reconstruction in merging point scenarios. It prevents virtual lane line deviations caused by misjudgment due to a single feature or historical average width compensation. It solves the problem of existing solutions using only the historical average lane width for mirror compensation, without considering the gradual change characteristics of lane width in merging point scenarios, resulting in insufficient compensation accuracy at the initial stage of bifurcation. This significantly improves the control accuracy and driving safety of lane keeping systems in merging point scenarios. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a flowchart illustrating the method in the embodiments of the diversion lane line anomaly detection and compensation method, apparatus, equipment and medium of the present invention; Figure 2 This is a schematic diagram of the hardware structure in the embodiments of the method, apparatus, equipment and medium for detecting and compensating for lane line anomalies at the diversion point of the present invention. Detailed Implementation

[0019] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present application.

[0020] First, some of the technical terms used in this application will be explained to help those skilled in the art understand this application.

[0021] Cubic polynomial model: refers to the mathematical expression for lane line geometry modeling. This is the core foundation for representing lane line parameters in this scheme, including information on lateral offset, heading angle, curvature, and rate of curvature change.

[0022] Lateral offset coefficient C0: refers to the lateral distance of the lane line model at the origin of the vehicle coordinate system. It is used to characterize the left and right position of the lane line and is a basic parameter for calculating the lane width.

[0023] Heading angle coefficient C1: refers to the first derivative coefficient of the lane line model, used to characterize the tangent direction or heading angle of the lane line, reflecting the change in the direction of the lane line.

[0024] Abnormal scenarios at traffic divergence points: These refer to driving scenarios where the right lane line suddenly diverges, disappears, or changes abruptly due to road topology forks at highway exits or interchanges. This is the main target operating condition addressed by this solution.

[0025] Compensation and reconstruction: This refers to the process of reconstructing virtual right lane line parameters by generating the lane width and higher-order coefficients before the bifurcation based on the left lane line parameters when an anomaly is detected in the right lane line, in order to maintain the continuity of the lane keeping function.

[0026] Dual-feature joint judgment: This refers to a mechanism that uses two independent features, namely the change in lane width and the change in the difference in the heading angle coefficients of the left and right lanes, to perform logical AND fusion to identify abnormal scenarios at the diversion point, thereby reducing the false detection rate.

[0027] Lane width before bifurcation: refers to the actual lane width value saved at the moment the scenario of first determining an abnormal bifurcation point is identified. It is used as the baseline width for compensation and reconstruction, replacing the traditional historical average width.

[0028] Compensation time window: refers to the threshold duration for lane line compensation reconstruction. In this solution, it is supported to be dynamically adjusted according to the quality of the left lane line (e.g., 2 to 3 seconds) to balance compensation accuracy and timeliness.

[0029] Virtual lane line: refers to the right lane line generated through compensation and reconstruction. Its identifier ID is assigned a preset value (such as -1) to inform the downstream control module that the lane line is an estimated value rather than a perceived measured value.

[0030] Existing autonomous driving lane keeping systems largely rely on high-precision maps or single visual features for lane departure detection and compensation. For example, high-precision map-based solutions are costly and heavily affected by signal obstruction, and are prone to failure in scenarios such as tunnels; vision-based solutions typically use historical average lane widths for mirror compensation, failing to consider the gradual changes in lane width at lane departure points, resulting in insufficient compensation accuracy at the initial stage of the departure. Therefore, there is an urgent need for a lane anomaly detection and compensation method that does not rely on high-precision maps and can adapt to the geometrical changes at lane departure points. Based on this, this application proposes a method, apparatus, and device for lane anomaly detection and compensation at lane departure points.

[0031] To make the objectives, technical solutions, and advantages of this application clearer, the embodiments of this application will be described in further detail below with reference to the accompanying drawings.

[0032] In a first aspect, embodiments of this application provide a method for detecting and compensating for lane line anomalies at a merging point.

[0033] In one embodiment, reference is made to Figure 1 , Figure 1 This is a flowchart illustrating the lane marking anomaly detection and compensation method at the merging point described in this application. Figure 1 As shown, a method for detecting and compensating for lane line anomalies at a merging intersection includes: S1. Obtain the parameters of the left lane line and the right lane line in front of the vehicle; S2. Based on the left lane line parameters and the right lane line parameters, determine the change in lane width and the change in the difference in the heading angle coefficient between the left and right lane lines; S3. If the change in lane width and the change in the difference between the heading angle coefficients of the left and right lanes both exceed the preset threshold, the current scenario is determined to be an abnormal scenario at the diversion point. S4. In response to the determination of an abnormal scenario at the diversion point, the parameters of the right lane line are compensated and reconstructed based on the parameters of the left lane line.

[0034] This application is mainly applied to scenarios with right-side divergence, such as highways and urban expressways. Therefore, when an anomaly is detected, the right-side lane line is directly reconstructed through compensation of the left-side lane line.

[0035] In this embodiment, by acquiring the parameters of the left and right lane lines and jointly analyzing the changing trends of width and heading angle, pure visual diversion detection without relying on high-precision maps is achieved, solving the detection failure problem caused by untimely updates or signal loss of high-precision maps. By compensating and reconstructing the right lane based on the parameters of the left lane line, continuous lane keeping is achieved when the right lane line is lost or changes abruptly, solving the problem of lane keeping system disengagement or vehicle lateral control oscillation caused by abnormal right lane lines, significantly improving the driving smoothness and safety of vehicles in high-speed diversion scenarios.

[0036] Further, in one embodiment, determining the change in lane width and the change in the difference in the heading angle coefficients of the left and right lane lines includes: Calculate the lane width and the difference in heading angle coefficients between the left and right lanes at multiple longitudinal distance sampling points; The first longitudinal distance sampling point is selected as the near-end reference point, and the second longitudinal distance sampling point is selected as the far-end reference point. The lane width change is determined based on the lane width difference between the far-end reference point and the near-end reference point. The change in the difference of the heading angle coefficients of the left and right lane lines is determined based on the difference between the far-end reference point and the near-end reference point.

[0037] In this embodiment, by adopting a differential calculation method with two reference points at the near and far ends, the problem of single-point detection being susceptible to local occlusion or noise interference is avoided. The spatial gradient change between the two points is used to characterize the evolution trend of lane geometry features. Compared with absolute value judgment, it can better reflect the unique "divergent" shape of the diversion point, effectively distinguishing between static road width changes and dynamic diversion point topology changes, and improving the robustness of feature extraction.

[0038] Further, in one embodiment, the multiple longitudinal distance sampling points include locations at 0 meters, 15 meters, 40 meters, 60 meters, 80 meters, and 100 meters distributed along the vehicle's direction of travel. A first longitudinal distance sampling point is selected as a near-end reference point, and a second longitudinal distance sampling point is selected as a far-end reference point, including: determining the 15-meter location as the near-end reference point and the 80-meter location as the far-end reference point. (Note: The 15-meter near-end data is reliable, while the 80-meter far-end data can sense trends and is still within the effective range of the camera; the 65-meter span between the two can smooth out noise.) In this embodiment, a 65-meter effective sensing baseline was constructed by selecting two specific distance points, 15 meters and 80 meters. Data at 15 meters has high confidence and serves as a reference anchor point, eliminating near-field measurement errors. At 80 meters, the baseline covers the starting area of ​​the diversion channel guide line, capturing the trend of lane widening. The combination of these two methods ensures data reliability while also considering forward-looking and trend-perceiving capabilities. It effectively smooths sensor noise, highlights the true geometric change trend, and avoids misjudgments caused by near-field jitter or far-field blurring.

[0039] Furthermore, in one embodiment, calculating the lane width includes calculating based on the lateral offset coefficient C0 of the left and right lane lines, using the following formula: ; in, The vertical distance is... and Distance Lateral offset coefficients of the left and right lane lines at the location.

[0040] In this embodiment, the lane width is calculated based on the C0 coefficient of a cubic polynomial model, which can accurately quantify the lateral spatial distance. This method directly utilizes lane line fitting parameters, eliminating the need for additional pixel-level edge detection. It is computationally efficient and conforms to the vehicle control model's definition of lane width, providing an accurate geometric basis for subsequent judgment of whether the lane has widened abnormally.

[0041] Furthermore, in one embodiment, calculating the difference in heading angle coefficients between the left and right lane lines includes calculating the difference in the first derivatives of the left and right lane lines, as shown in the formula: ; ; ; in, These are the heading angle coefficient, curvature coefficient, and rate of change of curvature coefficient for the lane line model, respectively.

[0042] In this embodiment, the divergence of the left and right lane lines is described from a kinematic constraint perspective by calculating the difference in the first derivative of the lane lines (i.e., the direction of the tangent to the heading angle). In the scenario of a split lane, the right lane line often deflects to the right, resulting in a significant increase in the difference in heading angle between the left and right lane lines. Introducing this feature compensates for the inability of a single width feature to distinguish between "lane widening" and "lane bifurcation," enhancing the depth of scene understanding.

[0043] Furthermore, in one embodiment, the compensation and reconstruction of the right lane line parameters based on the left lane line parameters includes: Before compensation and reconstruction, the normality of the left lane markings is checked; When the scenario is first identified as an abnormal divergence, the current lane width is saved as the lane width before the divergence. During the compensation and reconstruction process, the lateral offset coefficient of the right lane line is reconstructed based on the lane width before the fork.

[0044] In this embodiment, the normality test of the left lane line is used to prevent the use of abnormal left lane line parameters from contaminating the compensation result of the right lane line, thus ensuring the reliability of the compensation source data. By saving the lane width at the instant before the fork, the most realistic road geometry before the fork is preserved, avoiding the smoothing error that may be caused by using historical average values, so that the compensated lane line is more in line with the actual road shape in the early stage of the fork.

[0045] Furthermore, in one embodiment, reconstructing the lateral offset coefficient of the right lane line based on the lane width before the fork includes: The width of the lane before the fork is limited to a preset lane width range; Using the constrained pre-fork lane width and the lateral offset coefficient of the left lane line, calculate the lateral offset coefficient of the right lane line, so that the lateral distance between the reconstructed right lane line and the left lane line is equal to the constrained pre-fork lane width.

[0046] In this embodiment, a limiting mechanism ensures that the lane width value used for mirror compensation is always within a reasonable range of highway lane widths (e.g., 3.3 meters to 3.8 meters). This prevents extreme width values ​​(such as too narrow or too wide) caused by perceived noise from being used for reconstruction, avoiding the reconstructed virtual lane lines from deviating from the actual drivable area, thereby ensuring the safety boundary of vehicle lateral control.

[0047] Further, in one embodiment, calculating the lateral offset coefficient of the right lane line includes: ; in, The lateral offset coefficient of the reconstructed right lane line. The constrained lane width before the fork. This is the lateral offset coefficient for the left lane line.

[0048] At the same time, the heading angle coefficient C1, curvature coefficient C2, curvature change rate coefficient C3, and visible range of the left lane line are directly assigned to the corresponding parameters of the right lane line.

[0049] Then assign the ID of the right lane line identifier to a preset virtual lane line identifier (e.g., -1).

[0050] In this embodiment, a hybrid compensation strategy of "C0 retaining the original width and C1-C3 reusing the left side" is adopted. This maintains the correct lane width in the lateral position (reconstructed through C0) and the continuity of the road in terms of curvature (mirrored through C1-C3). This is because the curvature consistency of the left and right lane lines is still relatively high near the divergence point. Directly copying the higher-order parameters of the left side can ensure the smoothness of the virtual lane line and avoid angles or abrupt changes in the reconstructed lane line, so that the compensated right virtual lane line matches the actual road geometry perfectly.

[0051] Furthermore, in one embodiment, the left lane line parameters and the right lane line parameters are represented by a cubic polynomial model, including the lateral offset coefficient C0, the heading angle coefficient C1, the curvature coefficient C2, and the rate of change of curvature coefficient C3. The compensation and reconstruction of the right lane line parameters based on the left lane line parameters also includes: Assign the values ​​of C1, C2, C3 and the visible range of the left lane lines to the corresponding parameters of the right lane lines.

[0052] In this embodiment, the cubic polynomial model can accurately describe the geometry of lane lines, providing a mathematical basis for parameter compensation. This model is the standard input format for autonomous driving control algorithms; directly reconstructing the polynomial coefficients eliminates the need for subsequent coordinate transformations, reducing computational latency and facilitating direct use by downstream planning and control modules.

[0053] Furthermore, in one embodiment, the compensation reconstruction is performed only within a preset compensation time window, and the method further includes: Obtain the quality assessment results for the left lane line; If the quality assessment result of the left lane line is good, the duration threshold of the compensation time window will be extended from the first time threshold to the second time threshold. If the quality assessment result of the left lane line is a quality decline, then the duration threshold of the compensation time window is the first time threshold.

[0054] In this embodiment, an adaptive time window management based on the quality of the left lane line achieves a dynamic balance between compensation accuracy and time persistence. When the quality of the left lane line is good, it means that the reliability of the mirror source is high. Therefore, extending the compensation window (e.g., to 3 seconds) can provide more time for the recovery of the right lane line and reduce control disturbances caused by frequent switching. When the quality deteriorates, shortening the window can avoid the long-term use of unreliable compensation data and prevent error accumulation.

[0055] Furthermore, in one embodiment, the first time threshold is 2000 milliseconds, and the second time threshold is 3000 milliseconds.

[0056] Specifically, the default compensation time window is 2000 milliseconds; if the quality assessment result of the left lane line is detected as good (e.g., the quality indicator of the left lane line is true), the compensation time window will be extended to 3000 milliseconds.

[0057] The compensation and reconstruction process continues within the compensation time window. If the abnormal scenario at the diversion point is not eliminated by the time the window expires, the compensation and reconstruction process stops and the abnormal status counter is incremented. If the abnormality has been eliminated by the time the window expires, the abnormal status is reset.

[0058] In this embodiment, the 2-3 second time window covers the time required for a vehicle to pass through a typical diversion area. This adaptive mechanism is based on the reasonable assumption that "the better the quality of the left lane line, the higher the accuracy of its mirrored right lane line, and therefore the system can rely on mirror compensation for a longer period of time." This ensures the continuity of compensation while preventing the system from getting stuck in an infinite loop when anomalies persist through a timeout exit mechanism.

[0059] Furthermore, in one embodiment, after the compensation time window expires, it is detected whether the abnormal scenario at the diversion port has been eliminated; If the abnormal scenario at the diversion point has been eliminated, then reset the abnormal status. If the abnormal scenario at the diversion point is not eliminated, compensation and reconstruction will be stopped.

[0060] Specifically, after stopping compensation and reconstruction, the method also includes: Check whether the abnormal scenario at the diversion point has been eliminated; If the abnormal scenario at the diversion point has been eliminated, the abnormal duration counter will be reset to zero. If the abnormal scenario at the diversion point is not eliminated, the abnormal duration counter will be incremented by 1, and the abnormal handling process will be completely exited after the counter exceeds the preset upper limit.

[0061] In this embodiment, dynamic management of the duration of shunt port anomalies is achieved through anomaly state counter management. If the anomaly persists for an extended period (e.g., counter exceeds limit), it indicates that the anomaly may not be a transient shunt port scenario but rather another persistent fault. In this case, exiting the compensation process prevents the system from making incorrect decisions based on outdated data, avoids the cumulative errors caused by long-term use of outdated data, and ensures the healthy operation of the system state machine.

[0062] Furthermore, in one embodiment, the preset thresholds include a lane width change threshold and a heading angle coefficient difference change threshold.

[0063] The threshold for lane width variation is 0.9 meters (meaning the lane is more than 0.9 meters wider at 80 meters than at 15 meters). The threshold for the change in heading angle coefficient difference is 0.02 (that is, the difference in the left and right lane coefficients at 80 meters is more than 0.02 greater than that at 15 meters).

[0064] The current scenario is determined to be an abnormal scenario at the diversion point only if both of the above conditions are met simultaneously.

[0065] In this embodiment, by setting specific thresholds of 0.9 meters and 0.02 meters, and employing a logical AND dual-feature joint judgment mechanism, false triggers in non-diversion scenarios such as widening of ordinary lanes and widening of temporary lane lines in construction areas are effectively eliminated. For example, lanes before toll stations often widen, but the change in heading angle is not significant; this solution can avoid such false detections. Actual testing shows that the dual-feature joint judgment reduces the false detection rate by approximately 60% compared to a single width threshold judgment. 80%, significantly improving the detection accuracy in diversion scenarios.

[0066] Furthermore, in one embodiment, the preset lane width ranges from 3.3 meters to 3.8 meters.

[0067] The specific width restriction processing logic is as follows: if the width of the lane before the fork is less than 3.3 meters, then take 3.3 meters; if the width of the lane before the fork is greater than 3.8 meters, then take 3.8 meters; if it is within the range of 3.3 meters to 3.8 meters, then directly use the width of the lane before the fork.

[0068] (Note: This range is based on the standard lane width of highways to prevent outliers from causing compensation distortion.) In this embodiment, the range of 3.3 meters to 3.8 meters covers the width of most standard highway lanes. Through this hard constraint, even if there is a brief fluctuation in the width calculation before the fork in the preceding steps, the system can correct it to a reasonable physical range, further improving the robustness of the compensation parameters and preventing outliers from causing compensation distortion and thus causing vehicles to drift or deviate from their lanes.

[0069] Furthermore, in one embodiment, this application is applied to a specific scenario where an L2-level autonomous driving system equipped with a single forward-looking camera approaches a ramp exit / interchange fork in the road while driving on a highway.

[0070] System configuration: Hardware platform: NXP S32G vehicle domain controller; Perception sensor: Single forward-facing camera, resolution 1280×720, frame rate 30fps; Lane line model: cubic polynomial ; Sampling distance array: [0.0,15.0,40.0,60.0,80.0,100.0] meters.

[0071] The steps for detecting and compensating for lane marking anomalies at the diversion point are as follows: The first step is to calculate the lane width sequence: for each sampling distance x, the lane width is calculated using the lateral offset coefficient C0 of the left and right lane lines.

[0072] Longitudinal position of the left lane line: ; Longitudinal position of the right lane line: ; Lane width: .

[0073] The second step is to calculate the heading angle difference coefficient sequence: for each sampling distance x, calculate the difference in the first derivative (coefficient) of the left and right lane lines.

[0074] The left lane line coefficient is: ; The right lane line coefficient is: ; Heading angle difference coefficient: .

[0075] The third step is to calculate the proximal-distal variation: Take the lane width and heading angle difference coefficient at x=15 meters; Take the lane width and heading angle difference coefficient at x=80 meters; Lane width change = Lane width at 80 meters - Lane width at 15 meters; Change in heading angle difference coefficient = heading angle difference coefficient at 80 meters - heading angle difference coefficient at 15 meters.

[0076] The fourth step is to make a joint judgment based on two features: if the change in lane width is greater than 0.9 and the change in heading angle difference coefficient is greater than 0.02, then the divergence is determined to be true; record the time of the first anomaly and the lane width before the divergence (equal to the lateral offset coefficient of the left lane line minus the lateral offset coefficient of the right lane line at the previous moment).

[0077] Step 5, Lane line parameter compensation processing: Constrain the lane width before the fork to a range of 3.3 meters to 3.8 meters; check if the left lane line status is normal (no loss or jump on the left); if the left lane line status is normal, then perform mirror reconstruction of the right lane line: The lateral offset coefficient C0 of the right lane line is assigned the negative value of the absolute value of the difference between the width of the lane before the fork and the lateral offset coefficient C0 of the left lane line; The heading angle coefficient C1 of the right lane line is assigned the same value as the heading angle coefficient C1 of the left lane line; Assign the curvature coefficient C2 of the right lane line to the curvature coefficient C2 of the left lane line; The right lane curvature change rate coefficient C3 is assigned the same as the left lane curvature change rate coefficient C3. Assign the visible range of the right lane line to the visible range of the left lane line; Assign the ID value of the right lane line identifier to -1 (mark it as a virtual lane line).

[0078] Step 6, Time Window Management and Timeout Exit: The default time threshold is 2000 milliseconds; if the quality of the left lane line is good, the time threshold is increased by 1000 milliseconds (i.e., extended to 3 seconds); if the elapsed time is less than or equal to the time threshold, mirror compensation is performed (step 5); otherwise (i.e. after timeout): if the abnormality is not eliminated, the abnormality duration counter is incremented by 1; if the abnormality has been eliminated, the abnormality status is reset.

[0079] Key parameter configuration: Lane width difference threshold = 0.9; Coefficient difference threshold = 0.02; Lower limit of lane width = 3.3 meters; Upper limit of lane width = 3.8 meters; Default post-processing time threshold = 2000 milliseconds.

[0080] Secondly, embodiments of this application provide a lane line anomaly detection and compensation device for a merging intersection, comprising: The parameter acquisition module is used to acquire the parameters of the left lane line and the right lane line in front of the vehicle; The change determination module is used to determine the change in lane width and the change in the difference between the heading angle coefficients of the left and right lane lines based on the parameters of the left and right lane lines. The change in the difference between the heading angle coefficients of the left and right lane lines is used to characterize the change in the degree of divergence of the left and right lane lines along the driving direction. The scenario determination module is used to determine that the current scenario is an abnormal scenario at the diversion point when the changes in lane width and the difference in the heading angle coefficients of the left and right lanes both exceed preset thresholds. The compensation and reconstruction module is used to compensate and reconstruct the parameters of the right lane based on the parameters of the left lane when the scenario is determined to be an abnormal situation at the diversion point.

[0081] In this embodiment, the systematic implementation of lane line anomaly detection and compensation at diversion points is achieved through the collaborative work of the parameter acquisition module, change determination module, scene determination module, and compensation reconstruction module. This device solves the problem of lacking hardware support for method implementation, and its modular design clarifies the responsibilities of each functional unit, ensuring the streamlined execution of data acquisition, feature calculation, scene determination, and compensation reconstruction. If the system is deployed on an onboard domain controller, the modules can also operate in a distributed manner, as long as the communication latency meets real-time requirements, significantly improving the system's engineering adaptability.

[0082] Furthermore, in one embodiment, the change determination module is specifically used to: calculate the lane width and the difference in heading angle coefficients of the left and right lane lines at multiple longitudinal distance sampling points; select a first longitudinal distance sampling point as a near-end reference point and select a second longitudinal distance sampling point as a far-end reference point; determine the lane width change based on the lane width difference between the far-end reference point and the near-end reference point; and determine the change in the heading angle coefficient difference between the left and right lane lines based on the difference in the heading angle coefficients of the left and right lane lines between the far-end reference point and the near-end reference point.

[0083] In this embodiment, the change determination module is configured to execute a multi-distance sampling strategy, solving the problem that single-distance sampling is susceptible to noise interference. The module supports spatial gradient analysis at both near and far ends, ensuring the accuracy of feature calculation. This configuration provides a reliable data foundation for subsequent joint determination of dual features. If the vehicle speed changes, the module can also dynamically adjust the sampling distance to adapt to different look-ahead requirements, enhancing the environmental adaptability of the device.

[0084] Furthermore, in one embodiment, the multiple longitudinal distance sampling points include locations at 0 meters, 15 meters, 40 meters, 60 meters, 80 meters, and 100 meters distributed along the vehicle's driving direction; the change determination module is specifically used to determine the 15-meter location as the near-end reference point and the 80-meter location as the far-end reference point.

[0085] In this embodiment, the change determination module is configured to select two specific distance points, 15 meters and 80 meters, to construct an effective sensing baseline of 65 meters. Data at 15 meters has high confidence and serves as a reference anchor point, eliminating near-field measurement errors. At 80 meters, the module covers the starting area of ​​the diversion guide line, capturing the trend of lane widening. The combination of these two methods ensures data reliability while also considering forward-looking and trend-perceiving capabilities. It effectively smooths sensor noise, highlights the true geometric change trend, avoids misjudgments caused by near-field jitter or far-field blurring, and improves the module's output stability.

[0086] Furthermore, in one embodiment, the change determination module is specifically used to calculate the lane width based on the lateral offset coefficient C0 of the left and right lane lines, using the following formula: ; in, The vertical distance is... and Distance Lateral offset coefficients of the left and right lane lines at the location.

[0087] In this embodiment, the change determination module calculates the lane width based on the C0 coefficient of a cubic polynomial model, which can accurately quantify the lateral spatial distance. This module directly utilizes the lane line fitting parameters, eliminating the need for additional pixel-level edge detection. It boasts high computational efficiency and conforms to the vehicle control model's definition of lane width, providing accurate geometric basis for subsequent judgments on whether the lane has abnormally widened, and reducing the processor's computational load.

[0088] Furthermore, in one embodiment, the change determination module is specifically used to calculate the difference in the first derivatives of the left and right lane lines, using the following formula: ; ; ; in, These are the heading angle coefficient, curvature coefficient, and rate of change of curvature coefficient for the lane line model, respectively.

[0089] In this embodiment, the change determination module describes the divergence of the left and right lane lines from a kinematic constraint perspective by calculating the difference in the first derivative of the lane lines (i.e., the direction of the tangent to the heading angle). In the scenario of a split lane, the right lane line often deflects to the right, resulting in a significant increase in the difference in heading angle between the left and right lane lines. Introducing this feature compensates for the inability of a single width feature to distinguish between "lane widening" and "lane bifurcation," enhancing the device's depth of scene understanding and discrimination capability.

[0090] Furthermore, in one embodiment, the scene determination module is specifically used to determine the current scene as an abnormal diversion scene when the lane width change exceeds 0.9 meters and the heading angle coefficient difference change exceeds 0.02.

[0091] In this embodiment, the scene determination module is configured to perform joint determination using two features, solving the problem of high false detection rate in single-feature determination. The module has built-in threshold logic, which can effectively distinguish between genuine diversion scenarios and ordinary lane widening scenarios. This configuration significantly reduces the false detection rate in diversion scenarios. If road standards change, the module can also support online calibration and updating of threshold parameters to increase the model's adaptability and maintainability.

[0092] Furthermore, in one embodiment, the compensation and reconstruction module is specifically used to: perform a normality check on the left lane line before compensation and reconstruction; when the scenario of an abnormal divergence is determined for the first time, save the lane width at the current moment as the lane width before the divergence; and during the compensation and reconstruction process, reconstruct the lateral offset coefficient of the right lane line based on the lane width before the divergence.

[0093] In this embodiment, the compensation and reconstruction module is configured to perform a normality check on the left lane line to prevent the use of abnormal left lane line parameters from contaminating the compensation result of the right lane line, thus ensuring the reliability of the compensation source data. By saving the lane width at the instant before the fork, the most realistic road geometry before the fork is preserved, avoiding the smoothing error that may be caused by using historical average values. This makes the compensated lane line more closely match the actual road shape in the early stage of the fork, improving the input quality of the control system.

[0094] Furthermore, in one embodiment, the compensation and reconstruction module is specifically used to: limit the width of the lane before the fork, constraining it within a preset lane width range of 3.3 meters to 3.8 meters; and calculate the lateral offset coefficient of the right lane using the constrained lane width before the fork and the lateral offset coefficient of the left lane line, so that the lateral distance between the reconstructed right lane line and the left lane line is equal to the constrained lane width before the fork.

[0095] In this embodiment, the compensation and reconstruction module ensures that the lane width value used for mirror compensation is always within a reasonable range of highway lane widths through a limiting mechanism. This prevents extreme width values ​​(such as those that are too narrow or too wide) caused by perceived noise from being used for reconstruction, avoiding the reconstructed virtual lane lines from deviating from the actual drivable area, thereby ensuring the safety boundary of vehicle lateral control and enhancing the robustness of the system.

[0096] Furthermore, in one embodiment, the compensation and reconstruction module is specifically used to calculate the lateral offset coefficient of the right lane line, using the following formula: ; in, The lateral offset coefficient of the reconstructed right lane line. The constrained lane width before the fork. This is the lateral offset coefficient for the left lane line.

[0097] At the same time, the heading angle coefficient C1, curvature coefficient C2, curvature change rate coefficient C3, and visible range of the left lane line are directly assigned to the corresponding parameters of the right lane line.

[0098] Then assign the ID of the right lane line identifier to a preset virtual lane line identifier (e.g., -1).

[0099] In this embodiment, the compensation and reconstruction module adopts a hybrid compensation strategy of "C0 retaining the original width and C1-C3 reusing the left side," which maintains the correct lane width in the lateral position and the continuity of the road in the curvature. This is because the curvature consistency of the left and right lane lines is still relatively high near the divergence point. Directly copying the higher-order parameters of the left side can ensure the smoothness of the virtual lane line, avoiding angles or abrupt changes in the reconstructed lane line. This makes the compensated right virtual lane line match the actual road geometry, optimizing the generation quality of control commands.

[0100] Furthermore, in one embodiment, the compensation and reconstruction module is specifically used to: obtain the quality assessment result of the left lane line; if the quality assessment result of the left lane line is good, then extend the duration threshold of the compensation time window from 2000 milliseconds to 3000 milliseconds; if the quality assessment result of the left lane line is deteriorated, then keep the duration threshold of the compensation time window at 2000 milliseconds.

[0101] In this embodiment, the compensation and reconstruction module is configured to perform adaptive time window management, which solves the problem that a fixed time window cannot adapt to different lane line qualities; the module achieves a dynamic balance between compensation accuracy and time persistence. When the quality of the left lane line is good, it means that the reliability of the mirror source is high. Therefore, extending the compensation window can provide more time for the restoration of the right lane line, reduce control disturbances caused by frequent switching, and improve the driving experience.

[0102] Furthermore, in one embodiment, the compensation and reconstruction module is specifically used to: detect whether the abnormal scenario at the diversion port has been eliminated after the compensation time window expires; if the abnormal scenario at the diversion port has been eliminated, then reset the abnormal state; if the abnormal scenario at the diversion port has not been eliminated, then stop the compensation and reconstruction and increment the abnormal counter.

[0103] In this embodiment, the compensation and reconstruction module manages the duration of abnormalities at the shunt outlet by managing an abnormal state counter. If the abnormality persists for an extended period, it indicates that the problem may not be a transient shunt outlet scenario but rather another persistent fault. Exiting the compensation process in this case prevents the system from making incorrect decisions based on outdated data, avoids the accumulation of errors caused by the long-term use of outdated data, and ensures the healthy operation of the system state machine and the effective triggering of fault-tolerant mechanisms.

[0104] Furthermore, in one embodiment, the specific functional implementation of each module in the diversion lane line anomaly detection and compensation device corresponds to each step in the above-described diversion lane line anomaly detection and compensation method embodiment. For example, the parameter acquisition module executes the parameter acquisition step in the method embodiment, the change amount determination module executes the change amount determination step and related calculation formula in the method embodiment, the scene determination module executes the scene determination step and related threshold logic in the method embodiment, and the compensation reconstruction module executes the compensation reconstruction step and related time window management process in the method embodiment. The specific calculation formulas, parameter definitions, and beneficial effects are described in the first aspect embodiment and will not be repeated here.

[0105] In addition, the functions of each module in the above-mentioned lane line anomaly detection and compensation device correspond to the steps in the above-mentioned lane line anomaly detection and compensation method embodiment, and their functions and implementation processes will not be described in detail here.

[0106] Thirdly, embodiments of this application provide a lane line anomaly detection and compensation device for a merging point. The lane line anomaly detection and compensation device for a merging point can be a personal computer (PC), laptop computer, server, or other device with data processing capabilities.

[0107] Reference Figure 2 , Figure 2 This is a schematic diagram of the hardware structure of the lane line anomaly detection and compensation device at the merging point involved in the embodiments of this application. In the embodiments of this application, the lane line anomaly detection and compensation device at the merging point may include a processor, a memory, a communication interface, and a communication bus.

[0108] The communication bus can be of any type and is used to interconnect the processor, memory, and communication interface.

[0109] The communication interface includes input / output (I / O) interfaces, physical interfaces, and logical interfaces used for interconnecting internal components of the lane line anomaly detection and compensation device at the shunting exit, as well as interfaces used for interconnecting the device with other devices (such as other computing devices or user equipment). Physical interfaces can be Ethernet interfaces, fiber optic interfaces, ATM interfaces, etc.; user equipment can be displays, keyboards, etc.

[0110] Memory can be various types of storage media, such as random access memory (RAM), read-only memory (ROM), non-volatile RAM (NVRAM), flash memory, optical storage, hard disk, programmable ROM (PROM), erasable PROM (EPROM), electrically erasable PROM (EEPROM), etc.

[0111] The processor can be a general-purpose processor, which can call the lane line anomaly detection and compensation program stored in the memory and execute the lane line anomaly detection and compensation method provided in the embodiments of this application. For example, the general-purpose processor can be a central processing unit (CPU). The method executed when the lane line anomaly detection and compensation program is called can be referred to the various embodiments of the lane line anomaly detection and compensation method of this application, and will not be repeated here.

[0112] Those skilled in the art will understand that Figure 2 The hardware structure shown does not constitute a limitation of this application and may include more or fewer components than shown, or combine certain components, or have different component arrangements.

[0113] Fourthly, embodiments of this application also provide a computer-readable storage medium.

[0114] The present application provides a computer-readable storage medium storing a lane line anomaly detection and compensation program for a merging intersection, wherein when the merging intersection lane line anomaly detection and compensation program is executed by a processor, it implements the steps of the merging intersection lane line anomaly detection and compensation method described above.

[0115] The method implemented when the lane line anomaly detection and compensation procedure at the merging point is executed can be referred to in various embodiments of the lane line anomaly detection and compensation method at the merging point of this application, and will not be repeated here.

[0116] It should be noted that the sequence numbers of the embodiments in this application are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0117] The terms "comprising" and "having," and any variations thereof, in the specification, claims, and accompanying drawings of this application are intended to cover non-exclusive inclusion. For example, a process, method, apparatus, product, or device that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to such processes, methods, products, or devices. The terms "first," "second," and "third," etc., are used to distinguish different objects, etc., and do not indicate a sequence, nor do they limit "first," "second," and "third" to different types.

[0118] In the description of the embodiments of this application, terms such as "exemplary," "for example," or "for instance" are used to indicate examples, illustrations, or explanations. Any embodiment or design described as "exemplary," "for example," or "for instance" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of terms such as "exemplary," "for example," or "for instance" is intended to present the relevant concepts in a concrete manner.

[0119] In the description of the embodiments of this application, unless otherwise stated, " / " means "or". For example, A / B can mean A or B. The "and / or" in the text is merely a description of the relationship between related objects, indicating that there can be three relationships. For example, A and / or B can mean: A exists alone, A and B exist simultaneously, and B exists alone. In addition, in the description of the embodiments of this application, "multiple" means two or more.

[0120] In some processes described in the embodiments of this application, multiple operations or steps are included in a specific order. However, it should be understood that these operations or steps may not be executed in the order they appear in the embodiments of this application, or they may be executed in parallel. The sequence number of the operation is only used to distinguish different operations, and the sequence number itself does not represent any execution order. In addition, these processes may include more or fewer operations, and these operations or steps may be executed sequentially or in parallel, and these operations or steps may be combined.

[0121] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) as described above, and includes several instructions to cause a terminal device to execute the methods described in the various embodiments of this application.

[0122] The above are merely preferred embodiments of this application and do not limit the patent scope of this application. Any equivalent structural or procedural transformations made using the content of this application's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of this application.

Claims

1. A method for detecting and compensating for lane line anomalies at a merging point, characterized in that, include: Obtain the parameters of the left and right lane lines in front of the vehicle; Based on the left lane line parameters and the right lane line parameters, determine the change in lane width and the change in the difference in the heading angle coefficients of the left and right lane lines. If the change in lane width and the change in the difference in the heading angle coefficients of the left and right lanes both exceed the preset threshold, the current scenario is determined to be an abnormal scenario at the diversion point. In response to the determination of an abnormal diversion scenario, the parameters of the right lane line are compensated and reconstructed based on the parameters of the left lane line.

2. The method for detecting and compensating for lane line anomalies at a merging intersection as described in claim 1, characterized in that, The determination of the change in lane width and the change in the difference in the heading angle coefficients of the left and right lanes includes: Calculate the lane width and the difference in heading angle coefficients between the left and right lanes at multiple longitudinal distance sampling points; The first longitudinal distance sampling point is selected as the near-end reference point, and the second longitudinal distance sampling point is selected as the far-end reference point. The lane width change is determined based on the lane width difference between the far reference point and the near reference point; The change in the difference of the heading angle coefficients of the left and right lane lines is determined based on the difference between the heading angle coefficients of the left and right lane lines at the far reference point and the near reference point.

3. The method for detecting and compensating for lane line anomalies at a merging intersection as described in claim 1, characterized in that, The compensation and reconstruction of the right lane line parameters based on the left lane line parameters includes: Before compensation and reconstruction, the normality of the left lane markings is checked; When the scenario is first identified as an abnormal divergence, the current lane width is saved as the lane width before the divergence. During the compensation and reconstruction process, the lateral offset coefficient of the right lane line is reconstructed based on the width of the lane before the fork.

4. The method for detecting and compensating for lane line anomalies at a merging intersection as described in claim 3, characterized in that, The method of reconstructing the lateral offset coefficient of the right lane line based on the width of the lane before the fork includes: The width of the lane before the fork is limited to a preset lane width range; Using the constrained pre-fork lane width and the lateral offset coefficient of the left lane line, the lateral offset coefficient of the right lane line is calculated, such that the lateral distance between the reconstructed right lane line and the left lane line is equal to the constrained pre-fork lane width.

5. The method for detecting and compensating for lane line anomalies at a merging intersection as described in any one of claims 1-4, characterized in that, The parameters of the left lane line and the right lane line are represented by a cubic polynomial model, including the lateral offset coefficient C0, the heading angle coefficient C1, the curvature coefficient C2, and the rate of change of curvature coefficient C3. The compensation and reconstruction of the right lane line parameters based on the left lane line parameters also includes: The C1, C2, C3 and the visible range of the left lane line are respectively assigned the corresponding parameters of the right lane line.

6. The method for detecting and compensating for lane line anomalies at a merging intersection as described in claim 1, characterized in that, The compensation and reconstruction are performed only within a preset compensation time window, and the method further includes: Obtain the quality assessment results of the left lane line; If the quality assessment result of the left lane line is good, then the duration threshold of the compensation time window is extended from the first time threshold to the second time threshold. If the quality assessment result of the left lane line is a quality decline, then the duration threshold of the compensation time window is the first time threshold.

7. The method for detecting and compensating for lane line anomalies at a merging intersection as described in claim 6, characterized in that, After the compensation time window expires, check whether the abnormal scenario at the diversion port has been eliminated; If the abnormal scenario at the diversion port has been eliminated, then the abnormal state is reset; If the abnormal scenario at the diversion port is not eliminated, the compensation and reconstruction shall be stopped.

8. A lane marking anomaly detection and compensation device at a merging point, characterized in that, include: The parameter acquisition module is used to acquire the parameters of the left lane line and the right lane line in front of the vehicle; The change determination module is used to determine the change in lane width and the change in the difference in the heading angle coefficients of the left and right lane lines based on the left lane line parameters and the right lane line parameters. The change in the difference in the heading angle coefficients of the left and right lane lines is used to characterize the change in the degree of divergence of the left and right lane lines along the driving direction. The scenario determination module is used to determine that the current scenario is an abnormal diversion scenario when the change in lane width and the change in the difference between the heading angle coefficients of the left and right lanes both exceed a preset threshold. The compensation and reconstruction module is used to compensate and reconstruct the right lane line parameters based on the left lane line parameters in response to a scenario where the diversion point is deemed to be abnormal.

9. A lane marking anomaly detection and compensation device at a merging point, characterized in that, The lane line anomaly detection and compensation device for the merging point includes a processor, a memory, and a lane line anomaly detection and compensation program for the merging point stored in the memory and executable by the processor, wherein when the lane line anomaly detection and compensation program for the merging point is executed by the processor, it implements the steps of the lane line anomaly detection and compensation method for the merging point as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a lane line anomaly detection and compensation program for a merging intersection, wherein when the merging intersection lane line anomaly detection and compensation program is executed by a processor, it implements the steps of the lane line anomaly detection and compensation method for a merging intersection as described in any one of claims 1 to 7.

Images