Optimization method and device for HWA to enter transverse control mode, equipment and medium

Abstract

The invention discloses an optimization method and device for an HWA to enter a transverse control mode, equipment and a medium. The method comprises the steps that if a vehicle activates a transverse control mode for the first time, whether the vehicle meets a standard transverse control mode entering condition or not at present is judged; if yes, the vehicle is controlled to enter a transverse control mode, and lane line and bad lane line information is detected according to a preset lane line recognition algorithm to judge whether the vehicle exits the transverse control mode or not; if the vehicle exits the transverse control mode due to the bad lane line information and the transverse control mode is reactivated in the current ignition period, whether the vehicle currently meets a standard transverse control mode entering condition and an additional transverse control mode entering condition at the same time or not is judged; and if so, reactivating the transverse control mode, and continuously judging whether the vehicle exits the transverse control mode or not according to the lane line recognition algorithm. By increasing the index requirement for entering the transverse control mode, frequent entering and exiting of the transverse control mode are avoided, and the stability, comfort and safety of vehicle driving are improved.

Description

Technical Field

[0001] Embodiments of the present invention relate to the field of autonomous driving technology, and in particular to an optimization method, device, equipment and medium for HWA to enter a lateral control mode. Background Art

[0002] With the rapid development of intelligent driving technology, Highway Assist System (HWA) has become an important function to improve driving safety and comfort. The lateral control function of the HWA system relies on accurate recognition of lane lines and automatically adjusts the steering wheel to keep the vehicle in the center of the lane. However, the lateral control function of the current HWA system has a high dependence on clear and continuous lane lines, and faces many challenges in practical applications.

[0003] The existing HWA lateral control function has a relatively simple logic when entering and exiting, and lacks adaptability to complex road conditions. When the lane line is unclear or interfered by miscellaneous lines, the system is prone to frequently exit and re-enter the lateral control mode, resulting in an unstable driving experience and even potential safety hazards. For example, on public roads, due to lane line wear, construction markings or other miscellaneous interference, the system may not be able to accurately identify the lane line, resulting in frequent switching of control states, increasing the driver's operating burden and potential risks. In addition, the current intelligent driving perception system mainly recognizes lane lines based on camera image acquisition and feature extraction through algorithms. However, the existing algorithm has poor recognition performance for miscellaneous lines and is prone to misidentification. In scenarios with wide lanes or chaotic lane lines, the system may mistakenly identify miscellaneous lines as lane lines, causing the vehicle to follow the miscellaneous lines and the "dragon drawing" phenomenon of frequent steering wheel adjustments. This not only affects driving comfort, but may also pose a threat to driving safety.

[0004] Therefore, the performance and reliability of the lateral control function of the existing HWA system under complex road conditions are obviously insufficient, especially when the lane lines are unclear or there are interference from miscellaneous lines. The system's frequent advance and retreat and misidentification problems need to be solved urgently. In order to improve the stability and safety of the HWA system, it is urgent to optimize the entry and exit logic of the lateral control and enhance the system's ability to recognize miscellaneous lines to reduce the impact of poor lane line information on the lateral control function. Summary of the invention

[0005] The embodiments of the present invention provide a method, device, equipment and medium for optimizing HWA entering lateral control mode, aiming to solve the problem of frequent exit and entry into lateral control due to poor lane line information.

[0006] In a first aspect, an embodiment of the present invention provides an optimization method for HWA to enter a lateral control mode, the method comprising:

[0007] If the vehicle activates the lateral control mode for the first time, determining whether the vehicle currently meets the standard lateral control mode entry conditions;

[0008] If satisfied, control the vehicle to enter the lateral control mode, and detect lane lines and bad lane line information according to a preset lane line recognition algorithm to determine whether the vehicle exits the lateral control mode;

[0009] If the vehicle exits the lateral control mode due to the bad lane line information and reactivates the lateral control mode in the current ignition cycle, determining whether the vehicle currently satisfies both the standard lateral control mode entry condition and the additional lateral control mode entry condition;

[0010] If both conditions are met, the lateral control mode is reactivated, and the lane line recognition algorithm is used to continue to determine whether the vehicle exits the lateral control mode.

[0011] In a second aspect, an embodiment of the present invention further provides an optimization device for HWA to enter a lateral control mode, comprising a unit for executing the above method.

[0012] In a third aspect, an embodiment of the present invention further provides a computer device, which includes a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the above method when executing the computer program.

[0013] In a fourth aspect, an embodiment of the present invention further provides a computer-readable storage medium, wherein the storage medium stores a computer program, and the computer program can implement the above method when executed by a processor.

[0014] The embodiment of the present invention provides an optimization method, device, equipment and medium for HWA to enter the lateral control mode. The method includes: if the vehicle activates the lateral control mode for the first time, judging whether the vehicle currently meets the standard lateral control mode entry condition; if so, controlling the vehicle to enter the lateral control mode, and detecting the lane line and bad lane line information according to a preset lane line recognition algorithm to judge whether the vehicle exits the lateral control mode; if the vehicle exits the lateral control mode due to the bad lane line information and reactivates the lateral control mode in the current ignition cycle, judging whether the vehicle currently meets both the standard lateral control mode entry condition and the additional lateral control mode entry condition; if both are met, reactivating the lateral control mode, and continuing to judge whether the vehicle exits the lateral control mode according to the lane line recognition algorithm. The present application adds the additional lateral control mode entry condition when reactivating the lateral control mode, and requires both the standard lateral control mode entry condition and the additional lateral control mode entry condition to be met before the lateral control mode can be reactivated, thereby increasing the index requirements for entering the lateral control mode, thereby avoiding frequent entry and exit of the lateral control mode, and improving the stability, comfort and safety of vehicle driving. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying any creative work.

[0016] Figure 1 A schematic diagram of an application scenario of an optimization method for HWA entering a lateral control mode according to an embodiment of the present invention;

[0017] Figure 2 A schematic flow chart of the steps of the optimization method for HWA entering the lateral control mode provided by an embodiment of the present invention;

[0018] Figure 3 A schematic flow chart of the first sub-step of the optimization method for HWA entering the lateral control mode provided by an embodiment of the present invention;

[0019] Figure 4 A schematic flow chart of the second sub-step of the optimization method for HWA entering the lateral control mode provided by an embodiment of the present invention;

[0020] Figure 5 A schematic flow chart of the third sub-step of the optimization method for HWA entering the lateral control mode provided by an embodiment of the present invention;

[0021] Figure 6A schematic flow chart of the fourth sub-step of the optimization method for HWA entering the lateral control mode provided by an embodiment of the present invention;

[0022] Figure 7 A schematic diagram of a fifth sub-step flow chart of the optimization method for HWA entering the lateral control mode provided by an embodiment of the present invention;

[0023] Figure 8 A schematic flow chart of the sixth sub-step of the optimization method for HWA entering the lateral control mode provided by an embodiment of the present invention;

[0024] Fig. 9 A schematic block diagram of an optimization device for HWA entering a lateral control mode provided by an embodiment of the present invention;

[0025] Fig.10 A schematic block diagram of a computer device provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0026] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0027] It should be understood that when used in this specification and the appended claims, the terms "include" and "comprises" indicate the presence of described features, integers, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or combinations thereof.

[0028] It should also be understood that the terms used in this specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. As used in the specification of the present invention and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to include plural forms.

[0029] It should be further understood that the term "and / or" used in the present description and the appended claims refers to any and all possible combinations of one or more of the associated listed items, and includes these combinations.

[0030] As used in this specification and the appended claims, the term "if" may be interpreted as "when" or "upon" or "in response to determining" or "in response to detecting," depending on the context. Similarly, the phrases "if it is determined" or "if [described condition or event] is detected" may be interpreted as meaning "upon determination" or "in response to determining" or "upon detection of [described condition or event]" or "in response to detecting [described condition or event]," depending on the context.

[0031] In order to facilitate understanding of the solutions of the embodiments of the present invention, the technical meanings of the important terms below are first explained.

[0032] Highway Assist System (HWA) is an automatic driving assistance system suitable for highways. It combines adaptive cruise control (ACC) and lane centering functions to achieve automatic following, speed control and lane keeping on highways. The HWA system can also assist in lane change operations when the driver issues a lane change command, effectively reducing the driver's driving burden and improving driving safety and comfort.

[0033] The lateral control function in HWA is a key part to ensure the vehicle's stable driving on the highway. It monitors the vehicle's lateral position and lane lines through the vehicle's sensors, and uses the electronic power steering system (EPS) to apply appropriate lateral torque to the steering wheel to keep the vehicle in the center of the lane. When changing lanes, the lateral control system will safely assist the vehicle to complete the lane change according to the driver's instructions and surrounding traffic conditions, ensuring a smooth and safe lane change process.

[0034] Bad lane line information means that during the lane line recognition process, the system recognizes some messy lines instead of real lane lines. These messy lines may be caused by road construction, wear, stains, reflections, cracks, repair marks, shadows, other marks or objects on the road surface, etc. They will affect the accuracy of lane line recognition, causing the system to misjudge the position and shape of the lane, thereby affecting the normal operation of the vehicle's automatic driving assistance system (such as lane keeping, adaptive cruise, etc.).

[0035] Existing intelligent driving perception systems mainly rely on cameras to collect images and use feature extraction algorithms to identify and confirm lane lines. However, in complex road scenarios, such as Figure 1As shown in the figure, the system has insufficient recognition ability for road miscellaneous lines (such as broken, blurred, disappeared, worn marks, temporary marks, stains, etc.), and is prone to mistakenly judge miscellaneous lines as valid lane lines, resulting in abnormal exit of the HWA lateral control mode. At the same time, the triggering conditions of the existing lateral control mode are relatively simple and do not fully consider the complexity of the road environment. The control function may still be quickly reactivated after the interference of miscellaneous lines is eliminated. This "exit-entry" reciprocating process is particularly obvious in complex road conditions such as urban expressways and construction sections. It not only reduces system reliability, but also distracts the driver due to frequent handovers of control, significantly increasing operational risks. In order to solve the problem of frequent advance and retreat of the lateral control mode, an embodiment of the present invention proposes an optimization method for HWA to enter the lateral control mode, which introduces a strategy of using a perception system to identify and distinguish lane lines from other road markings or random miscellaneous lines (such as cracks, repair marks, stains, road shadows, etc.); introduces optimization conditions for HWA to enter the lateral control mode, which can more efficiently avoid frequent advance and retreat lateral control, thereby improving the driver's comfort experience and enhancing the ability of the HWA lateral control mode to be activated on any road; reduces the application scenarios where the HWA lateral control mode is not suitable for the use of this function, prevents the lateral control from entering in scenarios where the assisted driving function is not applicable, and improves the safety of the HWA lateral control mode. The optimization method for HWA to enter the lateral control mode is as follows:

[0036] See also Figure 2 , Figure 2 The flowchart of the optimization method for HWA to enter the lateral control mode provided by the embodiment of the present invention is as follows. Figure 2 As shown, the method includes the following steps: S110-S140.

[0037] S110, if the vehicle activates the lateral control mode for the first time, determining whether the vehicle currently meets the standard lateral control mode entry condition;

[0038] In this embodiment, the lateral control mode refers to the auxiliary driving function of the vehicle to achieve lane centering by automatically adjusting the steering wheel. The standard lateral control mode entry condition refers to the lane line required by the driving decision system and the comprehensive judgment rule that the vehicle needs to meet when the vehicle activates the lateral control for the first time. When the user activates the lateral control mode for the first time, if the road environment is complex (such as blurred lane lines, vehicle deviation from the center of the lane) or the vehicle is dynamically unstable (such as excessive lateral acceleration), directly entering the lateral control mode may cause vehicle control jitter or even deviate from the lane, causing safety hazards. Therefore, it is necessary to set a standard lateral control mode entry condition to control the vehicle to enter the lateral control mode in a suitable vehicle environment. Usually, when the user triggers the lateral control function for the first time on a highway or urban expressway (such as pressing the steering wheel control button), the driving decision system needs to determine in real time whether the current road conditions allow the function to be safely activated. The vehicle determines whether the vehicle enters the lateral control mode through the standard lateral control mode entry condition, and the standard lateral control mode entry condition may include, for example, lane line quality requirement conditions, lane line length requirement conditions, lane line position requirement conditions, lane line width and vehicle dynamic requirement conditions, etc. The requirements of these conditions are set when the vehicle faces a common vehicle scenario, so the threshold of the condition tends to be standard. By setting the reasonable standard lateral control mode entry condition, the vehicle needs to meet certain conditions to enter the lateral control mode, avoiding the entry condition being too loose, and easily entering the lateral control mode in an inappropriate scenario. At the same time, it also avoids the entry condition being very difficult, and it is difficult to enter the lateral control mode, and maintains moderate standard entry conditions. For example, it can avoid misactivation due to a short lane line or when the vehicle deviates from the lane, and it can also prevent forced activation in a wide lane or intense driving scenario. In a specific example, the vehicle is driving on a straight road on a highway, and the driver presses the steering wheel button for the first time to activate the lateral control mode. It is judged that the lane line and the vehicle meet the above-mentioned standard lateral control mode entry conditions at this time. The driving decision system allows the lateral control mode to be activated, and the vehicle keeps the lane centered steadily. Through this embodiment, it is possible to balance safety and usability, reduce the risk of misactivation, improve the safety of activation, and dynamically adapt to road changes.

[0039] In one embodiment, if Figure 3 As shown, the step S110 includes: S111-S113.

[0040] S111, determining whether the lane lines within the current preset continuous frame number meet the following conditions, including: the lane line quality is high, the length of the lane lines on both sides of the vehicle is greater than a first preset length threshold, the starting point of the lane line is located behind the front bumper of the vehicle, and the distance between the center point of the front bumper of the vehicle and the lane lines on both sides is greater than a preset distance threshold; and

[0041] S112, continuously determining whether the width of the lane line is within a preset width range and whether the lateral acceleration of the vehicle is less than a preset lateral inertia threshold;

[0042] S113: If so, determine that the vehicle meets the standard lateral control mode entry conditions.

[0043] In this embodiment, the determination of the entry conditions of the standard lateral control mode is realized through multi-dimensional parameter verification, which specifically includes: the system detects whether the lane line quality is high-level within a continuous preset number of frames (such as 5 frames) to ensure that the lane line is clear and continuous; verifies that the length of the lane lines on both sides of the vehicle is greater than the first preset length threshold (such as 1 meter) to exclude short or broken interference lines; verifies that the starting point of the lane line is located behind the front bumper of the vehicle to prevent misactivation due to the absence of valid lane lines in front; at the same time, calculates the lateral distance between the center point of the front bumper and the lane lines on both sides. If both are greater than the preset distance threshold (such as 1 meter), it is determined that the vehicle has not deviated from the center of the lane. In addition, the system continuously monitors whether the lane line width is within a preset range (such as 2.5-4.5 meters) to avoid activating control in abnormal lanes (such as construction areas or merging lanes), and detects in real time whether the vehicle's lateral acceleration is lower than the safety threshold (such as 0.3g) to prevent the lateral control from being activated during sharp turns or emergency lane changes. Exemplarily, the entry conditions of the standard lateral control mode include: 1. There are lane lines on both sides, and the lane lines on both sides must be of high quality; 2. The length of the lane lines on both sides is >1m; 3. The starting point of the lane line is behind the front bumper of the vehicle; 4. The distance between the center point of the front bumper and the lane lines on both sides is >1m; 5. Lane width: 2.5-4.5m; 6. Lateral acceleration <0.3g; The lane line is judged to be valid in 5 frames when the above conditions 1.2.3.4 are met. If the above conditions are met in consecutive frames, it is determined that the vehicle meets the standard entry conditions. For example, when driving on a highway, if the lane line is temporarily blurred due to rain erosion, the system detects that the quality of the lane line in 3 consecutive frames is not high or the length is less than 1 meter, and immediately refuses to activate the lateral control to avoid misoperation caused by data fluctuations. Through the coordinated verification of multi-dimensional parameters, this solution significantly improves the accuracy and robustness of entry condition judgment, ensuring that the system only starts the automatic control function in a safe and stable road environment.

[0044] S120: If the conditions are met, control the vehicle to enter the lateral control mode, and detect lane lines and bad lane line information according to a preset lane line recognition algorithm to determine whether the vehicle should exit the lateral control mode;

[0045] In this embodiment, after the vehicle enters the lateral control mode, the driving decision system continuously detects lane lines and bad lane line information through a preset lane line recognition algorithm, and determines whether to exit the lateral control mode based on the detection results. The lane line recognition algorithm refers to a recognition method that uses a vehicle-mounted visual sensor to analyze road images in real time, and integrates physical feature screening, geometric model fitting, pair verification, and time sequence consistency verification to distinguish valid lane lines from bad lane line information. Bad lane line information includes miscellaneous lines (such as road cracks, reflections, temporary markings, stains, shadows) that are determined by the algorithm to be interference or lane lines with abnormal detection (such as fractures, sudden changes in curvature, and blur). When the vehicle is driving in the lateral control mode, if the sensor misdetects due to blurred lane lines, interference from construction markings, or environmental noise, the vehicle control command may be abnormal (such as sudden steering or lane deviation) due to lane line recognition errors. At this time, it is necessary to dynamically determine whether to exit the lateral control to ensure safety. For example, on highways or urban roads, vehicles may misdetect false lane lines due to road reflections, or the algorithm may be confused due to the overlap of new and old markings in the construction area. If these miscellaneous lines cannot be identified and distinguished in time, the system may follow the wrong lane line. Therefore, it is necessary to dynamically determine whether to exit the lateral control mode by continuously detecting the lane line and bad lane line information. This step recognizes the real-time image through the lane line recognition algorithm, which can accurately distinguish the lane line and bad lane line information, and upload it to the driving decision system in real time. The driving decision system determines whether to exit the lateral control mode based on the bad lane information. For example, when the proportion of bad lane line information exceeds the preset proportion (such as 30%) or the effective lane line is continuously lost for more than the preset time (such as 2 seconds), the system immediately triggers the exit of the lateral control mode. Among them, compared with the traditional lane line detection, the lane line recognition algorithm can more effectively distinguish the real lane line and miscellaneous lines in the lane, and improve the accuracy and robustness of lane line recognition. Specifically, this step performs multi-dimensional analysis of real-time road data through a preset lane line recognition algorithm. Compared with the traditional lane line detection method with a single feature threshold, the accuracy of distinguishing between effective lane line and bad lane line information is significantly improved. Traditional solutions usually rely on single physical features such as color or width for screening, which are easily disturbed by reflections, shadows or road stains, resulting in a high false detection rate. This step uses a multi-level fusion verification mechanism, combining spatial geometric constraints and time series analysis in the feature extraction stage to dynamically exclude false lane lines caused by environmental noise or instantaneous occlusion. For example, in complex road scenes (such as overlap of old and new markings, strong light interference), the algorithm effectively suppresses the interference of false lane lines on the control system through multi-level consistency verification. Compared with traditional methods, the false detection rate of lane lines is effectively reduced, and the success rate of identifying bad lane line information is improved. When the system determines that the credible lane lines in the current environment are not enough to maintain lateral control stability (such as continuous loss of effective lane lines or too high a proportion of interference information), the control mode exit is immediately triggered to ensure driving safety.In a specific example, when the vehicle is traveling at 80 km / h in lateral control mode, it suddenly encounters strong reflections on the road surface, causing the camera to misdetect multiple false lane lines. The lane line recognition algorithm of this embodiment determines that it is bad information. When the proportion of bad information reaches 40%, the system exits lateral control within 200ms and prompts the driver to take over to prevent the vehicle from following the wrong lane line and deviating from the driving trajectory. Through this embodiment, the system achieves highly robust lane line tracking and real-time risk assessment during the lateral control process, taking into account both functional availability and safety, and can accurately identify reflective miscellaneous lines as interference information, maintain the stable operation of the lateral control mode, and trigger the exit logic when the real lane line failure is detected to ensure driving safety.

[0046] In one embodiment, if Figure 4 As shown, the step S120 includes: S121-S126.

[0047] S121, acquiring a road image captured by a vehicle-mounted visual sensor, and preprocessing the road image, performing feature extraction on the preprocessed road image to extract a set of candidate lane lines, wherein the preprocessing includes at least image filtering and image enhancement;

[0048] S122, performing feature screening on the lines in the candidate lane line set based on a preset feature screening condition to screen out the lane lines that meet the requirements for verification, wherein the preset feature screening condition is set by lane line width, color, and length;

[0049] S123, fitting the lines in the candidate lane line set by using a geometric model to fit a lane line to be verified that conforms to the geometric model, wherein the geometric model includes a straight line model or a curve model;

[0050] S124, verifying the legitimacy of the lane line to be verified according to the attribute information of the lane line to identify the legal lane line, wherein the attribute information includes the pairing of the lane line, the relationship between the lane line direction and the vehicle driving direction, and the context information of the vehicle driving;

[0051] S125, calculating the lane line change rate of the lane line to be verified based on the continuous multiple frame images, and verifying whether the lane line change rate is less than a preset change rate threshold, so as to determine the lane line that meets the preset change rate threshold;

[0052] S126. Determine the lines that do not meet the preset feature screening conditions, the geometric model, the attribute information and the preset change rate threshold as bad lane line information, and upload the bad lane information and the verified lane lines to the driving decision system of the vehicle.

[0053] In this embodiment, the lane line recognition algorithm of this embodiment improves the accuracy of distinguishing between real lane lines and bad lane line information through a multi-level verification mechanism. The specific process is as follows: the system collects road images through on-board visual sensors (such as front-view cameras), pre-processes the original road images, including Gaussian filtering to eliminate noise interference, and enhances image contrast through histogram equalization to make lane line edge features clearer; then extracts a set of candidate lane lines based on preset feature screening conditions (such as lane line width 0.1-0.3 meters, color white or yellow, length>1 meter). For example, cracks or reflective shadows with a width of less than 0.08 meters are directly excluded. Further, a geometric model is used to fit and verify the candidate lines: for straight lanes, the least squares method is used to fit the straight line equation, and the slope and curvature of the candidate lines are calculated; for curved scenes, a quadratic or cubic polynomial curve model is used for fitting, and the goodness of fit is evaluated through residual analysis (residual<0.1 meter), and unstructured miscellaneous lines that cannot be fitted (such as tree branch shadows) are excluded. The system combines lane attribute information to verify legitimacy, including pair verification, direction consistency verification, and context information matching verification. It checks whether the left and right lane lines appear in pairs and the spacing meets the road standard (such as 3.5 meters), whether the lane extension direction and the vehicle heading angle deviation are less than 5°, and whether the lane line position is consistent with the current lane topology (such as the number of lanes, ramp entrance) with reference to high-precision maps and real-time positioning information. In addition, the system calculates the lane line change rate based on 5 consecutive frames of images, including the lateral position change rate (such as lateral offset speed <0.2m / s) and curvature change rate (curvature fluctuation <0.05 / m), and filters abnormal data caused by instantaneous occlusion (such as vehicle occlusion) or sensor jitter. Finally, lines that do not meet the feature screening conditions, geometric models, attribute requirements, and change rate thresholds (such as temporary markings in construction areas and road reflective tape) are marked as bad lane line information and uploaded to the driving decision system together with the verified legal lane lines. In a specific example, the vehicle was driving on a highway with strong backlight. The on-board camera mistakenly detected multiple reflective tapes (width 0.05 meters, grayish white in color) due to strong light interference. In the preprocessing stage, some noise was suppressed by filtering and contrast enhancement. Feature screening excluded candidate lines with insufficient width. The remaining candidate lines were fitted with cubic curves, and it was found that their curvature residual reached 0.3 meters (exceeding the threshold of 0.1 meters), and were judged invalid. Combined with contextual information (the high-precision map shows that it is currently a straight road), the system further excludes lines with abnormal curvature. In the end, the algorithm only outputs real lane lines that meet the geometric model and direction consistency, and the proportion of bad lane line information drops to less than 5%. The driving decision system maintains stable operation in the lateral control mode. This embodiment significantly improves the lane line recognition accuracy in complex scenarios through a multi-level verification mechanism (feature screening → geometric model → attribute verification → timing analysis), reduces false detection and false exit, and ensures the accuracy and stability of lateral control instructions.

[0054] In one embodiment, if Figure 5 As shown, the step S120 also includes: S127-S128.

[0055] S127, performing image recognition on the preprocessed road image by using a pre-trained convolutional neural network model to identify lane lines and bad lane line information;

[0056] S128. Acquire road images and point cloud data collected by multiple vehicle-mounted visual sensors, align the point cloud data with the coordinate system of the road image, calculate the position deviation values ​​of the lane lines in the road image and in the point cloud data, respectively, and verify whether the position deviation values ​​are less than a preset position deviation threshold value, so as to determine the lane lines that meet the preset position deviation threshold value.

[0057] In this embodiment, the system further improves the lane recognition accuracy through the fusion of multimodal data (camera, radar, LiDAR) and deep learning technology: the pre-processed road image is input into the pre-trained convolutional neural network (CNN) model, which extracts local image features (such as edges and textures) through multi-layer convolution kernels, and uses pooling operations to enhance the spatial invariance of features, and gradually builds deep semantic understanding capabilities (such as distinguishing lane reflections from natural reflections). When training the model, a labeled data set containing different lighting, weather and road scenes (such as sunny lane lines, rainy day reflections, and snow landmarks) is used. The network parameters are optimized through the back propagation algorithm, so that the model can learn the common characteristics of lane lines (such as continuous linear structure, color consistency) and the interference pattern of miscellaneous lines (such as random breaks and irregular shapes). Finally, the model outputs a pixel-level lane line probability map and an interference area heat map, and distinguishes real lane lines from bad information (such as probability > 0.8 is determined as a lane line) based on the probability threshold. The probability map is annotated with the probability value (range 0 to 1) that each pixel belongs to the real lane line, and the heat map marks potential interference areas (such as high reflection or cracks). The system sets a probability threshold (such as 0.8) to binarize the probability map: the pixel area with a probability value ≥ 0.8 is determined to be the real lane line, and the candidate lane line contour is generated; at the same time, based on the interference area marked by the heat map, the overlapping part of the contour with the interference area is filtered out (if the overlapping area > 30%, it is removed). Of course, it is understandable that this is a way to use a convolutional neural network model to distinguish lane lines from bad lane line information. It is understandable that other models or methods can also be used to make the distinction. The system synchronously obtains the point cloud data generated by the laser radar or millimeter wave radar, aligns the point cloud coordinate system with the camera image coordinate system through the spatiotemporal calibration algorithm (such as synchronizing the external parameter matrix with the timestamp), extracts the corresponding lane line spatial position information in the point cloud, and calculates the lateral position deviation value between it and the image recognition result (for example, the lateral coordinate of the lane line in the image is X1, the corresponding position in the point cloud is X2, and the deviation ΔX = |X1-X2|). If ΔX exceeds the preset threshold (such as 0.15 meters), the lane line is determined to be a miscellaneous line misdetected by the sensor; otherwise, it is confirmed as a valid lane line. In a specific example, when the vehicle is driving on a highway on a rainy night, the camera misdetects the road surface reflection as a lane line due to water stains (image coordinates X1 = 3.2 meters). The laser radar point cloud data detects the real lane line position X2 = 3.5 meters due to penetrating the water film. The system calculates the deviation ΔX = 0.3 meters (exceeding the threshold of 0.15 meters), determines it as bad lane line information and excludes it, and finally maintains lateral control based on the lane line coordinates corrected by the radar point cloud.This embodiment uses the CNN model to deeply learn the essential characteristics of lane lines, combined with the spatial consistency verification of multi-sensor data, to significantly improve the ability to distinguish lane lines in complex scenarios, effectively solve the environmental limitations of a single sensor, reduce false detections and misjudgments caused by environmental interference, and ensure the stable operation of the lateral control mode under harsh conditions.

[0058] S130, if the vehicle exits the lateral control mode due to the bad lane line information and reactivates the lateral control mode in the current ignition cycle, determining whether the vehicle currently satisfies both the standard lateral control mode entry condition and the additional lateral control mode entry condition;

[0059] In this embodiment, the current ignition cycle refers to a complete operation cycle of the vehicle from starting to shutting down. The additional lateral control mode entry condition refers to the enhanced judgment condition that needs to be met when the vehicle reactivates the lateral control mode after exiting the lateral control mode due to poor lane line information, including stricter restriction requirements. When the vehicle exits the lateral control mode due to poor lane line information (such as temporary marking interference), if the driver attempts to reactivate within a short period of time, the traditional solution may repeatedly trigger the exit due to the lack of improvement in the environment, resulting in frequent system entry and exit, reducing user experience and increasing safety hazards. For example, in a construction section, the vehicle exits the lateral control due to the overlap of new and old markings. If the driver immediately attempts to reactivate, the system may misjudge again because the miscellaneous lines have not been completely eliminated, causing control command conflicts, and easily causing the "activation-exit-reactivation" cycle problem on complex roads. Therefore, it is necessary to introduce additional conditions when reactivating to ensure environmental stability and lane line reliability and avoid invalid activation. This step is determined by dual conditions. When reactivating, the system must simultaneously meet the standard lateral control mode entry conditions (such as basic parameters such as lane line quality, length, and position) and the additional lateral control mode entry conditions. The additional lateral control mode condition is set by historical data modeling, and its threshold is higher than the standard lateral control mode condition to filter short-term interference scenarios. The threshold of the additional lateral control mode condition can be dynamically adjusted according to the vehicle operating status (such as vehicle speed, road type) and environmental feedback (such as the number of recent exits). For example, if the vehicle exits the highway due to miscellaneous lines many times, the system automatically raises the threshold and enhances the activation threshold. In a specific example, after the vehicle exits the lateral control mode on the highway due to temporary marking interference, the driver attempts to reactivate without leaving the construction area. The system detects that the current lane line length is 20 meters (lower than the 25 meters required by the additional lateral control mode entry condition). Although the standard conditions are met, activation is still prohibited until the vehicle leaves the construction area and the lane line length is restored to 30 meters. Activation is allowed. Through this embodiment, the activation threshold is raised by using the additional lateral control mode entry condition, so that the driving decision system can automatically optimize the condition strictness according to the actual road conditions, balance safety and function availability, and reactivate the lateral control only in a reliable environment, reduce the driver's doubts about the stability of the system, enhance the function acceptance, and optimize the function robustness and user trust.

[0060] In one embodiment, if Figure 6 As shown, the step S130 includes: S131-S132.

[0061] S131, determining whether the lane line maintains the following conditions within a preset duration, including: the length of the lane lines on both sides of the vehicle is greater than a second preset length threshold and the confidence of the lane line is greater than a preset confidence threshold, wherein the second preset length threshold is greater than the first preset length threshold;

[0062] S132: If so, determine that the vehicle meets the additional lateral control mode entry condition.

[0063] In this embodiment, the setting of the additional lateral control mode entry condition is intended to prevent repeated misactivation within a short period of time after the vehicle exits the lateral control due to environmental interference by strengthening the stability and reliability verification of the lane line. When the vehicle exits the lateral control mode due to poor lane line information (such as construction markings or reflective interference), when the driver tries to activate it again, the system needs to continuously detect whether the lane line meets the additional conditions with higher requirements within a preset duration (such as 5 seconds): First, the length of the lane line on both sides of the vehicle needs to be greater than the second preset length threshold (such as 30 meters, much higher than the 1 meter of the first activation) to exclude the interference of short-term miscellaneous lines (such as temporary markings or cracks); second, the lane line confidence needs to be greater than the preset confidence threshold (such as 40%, higher than 10% of the first activation), which is calculated based on multi-sensor data fusion (such as the consistency of camera detection results and radar point cloud) and historical continuous frame verification results. During this process, the system analyzes the continuity of lane line length through continuous multi-frame images (such as 100ms interval per frame, length fluctuation <10% within 50 consecutive frames), and combines the time stability of confidence (such as confidence fluctuation <15% within 5 seconds) to ensure the strictness of the conditions and scene adaptability. If the above conditions are met within the preset duration, it is determined that the vehicle meets the conditions for entering the additional lateral control mode. In a specific example, the conditions for entering the additional lateral control mode include: 1. The average length of the left and right lane lines ≥30m; 2. The confidence of the left and right lane lines ≥40%; 3. The duration of the above conditions is 500ms-1s. The vehicle exits lateral control due to the overlap of old and new markings in the construction section. When the driver tries to reactivate, the system detects that the current lane line length is 25 meters (lower than the second preset threshold of 30 meters), and the confidence is only 30% (lower than the 40% threshold) due to blurred markings. It is determined that the additional conditions are not met and activation is refused; after the vehicle leaves the construction area, the lane line length is restored to 35 meters and the confidence is increased to 75%, and it is maintained for more than 1 second, and the system allows reactivation. When the intelligent assisted driving exits the lateral control mode due to lane line reasons, adding the judgment of the entry conditions of the additional lateral control mode will better screen the current road environment, prevent the self-vehicle from following the chaotic line control vehicle and frequently enter and exit the lateral control. Therefore, by introducing higher length and confidence thresholds and continuous verification mechanisms, this embodiment significantly reduces the probability of false activation in complex scenarios, ensuring that the lateral control function is reactivated only when the lane line is stable and reliable, avoiding functional oscillations caused by frequent advances and retreats, and improving the driver's trust in the system decision.

[0064] In other embodiments, the additional lateral control mode entry condition may also be other conditions, such as driver attention monitoring conditions, traffic environment risk assessment conditions, dynamic environment adaptation conditions, and vehicle state comprehensive judgment conditions.

[0065] Among them, for driver attention monitoring, when reactivating lateral control, if the driver is not paying attention (such as not looking forward for a long time), he may not be able to take over the vehicle in time, causing safety hazards. The driver's status is monitored in real time through the in-car camera or steering wheel grip sensor. If the driver's eyes deviate from the front for more than 2 seconds, or his hands leave the steering wheel for more than 10 seconds, the system will determine that the attention is insufficient. Before reactivating lateral control, the following conditions must be met at the same time: the driver's eyes return to the front within 1 second after the activation request; the hands continue to contact the steering wheel (torque>0.5Nm). Ensure that the driver is in a state where he can take over at any time when reactivating to reduce safety risks caused by distraction.

[0066] In terms of traffic environment risk assessment, reactivating lateral control may increase the risk of collision when traffic is dense or adjacent vehicles are cutting in at close range. The distance, speed and cutting-in intention of vehicles in adjacent lanes are monitored in real time through radar and cameras. Reactivation requires: the lateral distance between the vehicle and the vehicle in the adjacent lane is >1.5m; the relative speed of the adjacent vehicle is <20km / h (no emergency cutting behavior). Avoid activating lateral control in dangerous traffic environments to prevent the system from making emergency avoidance.

[0067] For dynamic environmental adaptation conditions, harsh environments such as rain, snow, and backlight may reduce sensor performance, and the activation conditions need to be adjusted dynamically. The camera analyzes the intensity of rain and snow, lighting conditions (such as the backlight index), or the humidity sensor detects the weather conditions. Dynamically adjust the activation threshold according to the environment: Rainy / snowy days: the lane line confidence threshold is increased from 40% to 60%; strong backlight: the lane line length threshold is increased from 30m to 50m; at night: enable infrared fill light to enhance the image, but the lane line reflection coefficient is required to be >0.3. Reduce the false activation rate of lateral control in harsh environments, and significantly enhance the ability to adapt to extreme scenarios.

[0068] In terms of comprehensive judgment conditions for vehicle status, activating lateral control may lead to control conflicts when the vehicle accelerates, decelerates or turns too aggressively. Real-time collection of vehicle speed, longitudinal acceleration, steering angular velocity and other parameters. Reactivation must meet the following conditions at the same time: vehicle speed fluctuation range <±10km / h (within 10 seconds); longitudinal acceleration absolute value <0.2g; steering angular velocity <10° / s. Activation is allowed only when the vehicle is in a stable driving state. After the lateral control is involved, the steering wheel shake phenomenon is reduced, improving safety.

[0069] In one embodiment, if Figure 7 As shown, the step S130 also includes: S1311-S1318.

[0070] S1311, determining a positioning error index according to a position deviation between a detected lane line and a real lane line, wherein the lane line is detected from a collected road image;

[0071] S1312, determining a shape error index according to a geometric difference between the detected lane line shape and the actual lane line shape;

[0072] S1313, determining a coverage index according to a ratio of the detected lane line length to the actual lane line length;

[0073] S1314, determining a missed detection rate indicator according to a ratio of the length of the undetected lane line to the length of the actual lane line;

[0074] S1315, determining a breakpoint quantity index according to the number of lane line breaks in a plurality of consecutive image frames within a preset time period;

[0075] S1316, determining a tracking stability index according to a change rate of lane lines between adjacent frame images;

[0076] S1317. Determine a robustness index according to lane line detection success rates in different scenarios in historical detection data;

[0077] S1318. Based on the preset weight coefficient, the positioning error index, the shape error index, the coverage index, the missed detection rate index, the breakpoint number index, the tracking stability index and the robustness index are weightedly summed to determine the confidence of the lane line.

[0078] In this embodiment, the calculation of lane line confidence is achieved through the fusion of multi-dimensional indicators, aiming to quantitatively evaluate the reliability and environmental adaptability of the detection results: First, the system calculates the positioning error index based on the position deviation (such as lateral offset) between the lane line detected in the road image collected by the camera and the real lane line, such as using mean square error (MSE) quantification or mean absolute error (MAE) to calculate the degree of deviation (such as the error value increases significantly when the deviation is > 0.2 meters); secondly, the shape error index is calculated by comparing the geometric shape difference (such as curvature radius deviation, slope angle difference) between the detected lane line and the real lane line (the real lane line can be obtained through manual annotation or high-precision map or multi-sensor fusion), such as the residual sum of squares between the detected curve and the real curve in the curved scene. The coverage index is calculated by the ratio of the effective length of the detected lane line to the real length (such as 70 meters out of 80 meters are detected, and the coverage is 87.5%), and the missed detection rate is its complement (12.5%). The number of breakpoints indicator counts the number of lane line breaks in multiple consecutive frames of images within a preset time period (such as within 10 seconds) (such as 1 point for each break), reflecting the continuity of the lane line. The tracking stability indicator calculates the rate of change of the lateral position or curvature of the lane line between adjacent frames (such as position change speed> 0.3m / s is considered unstable), and the robustness indicator is based on the weighted success rate of different scenes (such as rainy days and nighttime) in the historical detection data (such as the success rate of rainy day detection is 70%, weight 0.3). Finally, the system weighted sums the above indicators according to the preset weight coefficients (such as positioning error weight 0.2, coverage weight 0.25) to generate a comprehensive confidence score (0 to 100 points). When the score is higher than the threshold (such as 40 points), the lane line is determined to be reliable. In a specific example, when the vehicle is driving at night on a rainy day, the positioning error of the detected lane line increases due to the reflection of water stains (MSE=0.25, weight 0.2), the shape error increases significantly due to the curvature fitting deviation (residual=0.15, weight 0.3), the coverage rate drops to 65% (weight 0.1), the missed detection rate increases to 35% (weight 0.1), the number of breakpoints increases to 3 times due to lane line breaks (weight 0.05), the tracking stability is aggravated by the wet and slippery road surface fluctuations (rate of change 0.4m / s, weight 0.15), and the robustness index is based on the historical rainy day detection success rate (70%, weight 0.1). Based on the above indicators and their weights, the system calculates the comprehensive confidence score as: 0.2*(1-0.25)+0.3*(1-0.15)+0.1*0.65+0.1*(1-0.35)+0.05*(1-0.3)+0.15*(1-0.4)+0.1*0.7=0.15+0.255+0.065+0.065+0.035+0.09+0.07=0.73. Since the score of 73 is higher than the confidence threshold of 40, the system determines that the lane line is reliable and allows lateral control to be activated.Through the weight allocation and comprehensive scoring mechanism, this embodiment can still accurately evaluate the reliability of lane lines in complex scenarios, ensuring the strictness and adaptability of activation conditions. This embodiment quantifies the reliability of lane lines through multi-indicator fusion, significantly improves the detection robustness in complex scenarios, avoids misjudgment caused by the failure of simple indicators, and ensures that the lateral control activation conditions strictly match the real road environment.

[0079] In one embodiment, if Figure 8 As shown, the step S130 also includes: S1301-S1305.

[0080] S1301, acquiring a road image collected by a vehicle-mounted visual sensor, and preprocessing the road image, wherein the preprocessing includes at least noise suppression, brightness adjustment, and contrast enhancement;

[0081] S1302, performing lane line feature extraction on the pre-processed road image by using an edge detection algorithm to extract a lane line feature area;

[0082] S1303, detecting a lane line to be selected from the lane line feature area using Hough transform;

[0083] S1304: Based on the continuous multi-frame images, the lane line to be selected is verified according to the preset lane line geometric constraint condition, and the lane line that meets the geometric constraint condition is output as the detected lane line.

[0084] In this embodiment, lane lines need to be detected before calculating confidence. Lane line detection in this embodiment is achieved through image preprocessing and multi-level feature screening mechanism. It should be noted that the lane line detection method of this embodiment is different from the lane line recognition algorithm in the aforementioned step S120. Of course, it is understandable that in another embodiment, lane lines can also be detected by the lane line recognition algorithm in step S120 before calculating confidence. Specifically, first, the system obtains a road image captured by an on-board visual sensor (such as a front-view camera), suppresses noise through Gaussian filtering (such as removing raindrops or dust interference), and uses histogram equalization to adjust brightness and contrast to make the lane line edge features clearer; then, an edge detection algorithm (such as a Sobel operator or a Canny operator) is applied to the preprocessed image to extract the lane line feature area. The algorithm identifies significant edges (such as lane line boundaries) in the image through gradient calculation and dual threshold screening (such as a low threshold of 50 and a high threshold of 150). Furthermore, the Hough transform is used to detect the lane lines to be selected from the edge feature area: the Hough transform identifies the straight or curved structure in the image through parameter space mapping (such as converting from the Cartesian coordinate system to the polar coordinate system), and determines the candidate lines (such as lines with votes > 100) based on the voting mechanism. Finally, the system performs geometric constraint checks on the lane lines to be selected based on continuous multi-frame images, including lane line length (such as > 1 meter), continuity (such as break length < 0.2 meters) and curvature consistency (such as curvature fluctuation < 0.05 / m), excludes pseudo lane lines that do not meet the constraints (such as tree branch shadows or road cracks), and outputs lane lines that meet the geometric constraints as the final detection results. In a specific example, a vehicle is driving on an urban expressway at dusk. The camera captures a blurred image due to low light and street lamp reflections. In the preprocessing stage, the lane line contrast is enhanced through Gaussian filtering and histogram equalization; Canny edge detection extracts lane line boundary features, and Hough transform identifies multiple candidate lines; based on the geometric constraint verification of 5 consecutive frames of images, the system excludes short pseudo lines (length <1 meter) and interference lines with abnormal curvature (fluctuation >0.05 / m) caused by reflections, and finally outputs the real lane lines that meet the constraints for use by the driving decision system. This embodiment enhances image quality through preprocessing, combines edge detection and geometric constraint verification, significantly improves the accuracy and robustness of lane line detection under complex lighting and environmental noise, and ensures the recognition accuracy of the lane lines provided for confidence calculation.

[0085] S140: If all the above conditions are met at the same time, reactivate the lateral control mode, and continue to determine whether the vehicle exits the lateral control mode according to the lane line recognition algorithm.

[0086] In this embodiment, after the lateral control mode is reactivated, the driving decision system continues to detect the lane line and bad lane line information in real time through the lane line recognition algorithm, and the driving decision system evaluates the stability of the lateral control mode based on the detected lane line and bad lane line information to decide whether to exit again. In a specific example, after the vehicle reactivates the lateral control mode on an urban expressway, the driving decision system continues to use the above lane line recognition algorithm to detect that the lane line is clear and continuous (the proportion of bad lane line information is less than 10%), and the confidence of the effective lane line is stably maintained at more than 80%. The driving decision system determines that the current road environment meets the lateral control requirements, and continuously outputs smooth steering wheel adjustment instructions. The vehicle drives steadily along the center of the lane without triggering the exit logic. By continuing the high-precision lane line and interference information recognition capabilities, it ensures that the system only needs to maintain the lateral control function when responding to low-risk scenarios after reactivation, which not only avoids invalid exits caused by excessive sensitivity, but also provides a coherent assisted driving experience when the lane line is clear, enhancing functional practicality and user satisfaction.

[0087] To summarize, on the one hand, the perception system is used to identify and distinguish lane lines from other road markings or random lines (such as cracks, repair marks, stains, road shadows, etc.) to improve recognition performance; on the other hand, the entry conditions of the HWA lateral control mode are optimized, which can more efficiently avoid frequent forward and backward lateral control, thereby improving the driver's comfort experience and enhancing the ability of the HWA lateral control mode to be activated on any road; on the other hand, the application scenarios where the HWA lateral control mode is not suitable for the use of this function are reduced, preventing the entry of lateral control in scenarios where the assisted driving function is not applicable, and improving the safety of the HWA lateral control mode.

[0088] Fig. 9 FIG. 2 is a schematic block diagram of an optimization device 200 for HWA entering a horizontal control mode provided by an embodiment of the present invention. Fig. 9 As shown, corresponding to the above optimization method for HWA to enter the lateral control mode, the present invention also provides an optimization device 200 for HWA to enter the lateral control mode. The optimization device 200 for HWA to enter the lateral control mode includes a unit for executing the above optimization method for HWA to enter the lateral control mode, and the device can be configured in a computer device. Specifically, please refer to Fig. 9 The optimization device 200 for HWA to enter the lateral control mode includes: a first entry judgment unit 201, an exit detection unit 202, a second entry judgment unit 203 and a reactivation unit 204.

[0089] Among them, the first entry judgment unit 201 is used to judge whether the vehicle currently meets the standard lateral control mode entry condition if the vehicle activates the lateral control mode for the first time; the exit detection unit 202 is used to control the vehicle to enter the lateral control mode if it meets the condition, and detect the lane line and bad lane line information according to a preset lane line recognition algorithm to judge whether the vehicle exits the lateral control mode; the second entry judgment unit 203 is used to judge whether the vehicle currently meets both the standard lateral control mode entry condition and the additional lateral control mode entry condition if the vehicle exits the lateral control mode due to the bad lane line information and reactivates the lateral control mode within the current ignition cycle; the reactivation unit 204 is used to reactivate the lateral control mode if both conditions are met, and continue to judge whether the vehicle exits the lateral control mode according to the lane line recognition algorithm.

[0090] In one embodiment, the first entry determination unit 201 includes: a first determination subunit, a second determination subunit and a first determination unit.

[0091] Among them, the first judgment subunit is used to judge whether the lane line meets the following conditions within the current continuous preset number of frames, including: the lane line quality is high, the lane line length on both sides of the vehicle is greater than the first preset length threshold, the starting point of the lane line is located behind the front bumper of the vehicle, and the distance between the center point of the front bumper of the vehicle and the lane lines on both sides is greater than the preset distance threshold; and the second judgment subunit is used to continuously judge whether the width of the lane line is within the preset width range and whether the lateral acceleration of the vehicle is less than the preset lateral inertia threshold; the first judgment unit is used to judge that the vehicle meets the standard lateral control mode entry conditions if satisfied.

[0092] In one embodiment, the second entry determination unit 203 includes: a third determination subunit and a second determination unit.

[0093] Among them, the third judgment subunit is used to judge whether the lane line maintains the following conditions within a preset duration, including: the length of the lane lines on both sides of the vehicle is greater than a second preset length threshold and the confidence of the lane line is greater than a preset confidence threshold, wherein the second preset length threshold is greater than the first preset length threshold; the second judgment unit is used to judge whether the vehicle meets the conditions for entering the additional lateral control mode if the conditions are met.

[0094] In one embodiment, the second entry judgment unit 203 further includes: a positioning error unit, a shape error unit, a coverage rate unit, a missed detection rate unit, a breakpoint unit, a variation unit, a robustness unit and a weighting unit.

[0095] Among them, the positioning error unit is used to determine the positioning error index according to the position deviation between the detected lane line and the real lane line, wherein the lane line is detected in the collected road image; the shape error unit is used to determine the shape error index according to the geometric difference between the detected lane line shape and the real lane line shape; the coverage unit is used to determine the coverage index according to the ratio of the detected lane line length to the real lane line length; the missed detection rate unit is used to determine the missed detection rate index according to the ratio of the undetected lane line length to the real lane line length; the breakpoint unit is used to determine the breakpoint quantity index according to the number of times the lane line is broken in multiple consecutive frames of images within a preset time period; the change unit is used to determine the tracking stability index according to the change rate of the lane line between adjacent frame images; the robustness unit is used to determine the robustness index according to the lane line detection success rate in different scenarios in the historical detection data; the weighting unit is used to perform weighted summation of the positioning error index, the shape error index, the coverage index, the missed detection rate index, the breakpoint quantity index, the tracking stability index and the robustness index based on a preset weight coefficient to determine the confidence of the lane line.

[0096] In one embodiment, the second entry judgment unit 203 further includes: a first preprocessing unit, a feature extraction unit, a Hough transformation unit and a geometric constraint unit.

[0097] Among them, the first preprocessing unit is used to obtain a road image collected by a vehicle-mounted visual sensor and preprocess the road image, and the preprocessing at least includes noise suppression, brightness adjustment and contrast enhancement; the feature extraction unit is used to perform lane line feature extraction on the preprocessed road image through an edge detection algorithm to extract a lane line feature area; the Hough transform unit is used to detect a lane line to be selected from the lane line feature area using Hough transform; the geometric constraint unit is used to verify the lane line to be selected based on continuous multiple frame images according to preset lane line geometric constraint conditions, and output the lane line that meets the geometric constraint conditions as the detected lane line.

[0098] In one embodiment, the exit detection unit 202 includes: a second preprocessing unit, a screening unit, a fitting unit, a first verification unit, a second verification unit, an uploading unit, a convolution recognition unit and a third verification unit.

[0099] Among them, the second preprocessing unit is used to obtain the road image collected by the on-board visual sensor, preprocess the road image, and perform feature extraction on the preprocessed road image to extract a set of candidate lane lines, wherein the preprocessing at least includes image filtering and image enhancement; the screening unit is used to perform feature screening on the lines in the candidate lane line set based on preset feature screening conditions to screen out the lane lines to be verified that meet the requirements, wherein the preset feature screening conditions are set by the lane line width, color and length; the fitting unit is used to fit the lines in the candidate lane line set through a geometric model to fit the lane lines to be verified that meet the geometric model, wherein the geometric model includes a straight line or curve model; the first verification unit A unit is used to verify the legitimacy of the lane line to be verified according to the attribute information of the lane line to identify the legal lane line, wherein the attribute information includes the pairing of the lane line, the relationship between the lane line direction and the vehicle driving direction, and the context information of the vehicle driving; a second verification unit is used to calculate the lane line change rate of the lane line to be verified based on continuous multiple frame images, and verify whether the lane line change rate is less than a preset change rate threshold to determine the lane line that meets the preset change rate threshold; an uploading unit is used to determine the lines that do not meet the preset feature screening conditions, the geometric model, the attribute information and the preset change rate threshold as bad lane line information, and upload the bad lane information and the verified lane line to the driving decision system of the vehicle. A convolution recognition unit is used to perform image recognition on the preprocessed road image through a pre-trained convolutional neural network model to identify lane lines and bad lane line information; a third verification unit is used to obtain road images and point cloud data collected by multiple vehicle-mounted visual sensors, align the point cloud data with the coordinate system of the road image, calculate the position deviation value of the lane line in the road image and in the point cloud data respectively, and verify whether the position deviation value is less than a preset position deviation threshold to determine the lane line that meets the preset position deviation threshold.

[0100] The above-mentioned HWA optimization device 200 for entering the horizontal control mode can be implemented in the form of a computer program. The computer program can be used in the following example. Fig.10 Runs on the computer device shown.

[0101] See also Fig.10 , Fig.10 5 is a schematic block diagram of a computer device provided in an embodiment of the present application. The computer device 500 may be an electronic device with data processing and communication functions in a car.

[0102] See also Fig.10The computer device 500 includes a processor 502 , a memory and a network interface 505 connected via a system bus 501 , wherein the memory may include a non-volatile storage medium 503 and an internal memory 504 .

[0103] The non-volatile storage medium 503 may store an operating system 5031 and a computer program 5032. The computer program 5032 includes program instructions, and when the program instructions are executed, the processor 502 may execute an optimization method for HWA to enter a lateral control mode.

[0104] The processor 502 is used to provide computing and control capabilities to support the operation of the entire computer device 500 .

[0105] The internal memory 504 provides an environment for the operation of the computer program 5032 in the non-volatile storage medium 503. When the computer program 5032 is executed by the processor 502, the processor 502 can execute an optimization method for the HWA to enter the lateral control mode.

[0106] The network interface 505 is used to communicate with other devices over the network. Fig.10 The structure shown in the figure is merely a block diagram of a portion of the structure related to the solution of the present application, and does not constitute a limitation on the computer device 500 to which the solution of the present application is applied. The specific computer device 500 may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0107] The processor 502 is used to run a computer program 5032 stored in the memory to implement the steps of the above method.

[0108] It should be understood that in the embodiment of the present application, the processor 502 may be a central processing unit (CPU), and the processor 502 may also be other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. Among them, the general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0109] It can be understood by those skilled in the art that all or part of the processes in the method for implementing the above embodiment can be completed by instructing the relevant hardware through a computer program. The computer program includes program instructions, and the computer program can be stored in a storage medium, which is a computer-readable storage medium. The program instructions are executed by at least one processor in the computer system to implement the process steps of the embodiment of the above method.

[0110] Therefore, the present invention also provides a storage medium. The storage medium may be a computer-readable storage medium. The storage medium stores a computer program, wherein the computer program includes program instructions. When the program instructions are executed by a processor, the processor executes the steps of the above method.

[0111] The storage medium may be a USB flash drive, a mobile hard disk, a read-only memory (ROM), a magnetic disk, or an optical disk, etc., which are computer-readable storage media that can store program codes.

[0112] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in terms of function in the above description. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0113] In the several embodiments provided by the present invention, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of each unit is only a logical function division, and there may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed.

[0114] The steps in the method of the embodiment of the present invention can be adjusted in order, combined and deleted according to actual needs. The units in the device of the embodiment of the present invention can be combined, divided and deleted according to actual needs. In addition, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0115] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a storage medium. Based on this understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including several instructions for a computer device to execute all or part of the steps of the method described in each embodiment of the present invention.

[0116] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0117] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

[0118] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can easily think of various equivalent modifications or replacements within the technical scope disclosed by the present invention, and these modifications or replacements should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention shall be based on the protection scope of the claims.

Claims

1. An optimization method for HWA to enter lateral control mode, characterized in that: The method comprises: If the vehicle activates the lateral control mode for the first time, determining whether the vehicle currently meets the standard lateral control mode entry conditions; If satisfied, control the vehicle to enter the lateral control mode, and detect lane lines and bad lane line information according to a preset lane line recognition algorithm to determine whether the vehicle exits the lateral control mode; If the vehicle exits the lateral control mode due to the bad lane line information and reactivates the lateral control mode in the current ignition cycle, determining whether the vehicle currently satisfies both the standard lateral control mode entry condition and the additional lateral control mode entry condition; If both conditions are met, the lateral control mode is reactivated, and the lane line recognition algorithm is used to continue to determine whether the vehicle exits the lateral control mode.

2. The method according to claim 1, characterized in that The step of determining whether the vehicle currently meets the standard lateral control mode entry condition comprises: Determine whether the lane line within the current preset continuous frame number meets the following conditions, including: the lane line quality is high, the length of the lane line on both sides of the vehicle is greater than a first preset length threshold, the starting point of the lane line is located behind the front bumper of the vehicle, and the distance between the center point of the front bumper of the vehicle and the lane lines on both sides is greater than a preset distance threshold; and Continuously determining whether the width of the lane line is within a preset width range and whether the lateral acceleration of the vehicle is less than a preset lateral inertia threshold; If so, it is determined that the vehicle meets the standard lateral control mode entry condition.

3. The method according to claim 2, characterized in that The step of determining whether the vehicle currently satisfies both the standard lateral control mode entry condition and the additional lateral control mode entry condition comprises: Determining whether the lane line maintains the following conditions within a preset duration, including: the length of the lane lines on both sides of the vehicle is greater than a second preset length threshold and the confidence of the lane line is greater than a preset confidence threshold, wherein the second preset length threshold is greater than the first preset length threshold; If so, it is determined that the vehicle meets the additional lateral control mode entry condition.

4. The method according to claim 3, characterized in that The confidence of the lane line is determined by the following steps: Determining a positioning error index according to a position deviation between a detected lane line and a real lane line, wherein the lane line is detected from a collected road image; Determining a shape error index based on a geometric difference between a detected lane shape and a true lane shape; The coverage index is determined according to the ratio of the detected lane line length to the actual lane line length; The missed detection rate index is determined according to the ratio of the length of the undetected lane line to the length of the actual lane line; Determine the number of breakpoints according to the number of lane line breaks in a plurality of consecutive frames of images within a preset time period; Determine the tracking stability index based on the change rate of the lane lines between adjacent frame images; Determine the robustness index based on the lane line detection success rate in different scenarios in the historical detection data; Based on the preset weight coefficient, the positioning error index, the shape error index, the coverage index, the missed detection rate index, the breakpoint number index, the tracking stability index and the robustness index are weighted and summed to determine the confidence of the lane line.

5. The method according to claim 4, characterized in that The lane line is detected by the following steps: Acquire a road image collected by a vehicle-mounted visual sensor, and preprocess the road image, wherein the preprocessing includes at least noise suppression, brightness adjustment, and contrast enhancement; Performing lane line feature extraction on the preprocessed road image by using an edge detection algorithm to extract a lane line feature area; Detecting a lane line to be selected from the lane line feature area using Hough transform; Based on the continuous multi-frame images, the lane line to be selected is checked according to the preset lane line geometric constraint condition, and the lane line that meets the geometric constraint condition is output as the detected lane line.

6. The method according to any one of claims 1 to 5, characterized in that: The step of detecting lane lines and bad lane line information according to a preset lane line recognition algorithm to determine whether the vehicle exits the lateral control mode includes: Acquire a road image collected by a vehicle-mounted visual sensor, preprocess the road image, and perform feature extraction on the preprocessed road image to extract a set of candidate lane lines, wherein the preprocessing includes at least image filtering and image enhancement; Performing feature screening on the lines in the candidate lane line set based on a preset feature screening condition to screen out the lane lines that meet the requirements for verification, wherein the preset feature screening condition is set by lane line width, color and length; Fitting the lines in the candidate lane line set by means of a geometric model to fit a lane line to be verified that conforms to the geometric model, wherein the geometric model includes a straight line or a curve model; Performing legality verification on the lane line to be verified according to the attribute information of the lane line to identify the legal lane line, wherein the attribute information includes the pairing of the lane line, the relationship between the lane line direction and the vehicle driving direction, and the context information of the vehicle driving; Based on the continuous multiple-frame images, the lane line change rate of the lane line to be verified is calculated, and whether the lane line change rate is less than a preset change rate threshold is verified to determine the lane line that meets the preset change rate threshold; Lines that do not meet the preset feature screening conditions, the geometric model, the attribute information and the preset change rate threshold are determined as bad lane line information, and the bad lane information and the verified lane lines are uploaded to the driving decision system of the vehicle.

7. The method according to claim 6, characterized in that The method further comprises: Performing image recognition on the preprocessed road image through a pre-trained convolutional neural network model to identify lane lines and bad lane line information; Acquire road images and point cloud data collected by multiple vehicle-mounted visual sensors, align the point cloud data with the coordinate system of the road image, calculate the position deviation values ​​of the lane lines in the road image and in the point cloud data, respectively, and verify whether the position deviation values ​​are less than a preset position deviation threshold value, so as to determine the lane lines that meet the preset position deviation threshold value.

8. An optimization device for HWA entering lateral control mode, characterized in that: The method comprises a unit for executing the method according to any one of claims 1 to 7.

9. A computer device, characterized in that: The computer device comprises a memory and a processor, the memory stores a computer program, and the processor implements the method according to any one of claims 1 to 7 when executing the computer program.

10. A computer-readable storage medium, characterized in that: The storage medium stores a computer program, and when the computer program is executed by a processor, the method according to any one of claims 1 to 7 can be implemented.

Images