Method and system for suppressing false triggering of vehicle AEB in hail environment

Abstract

This invention discloses a method and system for suppressing false AEB (Autonomous Emergency Braking) triggering in hail environments. The method includes: simultaneously collecting multi-source environmental data output from multiple types of vehicle-mounted sensors; extracting multi-dimensional interference features based on the multi-source environmental data, weighting the calculations to obtain the hail environment probability to determine the hail condition, and simultaneously quantifying the hail intensity level; conducting five-layer joint verification of forward-detected targets to distinguish between hail-induced false targets and real obstacles; dynamically correcting the AEB collision time trigger threshold based on hail intensity, and quantifying the collision risk by integrating multiple driving parameters; switching control modes according to the target determination results, enabling a dedicated suppression strategy with simultaneous multi-parameter adjustment for hail-interfering targets, and executing standard AEB braking logic for high-risk real obstacles; and establishing a braking closed-loop correction mechanism, which, once a momentary false braking caused by hail occurs, smoothly releases pressure and compensates for power, correcting the vehicle's driving posture; thus significantly reducing the probability of false AEB triggering in hail environments.

Description

Technical Field

[0001] This invention belongs to the field of active safety assisted driving technology, and more specifically, relates to a method and system for suppressing the false triggering of AEB (Autonomous Emergency Braking) in hail environments. Background Technology

[0002] Automatic Emergency Braking (AEB), as a core active safety feature in vehicles, has been upgraded from the recommended national standard GB / T39901 to the mandatory national standard GB39901, becoming a mandatory feature for passenger cars. Current AEB systems rely on onboard forward-facing cameras, millimeter-wave radar, lidar, and other perception sensors to collect real-time information about the environment ahead of the vehicle. Through collision risk assessment algorithms, they identify real obstacles such as motor vehicles, pedestrians, and non-motorized vehicles, and actively output braking commands when the driver fails to take timely evasive action, thereby reducing the probability of a collision and mitigating the severity of collision damage.

[0003] Under normal sunny, rainy, and foggy weather conditions, traditional AEB perception fusion algorithms can reliably identify real road obstacles, and the braking control logic is reliable. However, in areas prone to hail, special weather conditions can easily generate dense hail particles, causing multi-dimensional and strong interference to the vehicle's perception system: on the one hand, high-speed falling hail particles obscure the camera's imaging field of view, causing a surge in image noise density, loss of target contour features, and blurred and distorted images; on the other hand, millimeter-wave radar receives a large amount of hail reflection clutter, forming a large number of discrete false point cloud targets, significantly reducing the sensor's signal-to-noise ratio and significantly decreasing the stability of target detection. Affected by the superposition of the above two interferences, traditional AEB is very likely to misidentify hail clusters as real obstacles ahead, triggering emergency braking unnecessarily when there is no collision risk.

[0004] Current industry technologies for AEB (Autonomous Emergency Braking) optimization in severe weather only address scenarios like regular rain and heavy fog. The mainstream approach involves simply adjusting sensor fusion weights and uniformly reducing the basic AEB trigger sensitivity. There is no integrated control scheme for targeted identification, target discrimination, and dynamic suppression specifically designed for extreme conditions like hail. Existing technologies have several inherent limitations: The hail-free environment identification mechanism cannot quantify the hail intensity level and only uses a fixed AEB trigger threshold, which cannot adapt to different hail interference intensities. The system lacks multi-dimensional verification logic to distinguish between false hail targets and real obstacles, and relies solely on a single frame image or single radar information to determine the target, resulting in a high probability of misjudgment. The braking control strategy is statically fixed. Once a target is detected, a fixed deceleration braking is uniformly executed. After a hailstorm is mistakenly triggered, the braking cannot be quickly released and power cannot be smoothly restored. The vehicle experiences obvious jerking and the driving smoothness is greatly reduced. Sudden, unannounced braking during hailstorms can significantly increase the risk of secondary traffic accidents such as rear-end collisions. Frequent and prolonged mis-braking can reduce drivers' trust in the AEB function, and may even lead to drivers actively disabling the AEB system, completely losing the ability to provide active safety protection.

[0005] Meanwhile, most existing optimization solutions rely on costly LiDAR to improve perception and anti-interference capabilities, resulting in high hardware modification costs and making it difficult to widely deploy in mass-produced L2-level assisted driving vehicles. In summary, existing AEB control strategies cannot effectively solve the technical challenge of high-frequency false triggering of the system in hail environments. There is an urgent need for a suppression control solution that is compatible with conventional onboard perception hardware, can accurately identify hail conditions, and adaptively adjusts the AEB braking logic. Summary of the Invention

[0006] To address the aforementioned problems in existing technologies, this invention provides a method and system for suppressing false triggering of AEB (Autonomous Emergency Braking) in hail environments. It synchronously collects multi-source environmental data using standard vehicle hardware such as cameras, millimeter-wave radar, and piezoelectric impact sensors. A weighted hail recognition model is constructed, integrating image noise, radar clutter, hail impact frequency, and temperature change characteristics to calculate the probability of hail environments and quantify hail intensity levels. A five-layer cross-validation logic, incorporating trajectory, multi-frame timing, speed, size, and multiple sensors, is designed to identify false hail targets. Based on hail intensity, the AEB's TTC (Total Traffic Response) trigger threshold, target confirmation frame count, and braking deceleration are dynamically and adaptively adjusted. The hail interference target adopts a graded suppression mode with multiple parameters synchronously adjusted, and adds a closed-loop recovery control logic for false braking with gradient pressure relief and power linear compensation. This invention only reuses the vehicle's existing perception hardware without adding a new LiDAR. It can accurately distinguish hail interference from real road obstacles, significantly reduce the frequency of AEB false triggering in hail weather, and adapt to light to heavy hail environments. It avoids vehicle instability and chain rear-end collisions caused by sudden braking. At the same time, the entire lightweight algorithm can be directly installed in L2-level mass-produced vehicles, taking into account the anti-interference capability in extreme weather and the complete braking protection performance of AEB under normal conditions. It has both control robustness and value for large-scale mass production applications.

[0007] To achieve the above objectives, one aspect of the present invention provides a decision-making control method for suppressing false triggering of AEB (Autonomous Emergency Braking) in hail environments, comprising the following steps: S1: Simultaneously collects multi-source environmental data through vehicle-mounted forward-looking camera, millimeter-wave radar, piezoelectric hail impact sensor, rain sensor, wheel speed and vehicle speed sensor, and ambient temperature sensor. S2. Construct a multi-feature fusion hail recognition model based on the multi-source environmental data, extract multi-dimensional interference features, and calculate the hail environment probability through weighted fusion; when the hail environment probability exceeds a preset judgment threshold, determine that the vehicle is in a severe hail driving condition, and calculate the hail intensity level based on the weighted fusion of multi-dimensional interference features. S3. For the target detected in front by the sensor in real time, execute five layers of verification logic in sequence: trajectory stability analysis, multi-frame target continuity verification, target velocity vector matching, target size change rate analysis, and multi-sensor cross-verification; output the judgment result of whether the target is a false hail target or a real obstacle. S4. Based on the hail intensity level, combined with the original factory TTC threshold for no severe weather and the preset adjustment coefficient, calculate the dynamic threshold and correct the AEB collision time trigger threshold in real time; at the same time, combine the vehicle's real-time speed, following distance and target TTC value to comprehensively quantify the collision risk level. S5. If the multi-layer verification determines that the target ahead is a false target due to hail interference, the hail-specific suppression mode is activated, and the dynamic AEB collision time trigger threshold calculated in step S4 is used as the braking determination basis under the current hail conditions. When the hail intensity level increases, the system automatically lowers the TTC trigger threshold, allowing AEB to be triggered only when the collision hazard is truly close, thereby improving the system's tolerance to hail clutter; S6. Set up a closed-loop correction logic for the vehicle's braking system. If the system identifies a false braking behavior caused by momentary hail interference, it will immediately release the vehicle's braking pressure, smoothly restore the vehicle's power output, and correct the vehicle's driving posture.

[0008] Furthermore, the multidimensional interference features mentioned in step S2 include image noise density, radar clutter density, hail impact frequency, and ambient temperature change rate.

[0009] Furthermore, the formula for calculating the probability of hail in step S2 is as follows: ,in, The preset weighting coefficients are obtained by calibrating each interference feature on a real vehicle. , The total number of environmental features participating in the integration; These are the environmental characteristic parameters involved in the fusion.

[0010] Furthermore, step S2 also includes: The calculated hail environment probability Compared with the system's preset judgment threshold Perform numerical comparison: like The vehicle is determined to be in severe hail conditions, and the hail intensity level calculation process continues. like If the current condition is determined to be normal sunny, light rain or fog, the hail intensity calculation is skipped directly, the vehicle controller switches to the original standard AEB control logic, and the subsequent hail suppression process is not started.

[0011] Furthermore, in step S2, the hail intensity level The value range is 0 to 5; It was a light hailstorm. Moderate hail. It was a severe hailstorm.

[0012] Further, the trajectory stability analysis unit verification in step S3 is as follows: extract the longitudinal and lateral displacement coordinates of the target for 5 consecutive frames and calculate the displacement variance; if the displacement variance is greater than a preset threshold, it means that the target's motion trajectory is random and is determined to be hail interference. The multi-frame continuity verification: count the number of frames the target continues to exist. If the target's lifespan is ≤3 frames, it is directly determined to be instantaneous hail particle interference. The target velocity vector matching verification involves extracting the longitudinal and lateral velocities of the target relative to the vehicle. If the velocity vector does not exhibit a continuous and stable change pattern, it is determined to be a false hail target. The target size change rate analysis: extract the horizontal and vertical dimensions of the target in multiple consecutive frames. If the size fluctuation exceeds the threshold of 30%, it is determined to be a hail particle cluster. The multi-sensor cross-verification unit verifies: synchronously retrieves the target detection results of the camera and radar. If the two types of sensors cannot detect the target in the same spatial location synchronously for three consecutive frames, it is directly determined to be hail interference. If a target meets at least two or more of the following interference characteristics: short lifespan, random trajectory, drastic size fluctuation, and no stable velocity vector, then the target is determined to be a hail-induced false target. If all five layers of verification pass, the target has a stable trajectory, can be detected synchronously across multiple consecutive frames, and exhibits smooth changes in speed and size, then it is determined to be a real obstacle.

[0013] Furthermore, the dynamic threshold calculation formula in step S4 is as follows: , The preset adjustment coefficient, The original TTC threshold is based on the absence of severe weather conditions.

[0014] Furthermore, step S4 combines real-time vehicle speed, following distance, and target TTC value to comprehensively quantify the collision risk level, including: The system reads the longitudinal relative distance between the vehicle and the target ahead from the radar in real time, and combines the real-time longitudinal speed of the vehicle and the relative speed of the target collected by the wheel speed sensor to calculate the current real collision time (TTC) in real time. In the linkage step S3, the obstacle determination result distinguishes the risk basis: if it is a false target of hail, it is directly determined that there is no effective collision risk, and only the suppression control parameters are output; if it is a real road obstacle, it is combined with dynamic thresholds. Conduct risk classification; Based on real-time TTC and dynamic thresholds The numerical difference is used to classify the collision risk into three levels: Low risk: Real-time TTC There is no immediate risk of collision; maintain normal following control. Medium risk: Real-time TTC There is a potential collision risk; activate brake warning. High risk: Real-time TTC The risk of collision is extremely high; prepare to issue a graded emergency braking command.

[0015] Furthermore, tightening the AEB trigger window in step S5 means replacing the basic TTC threshold of 1.8s under normal hail-free conditions with the value calculated using the formula. The calculated dynamic threshold; In step S5, extending the number of valid target confirmation frames means raising the threshold of 3 frames for valid determination under normal working conditions to 8 frames. Only when the same target is detected stably for 8 consecutive frames is it identified as a valid target with collision risk, thus filtering out short-term false point clouds that flash by momentarily during hail. In step S5, limiting the instantaneous maximum braking deceleration means locking the maximum braking deceleration of the entire vehicle to 0.25g. Even if the system briefly identifies a suspected risky target, it will not output high-intensity emergency braking. The standard AEB emergency braking control logic in step S5 includes: canceling all suppression parameters, restoring the original basic TTC threshold of 1.8s and 3-frame target confirmation threshold, and removing the 0.25g braking deceleration limit; When TTC approaches When the trigger threshold is reached, the vehicle controller outputs an audible and visual collision warning to alert the driver; when That is, when the TTC falls into the first threshold range, a light braking force is applied to pre-decelerate the vehicle; when When the risk of collision is extremely high, the vehicle controller releases the 0.25g braking deceleration limit and outputs the maximum braking deceleration allowed by the vehicle system to brake at full speed, thereby minimizing the collision speed.

[0016] A second aspect of the present invention provides a vehicle AEB false triggering suppression decision control system for hail environments, used to implement the aforementioned vehicle AEB false triggering suppression decision control method for hail environments, comprising: The environmental information acquisition module is used to drive the vehicle-mounted forward-looking monocular camera, millimeter-wave radar, piezoelectric hail impact sensor, rain sensor, wheel speed sensor, and ambient temperature sensor to simultaneously acquire raw environmental data from multiple sources. It also performs time-series alignment and outlier removal preprocessing on the raw data to output a standardized multi-source environmental dataset. The hail environment recognition module receives the standardized dataset output by the environmental information acquisition module, extracts multi-dimensional interference features such as image noise density, radar clutter density, hail impact frequency, and ambient temperature change rate, calculates the hail environment probability by weighted fusion to complete the hail condition determination, and outputs the hail intensity level by weighted fusion calculation. The obstacle authenticity determination module is used to receive target data from the camera and radar, and sequentially complete five layers of verification: trajectory stability analysis, multi-frame target continuity verification, target velocity vector matching, target size change rate analysis, and multi-sensor cross-verification. The output is the determination result of whether the target is a false hail target or a real obstacle. The dynamic risk assessment module is used to calculate a dynamic threshold based on the hail intensity level, the original factory TTC threshold without severe weather, and a preset adjustment coefficient, and to correct the AEB collision time trigger threshold in real time; at the same time, it combines the vehicle's real-time speed, following distance, and target TTC value to comprehensively quantify the collision risk level. The AEB false trigger suppression control module is used to simultaneously execute a combined control strategy of dynamically reducing the TTC trigger threshold, extending the number of valid target confirmation frames, and limiting the instantaneous maximum braking deceleration when identifying false hail targets, based on the threshold parameters output by the dynamic risk assessment module and the target determination results. When identifying real dangerous obstacles and the collision risk exceeds the dynamic AEB trigger threshold, it switches to output standard AEB emergency braking control commands. The brake recovery control module monitors the vehicle's braking execution status, identifies erroneous braking behavior caused by momentary hail interference, releases the vehicle's braking pressure in stages, and simultaneously outputs power compensation commands to smoothly restore the vehicle's power output and correct the vehicle's driving posture.

[0017] In summary, compared with the prior art, the above-described technical solutions conceived by this invention can achieve the following beneficial effects: (1) The method of the present invention can significantly reduce the probability of false AEB triggering under extreme hail conditions, and solve the core defect of false braking from the source. The existing technology has not built a hail-specific recognition system, and cannot distinguish hail clutter from real obstacles. Fixed trigger thresholds are prone to causing high-frequency and unwarranted sudden braking. The present invention establishes a weighted fusion hail recognition model based on multi-dimensional features of image, radar, impact, and temperature, which can accurately determine hail conditions and quantify interference intensity. At the same time, it filters false hail targets through a five-layer cross-verification mechanism of trajectory, time sequence, speed, size, and multiple sensors, and achieves effective differentiation between interference and obstacles. The actual measured AEB false braking rate under hail conditions has been greatly reduced, and the misjudgment problem caused by hail particles has been eliminated from the bottom layer of the algorithm.

[0018] (2) Traditional severe weather AEB schemes only use uniform and fixed trigger parameters, which cannot be adapted to different hail scenarios of light, moderate and heavy hail. They are prone to insufficient suppression and still brake falsely, or excessive suppression and loss of braking protection capability. This invention is based on the dynamic correction of the TTC trigger threshold, target confirmation frame number and maximum braking deceleration control parameters according to the hail intensity level. The stronger the hail interference, the stronger the suppression force is. When the interference weakens, the standard AEB logic is gradually restored. It can realize adaptive control under hail conditions of all intensity, taking into account both anti-interference capability and basic active safety functions.

[0019] (3) Existing solutions lack a smooth recovery mechanism after accidental braking. Instantaneous pressure relief and sudden power changes can cause vehicle pitching and jerking, resulting in a poor driving experience. This invention adds a braking-power closed-loop correction logic. After identifying accidental braking, it adopts an integrated control method of gradient pressure relief, linear torque synchronous compensation, and vehicle posture linkage correction to smoothly eliminate braking force and restore driving power. This avoids severe vehicle body fluctuations caused by sudden braking and acceleration, and significantly improves the smoothness of vehicle driving under severe road conditions such as hail.

[0020] (4) In hail weather, sudden braking without warning by AEB can easily cause following vehicles to react too late, leading to chain collisions and reducing road safety. This invention reduces ineffective emergency braking actions at the source by accurately filtering false hail targets that pose no collision risk, eliminating the danger of sudden deceleration during road driving; at the same time, closed-loop recovery control can weaken the magnitude of braking abrupt changes, leaving sufficient space for following vehicles to avoid collisions, significantly reducing the probability of secondary collisions and improving overall traffic safety in mixed traffic scenarios.

[0021] (5) Existing solutions for improving the accuracy of perception in adverse weather generally rely on expensive new hardware such as lidar, resulting in high vehicle modification costs and making it difficult to popularize them in ordinary L2-level vehicles. This invention fully reuses the vehicle's standard front-view camera, millimeter-wave radar, and conventional temperature and vibration sensing hardware, without the need for additional sensing equipment; the entire set of recognition and suppression algorithms is lightweight and has high real-time computing performance, and can be directly installed on the existing vehicle domain controller without modifying the original vehicle perception hardware architecture. It is compatible with all mass-produced assisted driving models and has extremely strong industrialization and promotion value. Attached Figure Description

[0022] Figure 1 This is a flowchart illustrating a vehicle AEB false trigger suppression decision control method under hail conditions, according to an embodiment of the present invention. Figure 2 This is a schematic diagram of the AEB false triggering suppression control logic in step S5 of a vehicle AEB false triggering suppression decision control method under hail conditions according to an embodiment of the present invention; Figure 3 This is a schematic diagram of the dynamic change curve of hail environmental impact factors in an embodiment of the present invention; Figure 4This is a schematic diagram of a vehicle AEB false trigger suppression decision control system under hail conditions, according to an embodiment of the present invention. Detailed Implementation

[0023] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention. Furthermore, the technical features involved in the various embodiments of this invention described below can be combined with each other as long as they do not conflict with each other.

[0024] like Figure 1 As shown, one aspect of the present invention provides a vehicle AEB false trigger suppression decision control method under hail conditions. The hardware includes a forward-looking monocular camera, a 77GHz millimeter-wave radar, a piezoelectric hail impact sensor, a rain sensor, wheel speed and vehicle speed sensors, and an ambient temperature sensor. The vehicle controller incorporates the control algorithm of the present invention, which includes the following steps: S1: Simultaneously collects multi-source environmental data through vehicle-mounted forward-looking camera, millimeter-wave radar, piezoelectric hail impact sensor, rain sensor, wheel speed and vehicle speed sensor, and ambient temperature sensor. S2. Based on the multi-source environmental data, construct a multi-feature fusion hail recognition model, extract multi-dimensional interference features such as image noise density, radar clutter density, hail impact frequency, and ambient temperature change rate, and calculate the hail environment probability through weighted fusion; when the hail environment probability exceeds a preset judgment threshold, determine that the vehicle is in a severe hail driving condition, and calculate the hail intensity level based on the weighted fusion of multi-dimensional interference features. S3. For the target detected in front by the sensor in real time, execute five layers of verification logic in sequence: trajectory stability analysis, multi-frame target continuity verification, target velocity vector matching, target size change rate analysis, and multi-sensor cross-verification; output the judgment result of whether the target is a false hail target or a real obstacle. S4. Based on the hail intensity level, combined with the original factory TTC threshold for no severe weather and the preset adjustment coefficient, calculate the dynamic threshold and correct the AEB collision time trigger threshold in real time; at the same time, combine the vehicle's real-time speed, following distance and target TTC value to comprehensively quantify the collision risk level. S5. If the multi-layer verification determines that the target ahead is a false target due to hail interference, the hail-specific suppression mode is activated, and a combined control strategy of tightening the AEB trigger window, extending the number of effective target confirmation frames, and limiting the instantaneous maximum braking deceleration is executed simultaneously. If multiple consecutive frames of verification identify a stable real obstacle, and the collision risk exceeds the dynamic AEB trigger threshold, the hail-specific suppression mode is immediately exited while ensuring the robustness of the original system, and the standard AEB emergency braking control logic is executed to ensure vehicle safety under real dangerous conditions. S6. Set up a closed-loop correction logic for the vehicle's braking system. If the system identifies a false braking behavior caused by momentary hail interference, it will immediately release the vehicle's braking pressure, smoothly restore the vehicle's power output, correct the vehicle's driving posture, and avoid continuous jerking, driving instability, and secondary accidents such as rear-end collisions.

[0025] The steps of the method of the present invention will be described in detail below.

[0026] (1) Multi-source environmental synchronous acquisition The vehicle controller drives all sensing hardware to collect data synchronously with a unified fixed sampling period. The multi-source data mentioned in step S1 includes at least the features of the road ahead image, radar echo point cloud features, hail impact frequency, rainfall intensity information, real-time vehicle speed, and real-time ambient temperature. In a specific embodiment of the present invention, the sampling period is set to 50ms to ensure that the acquisition timing of the six sensors is completely aligned, avoiding subsequent identification and judgment deviations caused by time misalignment of multi-source data; the six hardware channels synchronously acquire the corresponding original sensing data, and after the acquisition is completed, the vehicle controller uniformly performs time alignment and instantaneous jump outlier removal preprocessing to output a standardized multi-source environmental dataset, providing a regular input for subsequent hail environment recognition and obstacle authenticity verification. The acquisition content of each sensor is described below: Forward-facing camera: Real-time acquisition of 120° wide-angle road images in front of the vehicle, outputting image grayscale matrix, global pixel noise distribution information, foreground target contour features, and raw image clarity data; 77GHz millimeter-wave radar: Scans the entire space in front of the vehicle from 0 to 150m, and outputs raw echo point cloud dataset, clutter statistics per unit period, longitudinal / lateral velocity vectors of each detected target relative to the vehicle, and target circumscribed size parameters; Piezoelectric hail impact sensor: Multiple piezoelectric hail impact sensors are arranged in the windshield and hood area of ​​the vehicle to continuously collect the vibration amplitude and frequency generated by hail impacts, count the number of effective hail impacts per unit time, and output hail impact frequency time series data. Rain sensor: Collects signals of liquid water droplet concentration and rainfall intensity in the air to distinguish between liquid rain interference and solid hail particle interference, thereby helping to improve the accuracy of hail environment identification; Wheel speed sensor: synchronously collects the wheel speed pulse signals of the four wheels, and obtains the real-time longitudinal driving speed of the vehicle by filtering and solving the signals through the vehicle controller; Ambient temperature sensor: Real-time acquisition of continuous ambient temperature data outside the vehicle front, outputting time-series temperature values ​​to calculate the rate of decrease of ambient temperature per unit time. After the data collection is completed, the vehicle controller removes outlier values ​​with sudden jumps in a single frame from the raw data output by the six sensors, converts them to the same data dimension and numerical range, completes timestamp alignment and binding, and forms a set of standardized multi-source environmental datasets that are synchronously matched and free from time-series deviations. This dataset is then synchronously distributed to the subsequent hail environment recognition module and obstacle authenticity determination module. (2) Steps for Hail Environment Identification and Intensity Classification Based on the standardized multi-source environmental data collected in step S1, a hail environment identification model adapted to extreme hail conditions is constructed. This model extracts core interference features from the standardized multi-source environmental data: image noise density, radar clutter density, hail impact frequency, and ambient temperature change rate. The model then performs hail probability calculation, hail condition determination, and hail intensity classification step by step. Specifically, this includes: First, the feature parameters are standardized: Image noise density: The total number of pixels with abrupt gray-level changes in a single frame of the image is counted and divided by the total number of pixels in the image to obtain the image noise interference coefficient in the range of 0 to 1. Radar clutter density: The clutter density per unit area is obtained by dividing the total number of discrete clutter points with no stable motion pattern in a single sampling period by the area of ​​the radar's effective detection area. Hail impact frequency: Count the number of effective hail impact vibrations captured by the piezoelectric sensor within a 50ms sampling period, and output the impact frequency per unit time; Ambient temperature change rate: Extract 10 consecutive frames of time-series temperature data, calculate the difference in ambient temperature decrease per unit time, and obtain the temperature change rate characteristics; Each environmental characteristic parameter is normalized and mapped to the 0-1 interval to obtain dimensionless characteristic parameters. ; Secondly, a weighted fusion calculation is used to determine the probability of a hail environment: Preset weight coefficients for each interference feature , The weights were obtained through extensive hail-affected vehicle calibration and substituted into the formula. Calculate the probability of hail. ; Preset weighting coefficients for each interference feature obtained through actual vehicle calibration; The environmental feature parameters participating in the fusion; the above-mentioned interference feature parameters are all normalized before being substituted into the formula and mapped to the [0,1] interval; the vehicle controller completes the weighted summation operation and outputs the hail environment probability in the 0~1 interval; In a specific embodiment of the present invention, the environmental feature parameters involved in the fusion This includes: radar clutter density, image grayscale fluctuation, target flashing frequency, hail impact acoustic signature, and temperature drop characteristics; The calculated hail environment probability Compared with the system's preset judgment threshold Perform numerical comparison: like The vehicle is determined to be in severe hail conditions, and the hail intensity level calculation process continues. like If the current condition is determined to be normal sunny, light rain, or light fog, the hail intensity calculation is skipped, and the vehicle controller switches to the original standard AEB control logic, without initiating the subsequent hail suppression process; this embodiment presets a hail determination threshold. ; Finally, the weighted fusion calculation is used to determine the hail intensity level: When a vehicle is determined to be in hail conditions, the hail intensity level is calculated by weighted fusion of all the aforementioned standardized interference features. In a specific embodiment of the present invention, for a scenario determined to be in hail conditions, the hail intensity level is calculated by weighted fusion of all interference features. The calculation formula is: in, Hail intensity level; The clutter density of millimeter-wave radar is equal to the ratio of the number of radar clutter particles per unit time to the area of ​​the radar detection zone. Image noise density is equal to the ratio of the number of abnormal noise pixels to the total number of pixels. This represents the frequency of hail impacts. The rate of change of ambient temperature; The weighting coefficients for millimeter-wave radar clutter density; These are the weighting coefficients for the image noise density; The weighting coefficients for hail impact frequencies; The weighting coefficients are used to calculate the rate of change of ambient temperature. Each environmental characteristic is normalized and mapped to the interval of 0 to 1 to obtain dimensionless characteristic parameters.

[0027] Hail intensity level The value ranges from 0 to 5, with higher values ​​indicating stronger hail interference; the calculation process simultaneously distinguishes between light, moderate, and severe hail. It was a light hailstorm. Moderate hail. It was a severe hailstorm; After the calculation is completed, the vehicle controller will assign the hail intensity level. The dynamic collision risk assessment process in step S4 is synchronously transmitted as the core input parameter for dynamic correction of the AEB threshold. (3) Multi-level determination of the authenticity of obstacles In step S3, the vehicle controller receives the forward target detection list synchronously output by the forward-looking monocular camera and the 77GHz millimeter-wave radar in real time. For each independently detected target, it sequentially executes five layers of verification logic: trajectory stability analysis, multi-frame target continuity verification, target velocity vector matching, target size change rate analysis, and multi-sensor cross-verification. Based on the comprehensive results of the multi-layer verification, it distinguishes between hail-interferenced false targets and real road obstacles. The detailed execution process is as follows: First layer, trajectory stability analysis and verification: Extract the horizontal and vertical spatial coordinates of the current target for 5 consecutive frames and calculate the target displacement variance; a trajectory fluctuation judgment threshold is preset. If the displacement variance is greater than the threshold, it means that the target has no fixed route and the movement trajectory is random. The target is directly marked as having hail interference characteristics; if the displacement variance is less than or equal to the threshold, the target trajectory is judged to be stable and enters the next layer of verification. The second layer, multi-frame target continuity verification: This involves counting the total number of frames the target has existed from its first detection to the current frame, with a preset lifespan threshold of 3 frames. If the target's duration is ≤3 frames, it indicates the target only appears momentarily, and is identified as short-term hail particle interference, thus marking the interference characteristic. If the target's duration is >3 frames, it means the target can be continuously and stably captured by the sensor, and the verification proceeds to the next layer. The third layer, target velocity vector matching verification: extract the longitudinal and lateral relative velocities of the target relative to the vehicle in multiple consecutive frames, and fit the velocity change curve; if the velocity value does not have a continuous and stable change pattern and frequently jumps sharply, it is marked as hail interference feature; if the target velocity is continuous and stable and changes gently, proceed to the next layer of verification. Fourth layer, target size change rate analysis and verification: extract the outer horizontal width and vertical height of the target in multiple consecutive frames, and calculate the target size fluctuation range between adjacent frames; the preset size fluctuation threshold is 30%. If the size fluctuation range exceeds 30%, it means that the target is a discrete hail cluster with a size that fluctuates, and the interference feature is marked; if the size fluctuation range is ≤30%, the target outline is stable, and the final layer of verification is entered. Fifth layer, multi-sensor cross-validation: Simultaneously retrieve the target detection results of the camera and millimeter-wave radar in the same spatial area, and compare the spatial coordinate range of the target detected by the two types of sensors; if the camera and radar cannot detect the target synchronously in the same coordinate area within 3 consecutive frames, it is determined to be hail clutter interference generated by a single sensor, and the interference feature is marked; if the two sensors can detect the same spatial target synchronously for multiple consecutive frames, there is no interference mark. If the target meets at least two or more of the following interference characteristics when it comes to the five layers of verification results: short lifespan, random motion trajectory, drastic size fluctuation, and no stable velocity vector, then the target is determined to be a hail-induced false target. If no interference characteristics are detected in all five layers of verification, and the target trajectory, duration, speed, size, and dual-sensor matching are all stable and meet the standards, then the target is determined to be a real road obstacle such as a vehicle, pedestrian, or non-motorized vehicle. The vehicle controller synchronously transmits the target authenticity determination result to the dynamic collision risk assessment step S4 as the basis for judging the switching of AEB suppression strategy.

[0028] (4) Dynamic collision risk assessment In step S4, the vehicle controller receives the hail intensity level output in step S2. The obstacle authenticity determination result output in step S3 synchronously retrieves the real-time vehicle speed, radar-calculated following distance between the vehicle and the target vehicle, and real-time target collision time (TTC) collected in step S1, and performs calculations in two sub-processes: dynamic threshold calculation and collision risk quantification; specifically including: Dynamically adjust the AEB collision time trigger threshold based on hail intensity: retrieve the vehicle's original factory-calibrated TTC threshold under severe weather conditions. Adjustment coefficient with preset threshold Substitute into the dynamic threshold calculation formula The calculation is completed to obtain the dynamic AEB triggering threshold adapted to the current hail disturbance intensity. Among them, hail intensity level The larger the value, the higher the calculated dynamic trigger threshold. The smaller the value, the tighter the braking trigger window of the system, reducing the probability of misjudging braking in icy and snowy conditions; the vehicle's factory calibration does not include a benchmark TTC threshold for severe weather. It must not be less than the minimum TTC security threshold allowed by the system. This value is provided by each OEM; In a specific embodiment of the present invention, actual measurement Original factory basic threshold adjustment coefficient ; Calculate the dynamic TTC threshold: This means that under hail conditions, the AEB trigger TTC threshold is lowered from the usual 1.8s to 0.88s, reducing the braking trigger time window and avoiding the interference of small hailstones to trigger braking; The calculated dynamic TTC threshold The cached data is stored in the vehicle controller and used as the AEB braking judgment benchmark under the current hail conditions, replacing the conventional fixed basic threshold. Multi-dimensional parameters comprehensively quantify the collision risk level: Real-time reading of the longitudinal relative distance between the vehicle and the target in front, combined with the real-time longitudinal speed of the vehicle and the relative speed of the target collected by the wheel speed sensor, to calculate the current real collision time (TTC) in real time; In the linkage step S3, the obstacle determination result distinguishes the risk basis: if it is a false target of hail, it is directly determined that there is no effective collision risk, and only the suppression control parameters are output; if it is a real road obstacle, it is combined with dynamic thresholds. Conduct risk classification; Based on real-time TTC and dynamic thresholds The numerical difference is used to classify the collision risk into three levels: Low risk: Real-time TTC There is no immediate risk of collision; maintain normal following control. Medium risk: Real-time TTC There is a potential collision risk; activate brake warning. High risk: Real-time TTC The risk of collision is extremely high; prepare to issue a graded emergency braking command. The calculated dynamic AEB trigger threshold The comprehensive and quantified collision risk level and obstacle authenticity identification are simultaneously sent to the AEB false triggering grade suppression control process in step S5, serving as the core judgment basis for mode switching and braking output.

[0029] (5) Graded suppression control of AEB false triggering Step S5 includes AEB mode switching and graded suppression control sub-steps; the vehicle controller receives the target authenticity determination result output from step S3, and the dynamic AEB trigger threshold and collision risk level output from step S4, referring to... Figure 2 The control logic is executed differently for two target scenarios: hail false target suppression and real obstacle braking. The specific execution flow is as follows: Scenario 1: The target is determined to be a false target due to hail interference; activate the hail suppression mode. When the five-layer joint verification determines that the current detection target is a false target due to hail interference, the vehicle controller immediately activates the hail-specific suppression mode and simultaneously corrects the three sets of AEB core control parameters to form a combined anti-interference strategy: Tighten the AEB trigger window: Replace the base TTC threshold of 1.8s under normal hail-free conditions with the formula The calculated dynamic threshold; taking severe hail H=4.6 in this embodiment as an example, the dynamic TTC threshold is calculated to be 0.88s, that is, the dynamic TTC trigger threshold is adjusted from 1.8s to 0.88s. The higher the hail intensity, the smaller the dynamic TTC trigger threshold. AEB braking is only allowed to be activated when the collision time between the vehicle and the real obstacle is close to a dangerous state, thereby improving the system's tolerance to false targets such as hail clutter, reducing false braking caused by instantaneous clutter, and ensuring that AEB emergency braking can still be activated in time when a real collision risk occurs; Extend the number of valid target confirmation frames: Increase the threshold of 3 frames for valid determination under normal working conditions to 8 frames. Only when the same target is detected stably for 8 consecutive frames is it identified as a valid target with collision risk, thus filtering out short-term false point clouds that flash by momentarily during hail. Limiting instantaneous maximum braking deceleration: The maximum braking deceleration limit of the entire vehicle is locked at 0.25g. Even if the system briefly identifies a suspected risky target, it will not output high-intensity emergency braking to avoid severe vehicle jerking. After the parameters are updated, the vehicle controller continues to monitor the target ahead, does not output any graded braking commands, and only maintains the normal power driving logic; Scenario 2: If the target is determined to be a stable, real-world road obstacle, exit the suppression mode and execute standard AEB logic. If the five-layer verification results for multiple consecutive frames determine that the target is a real vehicle, pedestrian, or non-motorized vehicle, and the real-time collision TTC is less than the dynamic AEB trigger threshold and the collision risk reaches a high-risk level, the vehicle will perform a mode switching operation: under the premise of ensuring the robustness of the original system, it will immediately exit the hail-specific suppression mode and cancel all suppression parameters: restore the original basic TTC threshold of 1.8s, the 3-frame target confirmation threshold, and remove the 0.25g braking deceleration limit; Activate the standard AEB graded braking control logic and output control commands sequentially according to the collision risk gradient: Early warning phase: When TTC approaches When the trigger threshold is reached, the vehicle controller outputs an audible and visual collision warning to alert the driver; Mild pre-braking phase: when When the TTC falls into the first threshold range, it means that the collision risk has initially emerged but there is still sufficient room for avoidance, and a light braking force is applied to pre-decelerate the vehicle. Full emergency braking phase: when When the risk of collision is extremely high, the vehicle controller releases the 0.25g braking deceleration limit and outputs the maximum braking deceleration allowed by the vehicle system to brake at full speed, thereby minimizing the collision speed and avoiding or mitigating impact damage.

[0030] The vehicle controller cyclically reads the target verification results and dynamic thresholds with a 50ms sampling period, and switches between suppression mode and standard braking mode in real time; hail reduction, As the value decreases, the dynamic TTC threshold also decreases synchronously, and the suppression effect gradually weakens, achieving a smooth and abrupt switch in the control strategy.

[0031] The current mode identifier, the corrected TTC threshold, the target confirmation frame count, and the braking deceleration limit are synchronously transmitted to the vehicle brake controller and power controller to complete the issuance of commands to the underlying actuators.

[0032] In a specific embodiment of the present invention, the system determined that the target was hail interference, performed three parameter adjustments, and did not output any braking commands throughout the process. The vehicle drove smoothly without any sudden braking.

[0033] (6) Closed-loop recovery control for accidental braking In step S6, the vehicle controller monitors the brake pressure signal, longitudinal acceleration, and yaw rate output by the brake controller in real time, and establishes a brake-power closed-loop correction logic. For short-term false braking caused by hail noise, it performs segmented pressure relief, power compensation, and attitude correction throughout the entire process. The detailed execution process is as follows: Instantaneous false braking behavior identification and judgment: The vehicle controller continuously collects the braking execution status and target verification mark, and synchronously compares the target authenticity result output in step S3: If the system has determined that the current target is a false hail target, but the brake controller still outputs effective braking pressure and the longitudinal deceleration of the vehicle body exceeds the preset stability threshold, it is determined that a false braking caused by instantaneous hail interference has occurred, and the closed-loop recovery correction process is immediately started. Gradual and gradual release of vehicle braking pressure: Abandoning the control method of instantaneous complete pressure relief, a multi-level gradient pressure relief strategy is adopted; specifically, the master cylinder brake oil pressure is gradually reduced according to the preset pressure reduction step size, and the wheel braking force is slowly released in multiple stages to avoid the sudden disappearance of braking pressure, which would cause the vehicle to lurch forward or nose-dive violently. The pressure relief duration and pressure reduction of each stage are adaptively matched according to the current hail intensity level H. The higher the hail intensity, the more pressure relief stages there are and the smoother the pressure relief process is. Synchronous and smooth compensation of vehicle power output: During the synchronous period of graded release of braking pressure, the vehicle controller sends a linear torque compensation command to the power controller; specifically, the compensation torque is calculated and matched based on the current vehicle speed and remaining braking pressure, and the output power of the motor / engine is smoothly increased to offset the driving drag caused by the residual braking force. The amount of torque compensation decreases synchronously and linearly with the braking pressure, and there will be no sudden increase in power that causes the vehicle to lurch. Real-time vehicle driving posture correction: During the recovery process, the vehicle's yaw rate and longitudinal acceleration signals are continuously collected to judge the vehicle driving posture in real time. If the vehicle is detected to be veering or pitching, the difference in braking force between the left and right wheels is adjusted simultaneously and the power output distribution is slightly corrected to bring the vehicle's longitudinal and lateral motion parameters back to the stable range and suppress the tendency of the vehicle to become unstable during the braking process. Closed-loop recovery completion judgment and logic exit: Continuously monitor two indicators: braking pressure and vehicle acceleration. When the vehicle braking pressure returns to zero and the vehicle longitudinal acceleration is stable in the normal smooth driving range for a preset duration, the erroneous braking closed-loop correction process is judged to be complete. The vehicle controller stops torque compensation output, exits the closed-loop correction logic, switches to the normal constant speed driving control state, continuously monitors the braking signal, and waits for the next round of control command input. The complete closed-loop logic of segmented pressure relief + linear power compensation + attitude correction in step S6 of this invention can completely eliminate the phenomenon of continuous vehicle body jerking and sudden acceleration after rapid deceleration caused by instantaneous mis-braking during hail, and greatly reduce the risk of rear-end collisions caused by vehicles failing to avoid the impact in time.

[0034] The vehicle braking closed-loop correction logic is set up so that if the system experiences a brief erroneous braking action due to instantaneous strong hail noise interference, the braking recovery control module immediately executes the smooth recovery process: Rapidly reduce the master cylinder braking pressure and release the braking force in stages to avoid the vehicle from lurching forward due to the sudden and complete release of braking pressure. Simultaneously send a power compensation signal to the vehicle power controller to smoothly increase the output torque of the engine / motor and compensate for braking resistance; The system continuously monitors the vehicle's longitudinal acceleration and yaw rate. Once the vehicle's driving posture returns to stability, the power compensation is terminated, and normal power output is maintained.

[0035] Figure 3 This is a dynamic change curve of hail environmental impact factors, used to visually demonstrate the correlation between hail intensity, radar clutter density, and AEB false trigger probability over time during a continuous hailstorm.

[0036] The time-series changes of the curve are as follows: Between 0 and 120 seconds, the hail weather gradually intensified, and the hail intensity continued to rise. A large number of high-speed hail particles hit the vehicle's sensing area, causing a synchronous increase in millimeter-wave radar echo clutter, and the radar clutter density rose synchronously with the hail intensity. Affected by the continuously intensifying sensing interference, the stability of sensor target recognition continued to deteriorate, and the probability of false triggering of the AEB system increased rapidly. Around 120 seconds, the hail intensity reaches its peak, and the radar clutter density and the probability of AEB false triggering reach their maximum values ​​simultaneously. During this period, the vehicle perception system is most severely affected by hail. Traditional AEB is very likely to misjudge discrete hail particles as physical obstacles in front, thus generating invalid emergency braking actions. Between 120 and 200 seconds, the hail weather gradually weakened, the hail intensity continued to decrease, the amount of radar clutter decreased accordingly, the sensor's perception and recognition capabilities gradually recovered, and the probability of AEB false triggering decreased significantly and eventually returned to the normal low level range. The following technical conclusions can be drawn from this graph: Hail intensity and radar clutter density show a significant positive correlation; The higher the radar clutter density, the greater the probability of false triggering of the AEB system; Hail can cause continuous and strong interference to vehicle cameras and millimeter-wave radar, significantly reducing the stable recognition capability of the perception system. Traditional AEB control schemes that use fixed trigger thresholds are prone to frequent false braking under severe hail conditions; The integrated control mechanism for dynamic threshold adjustment and stratified false triggering suppression based on hail intensity grading proposed in this invention is both necessary and theoretically reasonable.

[0037] With the help of Figure 3 The temporal linkage and change pattern of the three shown can fully verify the complete technical route of the present invention: hail environment recognition → obstacle authenticity determination → dynamic collision risk assessment → AEB graded suppression control has a reliable theoretical basis and engineering feasibility, and can effectively solve the technical defects of the existing fixed threshold AEB in the frequent false triggering under extreme hail conditions.

[0038] like Figure 4 As shown, a second aspect of the present invention provides a vehicle AEB false triggering suppression decision control system for hail environments, used to implement the above-mentioned control method, comprising: The environmental information acquisition module is used to drive the vehicle-mounted forward-looking monocular camera, millimeter-wave radar, piezoelectric hail impact sensor, rain sensor, wheel speed sensor, and ambient temperature sensor to simultaneously acquire raw environmental data from multiple sources. It also performs time-series alignment and outlier removal preprocessing on the raw data to output a standardized multi-source environmental dataset. The hail environment recognition module receives the standardized dataset output by the environmental information acquisition module, extracts multi-dimensional interference features such as image noise density, radar clutter density, hail impact frequency, and ambient temperature change rate, calculates the hail environment probability by weighted fusion to complete the hail condition determination, and outputs the hail intensity level by weighted fusion calculation. The obstacle authenticity determination module is used to receive target data from the camera and radar, and sequentially complete five layers of verification: trajectory stability analysis, multi-frame target continuity verification, target velocity vector matching, target size change rate analysis, and multi-sensor cross-verification. The output is the determination result of whether the target is a false hail target or a real obstacle. The dynamic risk assessment module is used to calculate a dynamic threshold based on the hail intensity level, the original factory TTC threshold without severe weather, and a preset adjustment coefficient, and to correct the AEB collision time trigger threshold in real time; at the same time, it combines the vehicle's real-time speed, following distance, and target TTC value to comprehensively quantify the collision risk level. The AEB false trigger suppression control module is used to simultaneously execute a combined control strategy of dynamically reducing the TTC trigger threshold, extending the number of valid target confirmation frames, and limiting the instantaneous maximum braking deceleration when identifying false hail targets, based on the threshold parameters output by the dynamic risk assessment module and the target determination results. When identifying real dangerous obstacles and the collision risk exceeds the dynamic AEB trigger threshold, it switches to output standard AEB emergency braking control commands. The brake recovery control module monitors the vehicle's braking status, identifies erroneous braking behavior caused by momentary hail interference, releases the vehicle's braking pressure in stages, and simultaneously outputs power compensation commands to smoothly restore the vehicle's power output, correct the vehicle's driving posture, and avoid vehicle jerking and secondary accidents such as rear-end collisions.

[0039] It should be noted that the vehicle AEB false trigger suppression decision control system provided in this embodiment can be a computer program (including program code) running on a computer device. For example, the vehicle AEB false trigger suppression decision control system in hail environment is an application software. The vehicle AEB false trigger suppression decision control system in hail environment can be used to execute the corresponding steps in the above-mentioned method provided in the embodiment of this application.

[0040] Specific implementation examples: A full-vehicle road verification test was conducted under severe hail conditions. The test vehicle was equipped with a forward-looking monocular camera, a 77GHz millimeter-wave radar, a piezoelectric hail impact sensor, a rain sensor, wheel speed and vehicle speed sensors, and an ambient temperature sensor. The entire control algorithm of this invention was pre-embedded in the vehicle controller. The complete test process is as follows: The vehicle drove normally under continuous hail weather conditions. The vehicle controller drove the environmental information acquisition module to synchronously collect multi-source sensing data at fixed intervals. Among them, the forward-looking camera output the pixel features of the image in real time, the millimeter-wave radar continuously collected the echo point cloud of the entire front area, the ambient temperature sensor continuously recorded the temporal changes of atmospheric temperature, the piezoelectric sensor counted the frequency of hail impacts, the wheel speed sensor output the real-time vehicle speed, and the rain sensor distinguished between liquid rainfall and solid hail interference. The standardized dataset collected was input into the hail environment recognition model to complete the comprehensive working condition judgment.

[0041] I. Hail Environment Identification and Intensity Calculation Process: The multi-source interference characteristics observed in this test scenario were as follows: image noise density increased by 45% compared to normal clear conditions, the total amount of millimeter-wave radar clutter increased by 60%, the ambient temperature dropped by 8°C in a short period, and the frequency of hail impacts collected by the piezoelectric sensor increased significantly. Substituting these multi-dimensional characteristics into the weighted fusion formula, the probability of a hail environment was calculated to be 0.83. The system's built-in hail detection threshold is 0.75. When the probability of a hail environment exceeds the preset threshold, the vehicle controller determines that the vehicle is currently in a severe hail driving condition and simultaneously calculates the corresponding hail intensity level.

[0042] II. Five-layer authenticity verification process for the target ahead: The sensor detected a suspected target ahead in real time. The vehicle controller initiated a five-layer joint verification process, including trajectory stability, multi-frame continuity, velocity vector, size change, and cross-sensor verification. The verification results showed that the target's lifespan was only two frames, its trajectory was irregular, its size fluctuated drastically between adjacent frames, it lacked a continuous and stable relative velocity vector, and the forward-looking camera and millimeter-wave radar could not synchronize and continuously detect targets in the same spatial location. Based on all interference characteristics identified in the multi-layer verification, the system determined that the detected target was a false interference target formed by hail particles and posed no real collision risk.

[0043] III. Implementation of Dynamic Threshold Correction and Hierarchical Suppression Control: The dynamic risk assessment module retrieves the current hail intensity, adaptively adjusts the AEB control parameters based on a formula, and simultaneously activates a hail-specific suppression mode. The original factory TTC trigger threshold of 1.8s under no severe operating conditions was dynamically adjusted to 0.88s; The number of frames for effective target confirmation has been increased from the usual 3 frames to 8 frames, and only targets that are detected stably for multiple consecutive frames are determined to be effective risk sources. The instantaneous maximum braking deceleration of the entire vehicle is limited to 0.25g to mitigate sudden braking impact.

[0044] After the entire set of suppression strategies took effect, the vehicle controller did not output any emergency braking commands throughout the process, the vehicle maintained stable driving, and there was no unwarranted braking caused by hail.

[0045] IV. Verification through comparison with real obstacles: In the same test section, a stationary vehicle was artificially placed as a real obstacle. After the sensor detected the target, the five-layer verification results showed that the target had a long duration, a stable trajectory, and stable size and speed characteristics. Furthermore, the camera and radar could continuously identify it synchronously. The system automatically exited the hail suppression mode and restored the original standard AEB control parameters. When the collision time reached the trigger condition, the vehicle controller normally output a graded braking command, and the AEB active braking function was fully activated, effectively avoiding the risk of a real collision. Case test conclusion: This invention's integrated control scheme, which combines multi-feature hail recognition, multi-layer target discrimination, and dynamic threshold adaptive adjustment, can accurately distinguish between hail-induced false interference targets and real road obstacles. It completely avoids the problem of false AEB triggering in extreme hail environments, while not weakening the AEB active braking protection capability in real dangerous scenarios, verifying the reliability and engineering practical value of this solution.

[0046] A third aspect of the invention also provides a computer-readable storage medium storing a computer program that is executed by a processor to implement... Figure 1 The methods provided in each step are detailed in the implementation methods provided in the above steps, and will not be repeated here.

[0047] Those skilled in the art will readily understand that the above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A decision-making and control method for suppressing false triggering of AEB (Autonomous Emergency Braking) in hail environments, characterized in that, Includes the following steps: S1: Simultaneously collects multi-source environmental data through vehicle-mounted forward-looking camera, millimeter-wave radar, piezoelectric hail impact sensor, rain sensor, wheel speed and vehicle speed sensor, and ambient temperature sensor. S2. Construct a multi-feature fusion hail recognition model based on the multi-source environmental data, extract multi-dimensional interference features, and calculate the hail environment probability through weighted fusion; when the hail environment probability exceeds a preset judgment threshold, determine that the vehicle is in a severe hail driving condition, and calculate the hail intensity level based on the weighted fusion of multi-dimensional interference features. S3. For the target ahead detected by the sensor in real time, execute five layers of verification logic in sequence: trajectory stability analysis, multi-frame target continuity verification, target velocity vector matching, target size change rate analysis, and multi-sensor cross-verification. Output the determination result of whether the target is a false hailstone target or a real obstacle; S4. Based on the hail intensity level, combined with the original factory TTC threshold for no severe weather and the preset adjustment coefficient, calculate the dynamic threshold and correct the AEB collision time trigger threshold in real time; at the same time, combine the vehicle's real-time speed, following distance and target TTC value to comprehensively quantify the collision risk level. S5. If the multi-layer verification determines that the target ahead is a false target due to hail interference, then the hail-specific suppression mode is activated, and a combined control strategy of tightening the AEB trigger window, extending the number of effective target confirmation frames, and limiting the instantaneous maximum braking deceleration is executed simultaneously. If a stable, real obstacle is detected in multiple consecutive frames and the collision risk exceeds the dynamic AEB trigger threshold, the hail-specific suppression mode will be immediately exited and the standard AEB emergency braking control logic will be executed. S6. Set up a closed-loop correction logic for the vehicle's braking system. If the system identifies a false braking behavior caused by momentary hail interference, it will immediately release the vehicle's braking pressure, smoothly restore the vehicle's power output, and correct the vehicle's driving posture.

2. The vehicle AEB false triggering suppression decision control method under hail conditions according to claim 1, characterized in that: The multidimensional interference features mentioned in step S2 include image noise density, radar clutter density, hail impact frequency, and ambient temperature change rate.

3. The vehicle AEB false triggering suppression decision control method under hail conditions according to claim 2, characterized in that: The formula for calculating the probability of hail in step S2 is as follows: ,in, The preset weighting coefficients are obtained by calibrating each interference feature on a real vehicle. , The total number of environmental features participating in the integration; These are the environmental characteristic parameters involved in the fusion.

4. The vehicle AEB false triggering suppression decision control method under hail conditions according to claim 3, characterized in that: Step S2 also includes: The calculated hail environment probability Compared with the system's preset judgment threshold Perform numerical comparison: like The vehicle is determined to be in severe hail conditions, and the hail intensity level calculation process continues. like If the current condition is determined to be normal sunny, light rain or fog, the hail intensity calculation is skipped directly, the vehicle controller switches to the original standard AEB control logic, and the subsequent hail suppression process is not started.

5. The vehicle AEB false triggering suppression decision control method under hail conditions according to claim 4, characterized in that: In step S2, the hail intensity level The value range is 0 to 5; It was a light hailstorm. Moderate hail. It was a severe hailstorm.

6. A vehicle AEB false triggering suppression decision control method under hail conditions according to any one of claims 1-5, characterized in that: The trajectory stability analysis unit verification in step S3 is as follows: extract the longitudinal and lateral displacement coordinates of the target for 5 consecutive frames and calculate the displacement variance; if the displacement variance is greater than a preset threshold, it means that the target's motion trajectory is random and is determined to be hail interference. The multi-frame continuity verification: count the number of frames the target continues to exist. If the target's lifespan is ≤3 frames, it is directly determined to be instantaneous hail particle interference. The target velocity vector matching verification involves extracting the longitudinal and lateral velocities of the target relative to the vehicle. If the velocity vector does not exhibit a continuous and stable change pattern, it is determined to be a false hail target. The target size change rate analysis: extract the horizontal and vertical dimensions of the target in multiple consecutive frames. If the size fluctuation exceeds the threshold of 30%, it is determined to be a hail particle cluster. The multi-sensor cross-verification unit verifies: synchronously retrieves the target detection results of the camera and radar. If the two types of sensors cannot detect the target in the same spatial location synchronously for three consecutive frames, it is directly determined to be hail interference. If a target meets any one or more of the following characteristics: short lifespan, random trajectory, drastic size fluctuation, and no stable velocity vector, then the target is determined to be a false target caused by hail interference. If all five layers of verification pass, the target has a stable trajectory, can be detected synchronously across multiple consecutive frames, and exhibits smooth changes in speed and size, then it is determined to be a real obstacle.

7. A vehicle AEB false triggering suppression decision control method under hail conditions according to any one of claims 1-5, characterized in that: The formula for calculating the dynamic threshold in step S4 is as follows: , The preset adjustment coefficient, The original TTC threshold is based on the absence of severe weather conditions.

8. The vehicle AEB false triggering suppression decision control method under hail conditions according to claim 7, characterized in that: Step S4 combines real-time vehicle speed, following distance, and target TTC value to comprehensively quantify the collision risk level, including: The system reads the longitudinal relative distance between the vehicle and the target ahead from the radar in real time, and combines the real-time longitudinal speed of the vehicle and the relative speed of the target collected by the wheel speed sensor to calculate the current real collision time (TTC) in real time. In the linkage step S3, the obstacle determination result distinguishes the risk basis: if it is a false target of hail, it is directly determined that there is no effective collision risk, and only the suppression control parameters are output; if it is a real road obstacle, it is combined with dynamic thresholds. Conduct risk classification; Based on real-time TTC and dynamic thresholds The numerical difference is used to classify the collision risk into three levels: Low risk: Real-time TTC There is no immediate risk of collision; maintain normal following control. Medium risk: Real-time TTC There is a potential collision risk; activate brake warning. High risk: Real-time TTC The risk of collision is extremely high; prepare to issue a graded emergency braking command.

9. The vehicle AEB false triggering suppression decision control method under hail conditions according to claim 8, characterized in that, Step S5, tightening the AEB trigger window, means replacing the base TTC threshold of 1.8s under normal hail-free conditions with the value calculated using the formula. The calculated dynamic threshold; In step S5, extending the number of valid target confirmation frames means raising the threshold of 3 frames for valid determination under normal working conditions to 8 frames. Only when the same target is detected stably for 8 consecutive frames is it identified as a valid target with collision risk, thus filtering out short-term false point clouds that flash by momentarily during hail. In step S5, limiting the instantaneous maximum braking deceleration means locking the maximum braking deceleration of the entire vehicle to 0.25g. Even if the system briefly identifies a suspected risky target, it will not output high-intensity emergency braking. The standard AEB emergency braking control logic in step S5 includes: canceling all suppression parameters, exiting the hail false target suppression mode, restoring the standard AEB control strategy, and using the dynamic AEB trigger threshold calculated in step S4 as the current braking criterion, the 3-frame target confirmation threshold, and removing the 0.25g braking deceleration limit. When TTC approaches When the trigger threshold is reached, the vehicle controller outputs an audible and visual collision warning to alert the driver; when That is, when the TTC falls into the first threshold range, a light braking force is applied to pre-decelerate the vehicle; when When the collision risk is extremely high, the vehicle controller releases the 0.25g braking deceleration limit and outputs the maximum braking deceleration allowed by the vehicle system to brake at full speed, thereby minimizing the collision speed.

10. A decision-making control system for suppressing false triggering of AEB (Autonomous Emergency Braking) in hail environments, characterized in that, The method for suppressing false triggering of vehicle AEB in hail conditions as described in any one of claims 1-9 includes: The environmental information acquisition module is used to drive the vehicle-mounted forward-looking monocular camera, millimeter-wave radar, piezoelectric hail impact sensor, rain sensor, wheel speed sensor, and ambient temperature sensor to simultaneously acquire raw environmental data from multiple sources. It also performs time-series alignment and outlier removal preprocessing on the raw data to output a standardized multi-source environmental dataset. The hail environment recognition module receives the standardized dataset output by the environmental information acquisition module, extracts multi-dimensional interference features such as image noise density, radar clutter density, hail impact frequency, and ambient temperature change rate, calculates the hail environment probability by weighted fusion to complete the hail condition determination, and outputs the hail intensity level by weighted fusion calculation. The obstacle authenticity determination module is used to receive target data from the camera and radar, and sequentially complete five layers of verification: trajectory stability analysis, multi-frame target continuity verification, target velocity vector matching, target size change rate analysis, and multi-sensor cross-verification. The output is the determination result of whether the target is a false hail target or a real obstacle. The dynamic risk assessment module is used to calculate a dynamic threshold based on the hail intensity level, the original factory TTC threshold without severe weather, and a preset adjustment coefficient, and to correct the AEB collision time trigger threshold in real time; at the same time, it combines the vehicle's real-time speed, following distance, and target TTC value to comprehensively quantify the collision risk level. The AEB false trigger suppression control module is used to simultaneously execute a combined control strategy of dynamically reducing the TTC trigger threshold, extending the number of valid target confirmation frames, and limiting the instantaneous maximum braking deceleration when identifying false hail targets, based on the threshold parameters output by the dynamic risk assessment module and the target determination results. When identifying real dangerous obstacles and the collision risk exceeds the dynamic AEB trigger threshold, it switches to output standard AEB emergency braking control commands. The brake recovery control module monitors the vehicle's braking execution status, identifies erroneous braking behavior caused by momentary hail interference, releases the vehicle's braking pressure in stages, and simultaneously outputs power compensation commands to smoothly restore the vehicle's power output and correct the vehicle's driving posture.

Images