Lane departure early warning method and system and vehicle

Abstract

The invention discloses a lane departure early warning method and system and a vehicle, and relates to the technical field of vehicle auxiliary driving. The method comprises the following steps: inputting obtained driver state data, driver operation data and vehicle state data into a pre-trained lane departure risk prediction model to obtain a real-time lane departure risk degree; determining a target early warning sensitivity corresponding to the real-time lane departure risk degree according to a preset mapping relation between the lane departure risk degree and the early warning sensitivity; based on the target early warning sensitivity, dynamically adjusting a trigger threshold of lane departure early warning; and comparing the obtained lane line departure distance with the dynamically adjusted trigger threshold value, and judging whether to send out lane departure early warning or not according to a comparison result. According to the method, the trigger threshold value of lane departure early warning can be adaptively adjusted, the situations of false alarm and missing alarm in the using process are reduced, and the experience feeling and comfort of a driver are effectively improved.

Description

Technical Field

[0001] This invention belongs to the field of vehicle driver assistance technology, and particularly relates to a lane departure warning method, system and vehicle that can adaptively adjust the warning sensitivity according to the driver's state and driving style. Background Technology

[0002] Lane departure warning systems typically assess the lateral distance between the vehicle and the lane markings and issue an alert. However, different drivers have different driving habits: conservative drivers tend to stay near the center of the lane, while aggressive drivers tend to frequently change position within the lane, even driving close to the lane markings. Therefore, the degree of lane departure varies significantly between different types of drivers in daily driving. Using a fixed departure threshold for lane departure warnings could easily lead to aggressive drivers triggering the alert too frequently, causing excessive distraction; while conservative drivers might not trigger the warning due to smaller deviations, thus reducing the system's safety warning effectiveness.

[0003] Furthermore, traditional lane departure warning systems typically employ fixed sensitivity settings, failing to adequately consider driver operating states and individual driving style differences. This approach easily leads to false alarms, delayed warnings, or missed warnings in practical applications, severely impacting driver comfort and trust in the driver assistance system. Some drivers even choose to actively disable related driver assistance functions due to frequent false alarms, thereby weakening the system's role in enhancing driving safety.

[0004] On the other hand, a driver's ability to control the vehicle decreases when fatigued. If the warning system fails to react in time, the driving risk will increase significantly. Therefore, to make lane departure warning systems more intelligent and safer, it is necessary to dynamically adjust the warning sensitivity based on the driver's real-time state, especially the degree of fatigue, thereby improving the timeliness and accuracy of the system's warnings when the driver is fatigued. Summary of the Invention

[0005] To address the aforementioned deficiencies in the existing technology, the present invention aims to provide a lane departure warning method, system, and vehicle that adjusts the system sensitivity in real time according to the driver's state and driving style, making the warning more reflective of the actual vehicle deviation trend and reducing false alarms or missed alarms when dealing with different drivers.

[0006] This invention solves the above-mentioned technical problems through the following technical solution: a lane departure warning method, comprising:

[0007] Acquire vehicle lane departure distance, driver status data, driver operation data, and vehicle status data;

[0008] The driver status data, driver operation data, and vehicle status data are input into a pre-trained lane departure risk prediction model to obtain a real-time lane departure risk level.

[0009] Based on the preset mapping relationship between lane departure risk and warning sensitivity, the target warning sensitivity corresponding to the real-time lane departure risk is determined;

[0010] Based on the target warning sensitivity, the trigger threshold of the lane departure warning is dynamically adjusted; wherein, the higher the target warning sensitivity, the earlier the lane departure warning is triggered by the trigger threshold.

[0011] The obtained lane deviation distance is compared with the dynamically adjusted trigger threshold, and a lane departure warning is issued based on the comparison result.

[0012] This invention, by incorporating personalized factors such as driver status and driving style, provides customized lane departure warning services for drivers with different driving styles, including conservative and aggressive drivers. For aggressive drivers, false alarms are reduced when their condition is good; for all drivers, warnings are issued earlier when their condition is poor. This significantly reduces unnecessary interference, enhances drivers' trust in and willingness to use the driver assistance system, and avoids situations where drivers manually shut down the system due to frequent false alarms. This invention expands the decision-making basis from a single vehicle-lane geometry relationship to a multi-dimensional information fusion of driver status, driver behavior, and vehicle status. Compared to traditional methods, this multi-sensor data fusion decision-making mechanism can more comprehensively and accurately assess the overall driving risk environment, thereby making more precise and reasonable warning decisions and effectively reducing false alarms and missed alarms in complex driving scenarios.

[0013] This invention outputs a real-time lane departure risk level, i.e., the probability of a future lane departure event, through a lane departure risk prediction model. This allows the system to raise the warning level (i.e., improve warning sensitivity) in advance when the driver is in a poor state such as distraction or fatigue, but the vehicle has not yet seriously deviated from its lane. When the lane departure risk level is high, the system raises the trigger threshold to achieve an earlier warning, providing the driver with more reaction time, thereby more effectively avoiding accidents and realizing a leap from passive response to proactive prevention.

[0014] This invention uses lane departure risk as a quantitative indicator to balance safety and comfort / acceptability. In low-risk scenarios (such as when the driver is attentive and the handling is stable), the system maintains a low trigger threshold, prioritizing comfort / acceptability and avoiding unnecessary alarms. In high-risk scenarios (such as when the driver is fatigued or the handling is disordered), the system dynamically adjusts to a high trigger threshold, prioritizing safety and providing early and multiple warnings. This dynamic balancing mechanism enables the system to make optimal decisions under different operating conditions.

[0015] Furthermore, before acquiring the vehicle's lane departure distance, driver status data, driver operation data, and vehicle status data, the warning method further includes:

[0016] The vehicle's speed is obtained, and it is determined whether the vehicle's speed is greater than the preset speed. If so, the lane departure warning mechanism is activated, and the vehicle's lane departure distance, driver status data, driver operation data, and vehicle status data are obtained.

[0017] In low-speed driving situations such as traffic congestion, parking lot maneuvering, and intersection turns, it is normal and necessary for drivers to consciously and frequently cross lane lines. If the lane departure warning mechanism is activated in these situations, it will generate numerous unnecessary and frequent false alarms, severely interfering with the driver, reducing the driving experience, and potentially causing the driver to actively disable the system due to frequent false alarms, thus leaving protection unavailable when needed at high speeds. This invention effectively filters out a large number of false alarms in low-speed, non-risk scenarios by setting a preset speed threshold, ensuring that the system can work reliably and efficiently in truly dangerous medium-to-high-speed cruising conditions, thereby increasing driver acceptance and trust in the system.

[0018] Furthermore, the process of obtaining the lane line deviation distance includes:

[0019] Acquire an environmental image of the area in front of the vehicle, including lane lines;

[0020] Identify lane lines in the environmental image and mark the left and right lane lines, then convert the identified lane line pixel coordinates to a vehicle coordinate system with the vehicle center as the origin;

[0021] In the vehicle coordinate system, calculate the lateral distance between the vehicle center and the left and right lane lines, and take the smaller of the two as the final lane line deviation distance.

[0022] This invention transforms the lane line position in an image to the actual vehicle coordinate system through coordinate transformation, eliminating geometric distortion caused by camera perspective effects. This ensures that the calculated lane line deviation distance is a real physical quantity (not pixel distance), improving the accuracy and reliability of lane line deviation distance calculation. Based on the environmental image in front of the vehicle, this invention obtains the deviation distance. Compared to relying on sensors around the vehicle to perceive the deviation distance, this invention can determine the relative relationship between the vehicle's future trajectory and the lane lines earlier, achieving forward perception and prediction capabilities for early warning.

[0023] Furthermore, the training process of the lane departure risk prediction model includes:

[0024] During vehicle operation, driver status data, driver operation data, and vehicle status data are collected simultaneously.

[0025] The driver status data, driver operation data, and vehicle status data at each sampling time are preprocessed and labeled;

[0026] A sample dataset is constructed based on the preprocessed driver status data, driver operation data, and vehicle status data, along with their corresponding labels.

[0027] A deep learning model is selected, and the deep learning model is trained and validated using the sample dataset to obtain the lane departure risk prediction model.

[0028] The model training process of this invention is a supervised learning process. It teaches the deep learning model through a large number of paired samples of multimodal data (i.e., driver state data, driver operation data, and vehicle state data) and lane departure risk (high risk is labeled as 1, and low risk is labeled as 0), so that it can learn to extract key patterns that predict lane departure risk from complex and dynamic input information, and achieve accurate and personalized lane departure risk prediction.

[0029] Real-time prediction of lane departure risk based on a pre-trained lane departure risk prediction model significantly reduces the real-time computational load on the vehicle system, thereby ensuring the system's instantaneous response capability in high-speed driving scenarios.

[0030] Furthermore, the preset mapping relationship between lane departure risk level and warning sensitivity is configured such that the warning sensitivity increases monotonically with the increase of lane departure risk level.

[0031] Furthermore, the preset mapping relationship between lane departure risk and warning sensitivity is obtained through the following calibration method:

[0032] In a simulated driving environment or closed test site, test drivers with different driving styles simulate various driving states, and driver status data, driver operation data, and vehicle status data are collected under these driving states; at the same time, lane departure events are simulated.

[0033] The collected driver status data, driver operation data, and vehicle status data are input into the pre-trained lane departure risk prediction model to obtain the model prediction risk value corresponding to each sampling time.

[0034] For various simulated driving scenarios, the system provides test drivers with warning experiences of different warning sensitivities; and after each warning experience, the system records the test drivers' ratings of the current warning timing and frequency in terms of safety and comfort.

[0035] Data analysis is performed on the model's predicted risk values ​​and scores to determine the optimal warning sensitivity value for different risk ranges, thereby generating a mapping relationship between lane departure risk and warning sensitivity; wherein, the optimal warning sensitivity value refers to the highest warning sensitivity corresponding to the weighted sum of the test driver's scores on safety and comfort.

[0036] In this invention, the mapping relationship between lane departure risk and warning sensitivity is determined based on the driver's subjective feelings, so that the final determined mapping relationship between risk and sensitivity conforms to the user's psychological cognition and driving habits, fundamentally improving the user experience and acceptance of the system.

[0037] Furthermore, the dynamic adjustment formula for the trigger threshold is:

[0038] ;

[0039] in, This represents the trigger threshold dynamically adjusted based on the target warning sensitivity S; This represents the minimum trigger threshold when the warning sensitivity is 0. The adjustment coefficient is determined by the vehicle speed; the higher the vehicle speed, the larger the adjustment coefficient. S represents the target warning sensitivity.

[0040] The above dynamic adjustment formula achieves seamless dynamic adjustment of the trigger threshold, avoiding control abrupt changes caused by fixed thresholds or simple hierarchical thresholds. The dynamic adjustment formula is positively correlated with the vehicle speed through the adjustment coefficient. The higher the speed, the larger the adjustment coefficient, which significantly increases the trigger threshold, enabling early warning and multiple warnings, thus improving safety at high speeds.

[0041] Furthermore, the early warning method also includes dynamically adjusting the early warning mode based on the real-time lane departure risk level and a preset risk level threshold, specifically including:

[0042] The real-time lane departure risk level is compared with a preset risk level threshold.

[0043] When the real-time lane departure risk level is lower than the preset risk level threshold, the first warning mode is used to issue a warning.

[0044] When the real-time lane departure risk level exceeds the preset risk level threshold, a second warning mode is used to issue a warning.

[0045] The second warning mode is superior to the first warning mode in terms of warning intensity, sense of urgency, or number of warning modes used.

[0046] This invention constructs a progressive warning method from a single prompt to a strong warning, avoiding driver resentment caused by using high-intensity warnings in low-risk situations, while ensuring that drivers are effectively alerted through stronger and more redundant methods in high-risk situations. This invention accurately matches driving risks with warning resources, ensuring that the appropriate warning method is used in the appropriate scenario, and achieving the best balance between safety and user experience.

[0047] Furthermore, the first warning mode is one of the sound, visual and tactile warning modes, and the second warning mode is at least two of the sound, visual and tactile warning modes;

[0048] Alternatively, the first warning mode and the second warning mode may employ the same modality, with the second warning mode enhancing the warning intensity by increasing at least one parameter of the modality; wherein, if the modality is an audible warning, the second warning mode enhances the warning intensity by increasing the volume or frequency of the audible warning; if the modality is a visual warning, the second warning mode enhances the warning intensity by increasing the brightness of the visual warning or by making it flash; if the modality is a tactile warning, the second warning mode enhances the warning intensity by increasing the vibration frequency or amplitude of the tactile warning.

[0049] This invention improves the probability of drivers receiving warning information by introducing multimodal combined warnings, significantly reducing the risk of missed warnings and enhancing the reliability and robustness of warnings.

[0050] Based on the same concept, the present invention also provides a lane departure warning system, comprising:

[0051] The acquisition unit is used to acquire the vehicle's lane departure distance, driver status data, driver operation data, and vehicle status data.

[0052] The risk prediction unit is used to input the driver status data, driver operation data and vehicle status data into the pre-trained lane departure risk prediction model to obtain a real-time lane departure risk level.

[0053] The sensitivity determination unit is used to determine the target warning sensitivity corresponding to the real-time lane departure risk level based on the preset mapping relationship between lane departure risk level and warning sensitivity.

[0054] A threshold adjustment unit is used to dynamically adjust the trigger threshold of the lane departure warning based on the target warning sensitivity; wherein, the higher the target warning sensitivity, the earlier the lane departure warning is triggered by the trigger threshold.

[0055] The judgment unit is used to compare the acquired lane line deviation distance with the dynamically adjusted trigger threshold, and determine whether to issue a lane departure warning based on the comparison result.

[0056] Based on the same concept, the present invention also provides a vehicle equipped with the lane departure warning system described above.

[0057] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0058] The lane departure warning method and system provided by this invention can adaptively adjust the trigger threshold of lane departure warning based on driver status data, driver operation data and vehicle status data, reducing false alarms and missed alarms during use, effectively improving the driver's experience and comfort, improving the accuracy of warnings, and enhancing vehicle driving safety. Attached Figure Description

[0059] To more clearly illustrate the technical solution of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only one embodiment of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0060] Figure 1 This is a flowchart of the lane departure warning method in an embodiment of the present invention;

[0061] Figure 2 This is a diagram of the lane departure warning system architecture in an embodiment of the present invention. Detailed Implementation

[0062] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0063] The technical solution of the present invention will be described in detail below with reference to specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments.

[0064] Example 1

[0065] like Figure 1 As shown, the lane departure warning method provided in this embodiment includes the following steps:

[0066] Step S1: Obtain the vehicle speed, and when the vehicle speed is greater than the preset speed, activate the lane departure warning mechanism and proceed to step S2.

[0067] During vehicle operation, the system obtains the current vehicle speed in real time through the vehicle's CAN bus and continuously compares the vehicle speed with the preset speed (e.g., 60 km / h to 70 km / h).

[0068] If the vehicle's speed is lower than the preset speed, it is determined that the vehicle is in a low-speed driving condition. At this time, the system will suspend or not activate the lane departure warning mechanism. This means that even if the vehicle crosses the lane line, it will not trigger a lane departure warning, avoiding annoying false alarms in scenarios where the vehicle intentionally crosses lane lines, such as traffic congestion, parking lot maneuvering, or turning at intersections.

[0069] If the vehicle's speed is greater than or equal to the preset speed, the vehicle is determined to be in a high-speed driving condition. At this time, the system activates the lane departure warning mechanism and proceeds to step S2.

[0070] Step S2: Obtain the vehicle's lane departure distance, driver status data, driver operation data, and vehicle status data.

[0071] The lane departure warning method provided by this invention relies on real-time, synchronous acquisition and processing of multi-source data. Specifically, the lane departure distance is obtained through a vision-based environmental perception unit, and the specific process includes:

[0072] Step S2.1: Acquire an environmental image of the area in front of the vehicle that includes lane lines.

[0073] In this embodiment, the system continuously collects environmental images of the road ahead of the vehicle at a fixed frequency using a camera integrated inside the vehicle's windshield. These environmental images contain clear lane lines.

[0074] Step S2.2: Identify lane lines in the environmental image and mark the left and right lane lines.

[0075] The camera transmits environmental images to the onboard vision processing unit. The onboard vision processing unit runs a lane detection model, which accurately identifies lane line pixel regions from the environmental images. Lane line pixel regions located on the left side of the image are labeled as left lane lines, and lane line pixel regions located on the right side of the image are labeled as right lane lines.

[0076] Step S2.3: Convert the identified lane line pixel coordinates to the vehicle coordinate system with the vehicle center as the origin.

[0077] Because the original image suffers from perspective distortion, it cannot be directly used to measure distances in the real world. Therefore, an inverse perspective transformation technique is employed to transform the marked left and right lane lines in the image from the image pixel coordinate system to a vehicle coordinate system with the vehicle's center as the origin. In the vehicle coordinate system, the X-axis points to the right side of the vehicle, the Y-axis points directly in front of the vehicle, and the origin is located at the vehicle's centroid.

[0078] Step S2.3: In the vehicle coordinate system, calculate the lateral distance between the vehicle center and the left and right lane lines, and take the smaller of the two as the final lane line deviation distance.

[0079] In the vehicle coordinate system, calculate the lateral and vertical distances from the vehicle center to the left and right lane lines respectively, and take the smaller of the two lateral and vertical distances as the lane deviation distance. The lane deviation distance intuitively reflects the lateral distance between the vehicle center and the nearest lane line, and is the most direct geometric quantity for judging the risk of deviation.

[0080] In another specific implementation, the process of obtaining the lane deviation distance is as follows: after obtaining the lateral distance between the vehicle center and the lane line, the lateral distances between the outer edges of the left and right wheels and the corresponding lane lines are calculated based on the pre-stored vehicle width parameters. The final lane deviation distance is the smaller of these two lateral distances.

[0081] In this embodiment, driver status data is mainly obtained through a driver monitoring system (DMS). Driver status data includes, but is not limited to, driver facial expressions (e.g., eye closure, blinking frequency, whether the gaze is off the road, yawning), head posture (e.g., turning angle and nodding frequency), and actions (e.g., rubbing eyes, not looking ahead, and other fatigue or distraction actions).

[0082] In this embodiment, driver operation data is read in real time via the vehicle's Controller Area Network (CAN bus). Driver operation data includes, but is not limited to, steering wheel operation data (e.g., steering wheel torque / angular velocity) and pedal operation data (e.g., accelerator pedal opening, accelerator pedal opening change rate, brake pedal opening, and brake pedal opening change rate).

[0083] In this embodiment, vehicle status data is acquired from various onboard sensors and controllers via the CAN bus. Vehicle status data includes, but is not limited to, lateral acceleration, yaw rate, distance to and relative speed to vehicles in the preceding and adjacent lanes.

[0084] Driver status data, driver operation data, and vehicle status data provide comprehensive and accurate input for the subsequent lane departure risk prediction model. This collaborative work of multi-source heterogeneous data is the fundamental guarantee for the high accuracy and high reliability of the intelligent early warning system of this invention.

[0085] Step S3: Input the driver status data, driver operation data and vehicle status data into the pre-trained lane departure risk prediction model to obtain a real-time lane departure risk level.

[0086] In one specific embodiment of the present invention, the training process of the lane departure risk prediction model includes:

[0087] Step S3.1: During vehicle operation, driver status data, driver operation data, and vehicle status data are collected simultaneously.

[0088] In this embodiment, driver status data includes, but is not limited to, driver facial expressions (e.g., eyelid closure, blinking frequency, whether the gaze is off the road, yawning), head posture (e.g., yaw angle and nodding frequency), and actions (e.g., rubbing eyes, not looking ahead, and other signs of fatigue or distraction). Driver operation data includes, but is not limited to, steering wheel operation data (e.g., steering wheel torque / angular velocity) and pedal operation data (e.g., accelerator pedal opening, accelerator pedal opening rate of change, brake pedal opening, and brake pedal opening rate of change). Vehicle status data includes, but is not limited to, lateral acceleration, yaw rate, distance to and relative speed to vehicles in the preceding and adjacent lanes.

[0089] Step S3.2: Preprocess and label the driver status data, driver operation data, and vehicle status data at each sampling time.

[0090] In this embodiment, preprocessing includes handling missing values, noise, and outliers, and smoothing and synchronizing the data. Labeling involves assigning tags to the driver status data, driver operation data, and vehicle status data at each sampling time after preprocessing. In this embodiment, the tag is 1 (high risk) or 0 (low risk). If, within a certain period (a few seconds) after collecting the driver status data, driver operation data, and vehicle status data at that sampling time, the vehicle unintentionally deviates from its lane, then the data at that sampling time is labeled with a 1; otherwise, it is labeled with a 0.

[0091] Step S3.3: Construct a sample dataset based on the preprocessed driver status data, driver operation data, and vehicle status data and their corresponding labels.

[0092] Each sample in the sample dataset includes an input and a desired output. The input consists of driver status data, driver operation data, and vehicle status data at a single sampling time. The desired output is the label for the corresponding sampling time.

[0093] Step S3.4: Select a deep learning model, train and validate the deep learning model using the sample dataset, and obtain the lane departure risk prediction model.

[0094] In this embodiment, the deep learning model is a recurrent neural network (e.g., LSTM or GRU) or a state machine.

[0095] Step S4: Determine the target warning sensitivity corresponding to the real-time lane departure risk level based on the preset mapping relationship between lane departure risk level and warning sensitivity.

[0096] The lane departure risk prediction model outputs a real-time lane departure risk level between 0 and 1, indicating the probability of a lane departure event. A higher real-time lane departure risk level indicates a higher probability of a lane departure event and a higher warning sensitivity; that is, the warning sensitivity increases monotonically with the lane departure risk level. Therefore, the warning sensitivity can be set to a range of 0 to 1, corresponding to the lane departure risk level. For example, when the lane departure risk level is 0, the warning sensitivity is 0; when the lane departure risk level is 0.1, the warning sensitivity is 0.1; when the lane departure risk level is 0.2, the warning sensitivity is 0.2; when the lane departure risk level is 0.3, the warning sensitivity is 0.3; ...; when the lane departure risk level is 1, the warning sensitivity is 1.

[0097] In another specific implementation, the preset mapping relationship between lane departure risk and warning sensitivity can also be obtained through calibration. The specific calibration steps include:

[0098] Step S4.1: In a simulated driving environment or closed test site, test drivers with different driving styles simulate various driving states and collect driver state data, driver operation data and vehicle state data under the driving states; at the same time, lane departure events are simulated.

[0099] The driving environment is simulated using a fixed base or motion simulator equipped with a real vehicle cockpit, high-resolution projection system, force feedback steering wheel, and vehicle dynamics model. Alternatively, in a closed test track, existing standard lane lines are laid out or utilized, and real vehicles equipped with a complete data acquisition system are used for testing.

[0100] Recruit a sufficient number of test drivers and, based on their self-reports and pre-screened driving assessments, categorize them into two driving styles: conservative and aggressive.

[0101] Driving states include normal alertness, fatigue, and distraction. For example, distraction can be simulated by having the driver perform other tasks (answering the phone, operating other devices in the vehicle), and fatigue can be induced by driving for extended periods.

[0102] In simulated driving environments or closed test tracks, vehicle deviation events can be induced through programming or route design. For example, when the driver is fatigued or distracted, the vehicle can be slowly deviated from its lane line on curves or straight sections.

[0103] Step S4.2: Input the collected driver status data, driver operation data and vehicle status data into the pre-trained lane departure risk prediction model to obtain the model prediction risk value corresponding to each sampling time.

[0104] Step S4.3: Provide test drivers with warning experiences of different warning sensitivities for various simulated driving scenarios; and after each warning experience, record the test drivers' ratings of the current warning timing and frequency in terms of safety and comfort.

[0105] In each type of driving scenario, the lane departure warning mechanism will operate based on a pre-set warning sensitivity and record the test driver's rating of the current warning timing (i.e., when the warning is issued) and frequency (i.e., the number of warnings) in terms of safety and comfort. In the same type of driving scenario, each test driver will experience the warning effect under different pre-set warning sensitivities in turn and record the rating.

[0106] Step S4.4: Perform data analysis on the model's predicted risk level and score, determine the optimal warning sensitivity value for different risk level intervals, and then generate the mapping relationship between lane departure risk level and warning sensitivity.

[0107] Based on the model's predicted risk level and scores, an optimal warning sensitivity value is determined for each risk level interval. This optimal warning sensitivity value refers to the highest warning sensitivity corresponding to the weighted sum of the test drivers' scores on safety and comfort.

[0108] For example, for a risk level of R, three warning sensitivities were tested. At warning sensitivity A1, the mean safety score of all test drivers was S1, and the mean comfort score was C1. At warning sensitivity A2, the mean safety score of all test drivers was S2, and the mean comfort score was C2. At warning sensitivity A3, the mean safety score of all test drivers was S3, and the mean comfort score was C3. The weighted sum of the mean safety score and the mean comfort score was calculated for each warning sensitivity. The warning sensitivity corresponding to the highest weighted sum was taken as the optimal warning sensitivity value, thus obtaining the risk level R and its corresponding optimal warning sensitivity value.

[0109] Step S5: Based on the target warning sensitivity, dynamically adjust the trigger threshold of the lane departure warning.

[0110] The higher the target warning sensitivity, the earlier the trigger threshold is adjusted to trigger the warning. In this embodiment, the dynamic adjustment formula for the trigger threshold is:

[0111] (1)

[0112] in, This represents the trigger threshold dynamically adjusted based on the target warning sensitivity S; This represents the minimum trigger threshold when the warning sensitivity is 0. The adjustment coefficient is determined by the vehicle speed; the higher the vehicle speed, the larger the adjustment coefficient. S represents the target warning sensitivity. The adjustment coefficient can also be set based on experience.

[0113] Minimum trigger threshold when lane deviation distance is equal to the lateral distance from the vehicle center to the lane line. This is the sum of the safe lateral distance (e.g., 0.1m) between the outer edge of the wheel and the corresponding lane line, and the vertical distance from the vehicle center to the outer edge of the wheel. Since the warning threshold may be set inside or outside the lane line, this embodiment defines the lane line deviation distance as the lateral distance from the vehicle center to the lane line, making data processing simpler and more intuitive.

[0114] Step S6: Compare the obtained lane deviation distance with the dynamically adjusted trigger threshold, and determine whether to issue a lane departure warning based on the comparison result.

[0115] When the lane deviation distance is less than or equal to the dynamically adjusted trigger threshold, it indicates that lane departure will occur, and a lane departure warning will be issued.

[0116] In one specific embodiment of the present invention, while issuing a lane departure warning, the warning method further includes dynamically adjusting the warning mode based on the real-time lane departure risk level and a preset risk level threshold, specifically including:

[0117] Compare the real-time lane departure risk level with the preset risk level threshold;

[0118] When the real-time lane departure risk level is lower than the preset risk level threshold, the first warning mode is used to issue a warning.

[0119] When the real-time lane departure risk level exceeds the preset risk level threshold, the second warning mode is used to issue a warning.

[0120] The second early warning mode is higher than the first early warning mode in terms of warning intensity, sense of urgency, or number of warning modes used.

[0121] In one specific implementation, the first warning mode is one of the following: sound (e.g., an audible alarm), visual, and tactile warnings (e.g., a vibrator installed on the steering wheel or seat); the second warning mode is at least two of the following: sound, visual, and tactile warnings. For example, the first warning mode is an audible warning, and the second warning mode is both an audible and a visual warning.

[0122] In another specific implementation, the first warning mode and the second warning mode adopt the same mode, and the second warning mode enhances the warning intensity by increasing at least one parameter of the mode.

[0123] Specifically, if the first warning mode is an audible warning, then the second warning mode is also an audible warning, and the warning intensity is enhanced by increasing the volume or frequency of the audible warning; if the first warning mode is a visual warning, then the second warning mode is also a visual warning, and the warning intensity is enhanced by increasing the brightness of the visual warning or making it flash; if the first warning mode is a tactile warning, then the second warning mode is also a tactile warning, and the warning intensity is enhanced by increasing the vibration frequency or amplitude of the tactile warning.

[0124] Compared to traditional fixed-sensitivity schemes, this invention dynamically adjusts the trigger threshold based on the target warning sensitivity, making the lane departure warning function more intelligent. It dynamically adjusts the warning timing according to the driver's state, operation, and vehicle status, effectively reducing false alarms and missed alarms, and improving the user experience. Compared to traditional single-prompt schemes, this invention dynamically adjusts the warning method based on real-time lane departure risk and a preset risk threshold. Furthermore, when the real-time lane departure risk exceeds the preset risk threshold, a higher-level warning mode is employed, enhancing driving safety.

[0125] Example 2

[0126] like Figure 2 As shown, the lane departure warning system provided in this embodiment includes an acquisition unit, a risk prediction unit, a sensitivity determination unit, a threshold adjustment unit, and a judgment unit.

[0127] The acquisition unit is used to acquire the vehicle's lane departure distance, driver status data, driver operation data, and vehicle status data.

[0128] The risk prediction unit is used to input the driver status data, driver operation data and vehicle status data acquired by the acquisition unit into the pre-trained lane departure risk prediction model to obtain a real-time lane departure risk level.

[0129] The sensitivity determination unit is used to determine the target warning sensitivity corresponding to the real-time lane departure risk output by the prediction model, based on the preset mapping relationship between lane departure risk and warning sensitivity.

[0130] The threshold adjustment unit is used to dynamically adjust the trigger threshold of the lane departure warning based on the target warning sensitivity determined by the sensitivity determination unit; wherein, the higher the target warning sensitivity, the earlier the lane departure warning is triggered.

[0131] The judgment unit is used to compare the acquired lane line deviation distance with the dynamically adjusted trigger threshold, and determine whether to issue a lane departure warning based on the comparison result.

[0132] In one specific implementation, the early warning system further includes a sensing unit, a driver monitoring unit, and an early warning unit.

[0133] The perception unit is used to acquire environmental images containing lane lines in front of the vehicle; identify lane lines in the environmental images and mark the left and right lane lines; transform the pixel coordinates of the identified lane lines to a vehicle coordinate system with the vehicle center as the origin; in the vehicle coordinate system, calculate the lateral distance between the vehicle center and the left and right lane lines, and take the smaller of the two as the final lane line deviation distance. Specifically, the perception unit includes a camera and an onboard vision processing unit.

[0134] The driver monitoring unit is used to monitor driver status data. Driver operation data and vehicle status data can be obtained through existing vehicle controllers.

[0135] The warning unit includes audible warnings, visual warnings, and / or tactile warnings.

[0136] In some specific embodiments of the present invention, the warning system may incorporate the features of the lane departure warning method in Embodiment 1 of the present invention, and vice versa, which will not be repeated here.

[0137] Example 3

[0138] This embodiment also provides an electronic device, which includes: a memory, a processor, and a computer program or instructions stored in the memory. The processor executes the computer program or instructions to implement the lane departure warning method in Embodiment 1 of the present invention.

[0139] Although not shown, the electronic device includes a processor that can perform various appropriate operations and processes based on programs and / or data stored in read-only memory (ROM) or loaded from a storage portion into random access memory (RAM). The processor can be a multi-core processor or may contain multiple processors. In some embodiments, the processor may include a general-purpose main processor and one or more specialized coprocessors, such as a central processing unit, graphics processing unit (GPU), neural network processor (NPU), digital signal processor (DSP), etc. Various programs and data required for device operation are also stored in RAM. The processor, ROM, and RAM are interconnected via a bus. Input / output (I / O) interfaces are also connected to the bus.

[0140] The processor and memory described above are used together to execute programs / instructions stored in the memory. When the program / instructions are executed by the computer, they can implement the methods, steps, or functions described in the above embodiments.

[0141] Although not shown, embodiments of the present invention also provide a computer-readable storage medium having a computer program or instructions stored thereon, which, when executed by a processor, implements the lane departure warning method of Embodiment 1 of the present invention.

[0142] Readable storage media include both permanent and non-permanent, removable and non-removable media that can store information by any method or technology. Information can be computer-readable instructions, data structures, program modules, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0143] The above description only discloses specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or modifications that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention.

Claims

1. A lane departure warning method, characterized in that, The early warning method includes: Acquire vehicle lane departure distance, driver status data, driver operation data, and vehicle status data; The driver status data, driver operation data, and vehicle status data are input into a pre-trained lane departure risk prediction model to obtain a real-time lane departure risk level. Based on the preset mapping relationship between lane departure risk and warning sensitivity, the target warning sensitivity corresponding to the real-time lane departure risk is determined; Based on the target warning sensitivity, the trigger threshold of the lane departure warning is dynamically adjusted; wherein, the higher the target warning sensitivity, the earlier the lane departure warning is triggered by the trigger threshold. The obtained lane deviation distance is compared with the dynamically adjusted trigger threshold, and a lane departure warning is issued based on the comparison result.

2. The lane departure warning method according to claim 1, characterized in that, Before acquiring the vehicle's lane departure distance, driver status data, driver operation data, and vehicle status data, the warning method further includes: The vehicle's speed is obtained, and it is determined whether the vehicle's speed is greater than the preset speed. If so, the lane departure warning mechanism is activated, and the vehicle's lane departure distance, driver status data, driver operation data, and vehicle status data are obtained.

3. The lane departure warning method according to claim 1, characterized in that, The process of obtaining the lane line deviation distance includes: Acquire an environmental image of the area in front of the vehicle, including lane lines; Identify lane lines in the environmental image and mark the left and right lane lines, then convert the identified lane line pixel coordinates to a vehicle coordinate system with the vehicle center as the origin; In the vehicle coordinate system, calculate the lateral distance between the vehicle center and the left and right lane lines, and take the smaller of the two as the final lane line deviation distance.

4. The lane departure warning method according to claim 1, characterized in that, The training process of the lane departure risk prediction model includes: During vehicle operation, driver status data, driver operation data, and vehicle status data are collected simultaneously. The driver status data, driver operation data, and vehicle status data at each sampling time are preprocessed and labeled; A sample dataset is constructed based on the preprocessed driver status data, driver operation data, and vehicle status data, along with their corresponding labels. A deep learning model is selected, and the deep learning model is trained and validated using the sample dataset to obtain the lane departure risk prediction model.

5. The lane departure warning method according to claim 1, characterized in that, The preset mapping relationship between lane departure risk and warning sensitivity is configured such that the warning sensitivity increases monotonically with the increase of lane departure risk. Alternatively, the preset mapping relationship between lane departure risk and warning sensitivity can be obtained through the following calibration method: In a simulated driving environment or closed test site, test drivers with different driving styles simulate various driving states, and driver state data, driver operation data, and vehicle state data are collected under these driving states; at the same time, lane departure events are simulated. The collected driver status data, driver operation data, and vehicle status data are input into the pre-trained lane departure risk prediction model to obtain the model prediction risk value corresponding to each sampling time. For various simulated driving scenarios, it provides test drivers with a warning experience with different warning sensitivities; After each warning experience, the test driver's rating of the current warning timing and frequency in terms of safety and comfort was recorded; Data analysis is performed on the model's predicted risk values ​​and scores to determine the optimal warning sensitivity value for different risk ranges, thereby generating a mapping relationship between lane departure risk and warning sensitivity; wherein, the optimal warning sensitivity value refers to the highest warning sensitivity corresponding to the weighted sum of the test driver's scores on safety and comfort.

6. The lane departure warning method according to claim 1, characterized in that, The dynamic adjustment formula for the trigger threshold is: ； in, This represents the trigger threshold dynamically adjusted based on the target warning sensitivity S; This represents the minimum trigger threshold when the warning sensitivity is 0. The adjustment coefficient is determined by the vehicle speed; the higher the vehicle speed, the larger the adjustment coefficient. S represents the target warning sensitivity.

7. The lane departure warning method according to any one of claims 1 to 6, characterized in that, The early warning method further includes dynamically adjusting the early warning mode based on the real-time lane departure risk level and a preset risk level threshold, specifically including: The real-time lane departure risk level is compared with a preset risk level threshold. When the real-time lane departure risk level is lower than the preset risk level threshold, the first warning mode is used to issue a warning. When the real-time lane departure risk level exceeds the preset risk level threshold, a second warning mode is used to issue a warning. The second warning mode is superior to the first warning mode in terms of warning intensity, sense of urgency, or number of warning modes used.

8. The lane departure warning method according to claim 7, characterized in that, The first warning mode is one of the sound, visual and tactile warning modes, and the second warning mode is at least two of the sound, visual and tactile warning modes. Alternatively, the first warning mode and the second warning mode may employ the same modality, with the second warning mode enhancing the warning intensity by increasing at least one parameter of the modality; wherein, if the modality is an audible warning, the second warning mode enhances the warning intensity by increasing the volume or frequency of the audible warning; if the modality is a visual warning, the second warning mode enhances the warning intensity by increasing the brightness of the visual warning or by making it flash; if the modality is a tactile warning, the second warning mode enhances the warning intensity by increasing the vibration frequency or amplitude of the tactile warning.

9. A lane departure warning system, characterized in that, The early warning system includes: The acquisition unit is used to acquire the vehicle's lane departure distance, driver status data, driver operation data, and vehicle status data. The risk prediction unit is used to input the driver status data, driver operation data and vehicle status data into the pre-trained lane departure risk prediction model to obtain a real-time lane departure risk level. The sensitivity determination unit is used to determine the target warning sensitivity corresponding to the real-time lane departure risk level based on the preset mapping relationship between lane departure risk level and warning sensitivity. A threshold adjustment unit is used to dynamically adjust the trigger threshold of the lane departure warning based on the target warning sensitivity; wherein, the higher the target warning sensitivity, the earlier the lane departure warning is triggered by the trigger threshold. The judgment unit is used to compare the acquired lane line deviation distance with the dynamically adjusted trigger threshold, and determine whether to issue a lane departure warning based on the comparison result.

10. A vehicle, characterized in that, The vehicle is equipped with the lane departure warning system as described in claim 9.

Images