Windshield visor peeling and vehicle emergency control method

Abstract

The present application belongs to the field of road vehicle control system, and relates to a windshield shelter stripping and vehicle emergency control method. The method accurately identifies the material and adhesion state of the shelter, combines the material stripping difficulty factor, shelter coverage area and current environment to assess the stripping feasibility of the shelter. When the shelter meets the stripping conditions, the system actively performs static electricity elimination and vehicle posture control to remove the shelter and quickly restore the driver's vision. When the shelter cannot be stripped, the system automatically switches to the emotion soothing, control management and vehicle-road collaborative compensation mode to guide the driver to perform the side parking operation. This double-path design of "treating the symptoms when the shelter can be stripped and treating the root cause when the shelter cannot be stripped" covers all possible working conditions of the highway foreign object sheltering scene and avoids the limitations of the "one-size-fits-all" alarm of the prior art.

Description

Technical Field

[0001] This invention belongs to the field of road vehicle control systems, and specifically relates to a method for removing obstructions from windshields and for emergency vehicle control. Background Technology

[0002] When a vehicle's visibility is suddenly obstructed by a foreign object while driving on the eastern expressway, it can easily cause panic and improper operation by the driver, leading to a traffic accident. Currently, the solution to this scenario is to remove the obstruction promptly. Most existing removal methods rely on windshield wipers, but their ability to remove large areas of tarpaulins, cardboard, etc., is limited.

[0003] Chinese patent CN116039565B proposes a method, equipment, storage medium, and device for cleaning a windshield. The method determines and analyzes the target visual area where foreign objects are located based on the cleaning equipment, and cleans the dirt in a targeted manner based on different analysis results to improve the cleaning effect. The method identifies the factors that cause obstruction, but the cleaning strategy mainly relies on adjusting the wipers, which is relatively simple.

[0004] Chinese patent CN121716648A proposes a method, system, and vehicle for removing obstructions from a car windshield, but this method relies on the sunroof and vehicles approaching from the side, and its applicability is insufficient.

[0005] Chinese patent CN121404246A proposes a method, system, and vehicle for removing windshield obstructions. The method includes: triggering an emergency mode when the windshield of a target vehicle is detected to be covered by an obstruction; acquiring vehicle status parameters of the target vehicle and cooperating vehicles while the target vehicle is in emergency mode; generating a collaborative adjustment strategy based on the vehicle status parameters of the target vehicle and cooperating vehicles; and adjusting the driving state of the target vehicle and cooperating vehicles based on the collaborative adjustment strategy to change the flow field distribution in the windshield area of ​​the target vehicle, thereby removing the obstruction. This method can proactively utilize surrounding cooperating vehicles to quickly and automatically remove obstructions in emergency scenarios where a large obstruction suddenly covers the windshield while the target vehicle is traveling at high speed. However, this method still has the following problems: (1) The emergency strategy is simple. For obstructions that cannot be removed in a short period of time, it ignores the driver's emotions and does not design an appropriate emergency response plan based on the driver's actual driving state and psychological feelings, which is not humane enough; (2) It lacks visual adaptation protection. Different types of obstructions have different light-blocking effects. When an obstruction with good light-blocking effect is quickly removed, the sudden change in external light can easily cause discomfort such as blurred vision and glare to the driver. (3) Relying on external vehicles and using surrounding cooperating vehicles to adjust the driving state to change the flow field distribution, but it cannot be guaranteed that there will always be available cooperating vehicles around during the driving process. In summary, the existing methods for removing windshield obstructions still have many shortcomings. Summary of the Invention

[0006] In view of the shortcomings and deficiencies of the existing technology, the purpose of this invention is to provide a method for removing obstructions from windshields and for emergency vehicle control. This method accurately identifies the material and attachment status of the obstruction and innovatively introduces the peeling feasibility index (DI) as a grading judgment criterion. When the obstruction is peelable, the system actively performs static electricity elimination and vehicle attitude control to remove the obstruction and quickly restore the driver's vision. When the obstruction cannot be peeled, the system automatically switches to an emotional calming, control management, and vehicle-road cooperative compensation mode to guide the driver to perform a pullover operation.

[0007] To achieve the above objectives, the present invention adopts the following technical solution: A method for removing obstructions from windshields and controlling vehicles in an emergency, the method comprising the following steps: Step S1. Combine the vibration spectrum of the obstruction hitting the glass collected by the solid-state acoustic sensor with the machine learning model to identify the material type of the obstruction and map it to the material peeling ease factor. Step S2. Use a planar capacitive electrostatic sensor to detect the amount of electrostatic charge between the obstruction and the windshield, estimate the electrostatic attraction force, and assess the feasibility of removing the obstruction by combining the material peeling ease factor, the coverage area of ​​the obstruction, and the current environment; if it is worth trying to remove, proceed to step S3; otherwise, proceed to step S4. Step S3. Use an ion generator to eliminate static electricity and verify static residue; estimate the shading rate of the obstruction, and based on the estimated shading rate and ambient light, obtain the target initial transmittance of the windshield before peeling it off, and reduce the transmittance of the windshield to the target initial transmittance before the peeling action is performed; determine the peeling action mode based on real-time ambient wind information and vehicle status, select the optimal vehicle attitude control action, and use inertial force and aerodynamic force to peel off the obstruction; at the same time, perform driver line-of-sight adaptive composite control, calculate the target transmittance, and after the peeling action is completed, control the windshield transmittance to transition from the target initial transmittance to the target transmittance according to an exponential law; Step S4. Assess the driver's emotional state and implement graded control intervention based on the stress level; at the same time, send a compensatory vision request to the roadside unit within the communication range via C-V2X, select the RSU screen that meets the conditions, issue a rear warning and guide the driver to perform the operation of pulling over.

[0008] As a preferred embodiment of the present invention, in step S1, a solid-state acoustic sensor is arranged in each of the four corner areas of the inner surface of the windshield. The signals collected by each solid-state acoustic sensor are pre-emphasized and filtered, then windowed in frames, and the power spectral density is calculated using the Welch method. The relative energy of the frequency band is extracted, the spectral centroid is calculated, and the logarithmic attenuation rate is extracted from the time-domain signal envelope to obtain the feature vector of each solid-state acoustic sensor channel. The time-domain signal envelope is obtained by taking the modulus after performing a Hilbert transform on the pre-emphasized and filtered signal. The feature vectors of the four solid-state acoustic sensors are then concatenated and input into the ELM classifier, which outputs the probability of the corresponding material category. The occlusion material type is determined based on the maximum probability value and mapped to the material peeling ease factor.

[0009] As a preferred embodiment of the present invention, in step S2, four planar capacitive electrostatic sensors are arranged along the four edges of the outer surface of the windshield. Each of the four planar capacitive electrostatic sensors independently detects the change in static charge in its respective area, calculates the equivalent charge in the area where each planar capacitive electrostatic sensor is located, and takes the largest absolute value of the four equivalent charge as the equivalent static charge of the entire shield, which is used to estimate the electrostatic adsorption force.

[0010] As a preferred embodiment of the present invention, the method for determining the feasibility of stripping in step S2 is as follows: ; in, Feasibility index for divestiture; Material peeling ease factor; : The preset maximum threshold for electrostatic adsorption force; The magnitude of the electrostatic attraction between the obstruction and the windshield. : Environmental wind field auxiliary factor, initially set to 0, and updated according to actual wind conditions; Non-electrostatic adhesion factor, based on the proportion of the area covered by the obstruction and correction for ambient humidity, with a value range of... ; All are weighting coefficients.

[0011] As a preferred embodiment of the present invention, the ion generator in step S3 is positioned close to the base of the rearview mirror inside the vehicle, with its ion release port facing the inner surface of the windshield and the release direction forming an angle of 30° to 45° with the plane of the windshield. When using the ion generator to eliminate static electricity, the total number of ions that the ion generator needs to release is first calculated based on the equivalent static charge of the entire obstruction, thereby determining the working time of the ion generator. Then, the ion generator is activated to release ions with opposite charge polarity to the obstruction onto the inner surface of the windshield, neutralizing the static charge on the outer surface area.

[0012] As a preferred embodiment of the present invention, in step S3, the stripping action mode is determined according to the current environmental wind conditions. The stripping action mode is divided into three types according to tailwind, headwind, and crosswind. Then, based on real-time environmental wind information and vehicle status, the lateral inertial force on the obstruction under vehicle attitude control is calculated. Lift / shear force generated by airflow on obstructions The system measures the longitudinal inertial force on the obstruction and determines whether the current peeling action can be successfully performed. If it can, the peeling action is executed immediately. If it cannot, the attitude control parameters are adjusted within the safety constraints and recalculated and verified again. If the parameters have been adjusted to the preset safety upper limit and the peeling conditions still cannot be met, the peeling attempt is deemed to have failed, the active peeling is abandoned, and the process proceeds to step S4 and executes the safe side-by-side procedure.

[0013] As a preferred embodiment of the present invention, in step S3, the stripping result is first calculated based on the target transmittance and the target initial transmittance. The system measures the feedforward transmittance per second; simultaneously, it uses an in-vehicle camera to monitor the driver's pupil diameter in real time, and uses PID control to perform closed-loop correction of the feedforward transmittance, so that the driver's pupil diameter is eventually stabilized in a comfortable range and the transmittance reaches the target transmittance.

[0014] In a preferred embodiment of the present invention, step S4 first obtains the driver's facial stress score, physiological stress score, vocal stress score, and behavioral stress score, and then weights and fuses these four scores to obtain the driver's emotional stress index. ,like If so, then complete control is granted; if If so, then relinquish control and initiate proactive reassurance; if Rapid acceleration is prohibited, and single steering angles are limited to >15°; if If the vehicle controller immediately takes over the throttle and brakes, the speed will be reduced to below 40 km / h with a deceleration of 0.2g.

[0015] As a preferred embodiment of the present invention, the RSU image in step S4 must simultaneously satisfy both the view consistency assessment and the latency assessment. If the latency assessment is not satisfied, a request to reduce the image quality is first sent to the current RSU. If the image quality is still not satisfied after reducing the image quality, a compensation request is resent to other roadside units. If an RSU image that meets the conditions still cannot be obtained, the forward-looking camera image of the vehicle ahead is requested to be shared via V2V communication, and the view deviation and total latency are recalculated to assess availability. If a compensation image that meets the conditions still cannot be obtained, the system is downgraded to blind spot monitoring guidance mode.

[0016] As a preferred embodiment of the present invention, in step S4, when guiding the vehicle to safely pull over to the side of the road, the current lane and position of the vehicle are first determined. Then, all parking areas within a specified range ahead are retrieved from the map database. The comprehensive score of each candidate area is calculated, and the candidate area with the highest comprehensive score is selected as the target parking point. For vehicles that have acquired compensation images, the driver is guided through a head-up display, voice prompts, and instrument panel auxiliary display until the target parking area is reached. If the current blind spot monitoring guidance mode is in operation, the driver is guided to perform the following operations only through voice prompts and instrument panel auxiliary display: First, maintain stable driving in the current lane and gradually decelerate. When the blind spot radar confirms that the right lane is safe, the driver is prompted to change lanes to the right. This process is repeated until the vehicle enters the target parking area. If there are vehicles in the right lane and the driver cannot change lanes, the driver is prompted to maintain the current lane and change lanes when conditions permit.

[0017] As a further preferred embodiment of the present invention, the expression for the non-electrostatic adhesion factor in step S2 is: ; in, The effective windward area of ​​the obstruction on the windshield. Total visible area of ​​the windshield. The current relative humidity. The baseline relative humidity is and These are the weighting coefficients; The expression for the environmental wind field auxiliary factor is: ; ; in, The relative airflow velocity felt on the surface of the obstruction; Reference wind speed; Horizontal composite wind speed; : The absolute value of the angle between the wind direction and the vehicle's longitudinal axis; : Vehicle speed relative to the ground scalar.

[0018] Advantages and beneficial effects of the present invention: (1) This invention accurately identifies the material and adhesion status of the obstruction. The vibration spectrum of the obstruction hitting the glass, collected by the solid-state acoustic sensor, is combined with the machine learning model to identify the material type of the obstruction. At the same time, a planar capacitive electrostatic sensor is used to detect the amount of static charge between the obstruction and the glass, and the ease of peeling off the obstruction is comprehensively evaluated, providing a basis for subsequent decision-making.

[0019] (2) This invention proposes a tiered emergency response architecture, with two parallel paths: proactive removal of detachable obstructions and collaborative extrication for non-detachable obstructions. It innovatively introduces a detachment feasibility index (DI) as the tiered judgment criterion. When the obstruction is detachable, the system proactively performs electrostatic discharge and vehicle attitude control to remove the obstruction, quickly restoring the driver's visibility. When the obstruction is non-detachable, the system automatically switches to an emotional calming, control management, and vehicle-road cooperative compensation mode, guiding the driver to pull over. This dual-path design, which addresses both detachable and non-detachable obstructions, covers all possible scenarios of foreign object obstruction on highways, avoiding the limitations of the existing "one-size-fits-all" alarm system.

[0020] (3) This invention pioneers an active obstacle removal method based on vehicle attitude control, filling the technological gap in non-contact foreign object removal. Existing technologies rely on windshield wipers or passive airflow to remove foreign objects, which are extremely ineffective for removing large areas of tarpaulins, wet cardboard, and other attached materials. This invention actively removes foreign objects by controlling the vehicle to perform micro-frequency serpentine steering, instantaneous acceleration and deceleration vibrations, and other attitude changes, utilizing the coupling effect of inertial force and aerodynamics. It also innovatively introduces real-time environmental wind field perception to adaptively select the optimal removal action. This method does not rely on additional actuators and can be implemented by reusing the vehicle's existing chassis control system, combining high efficiency and economy.

[0021] (4) This invention establishes a visual adaptation protection mechanism of "feedforward pre-darkening + pupil feedback closed loop" to prevent secondary glare injury at the moment of removal. In the prior art, after the obstruction is removed, the driver's pupil needs to switch from dark adaptation to light adaptation instantly, which can easily cause secondary accidents due to glare. Before removal, this invention pre-reduces the light transmittance of the smart dimming glass by detecting the light shading rate, and after removal, it smoothly transitions with an exponential curve. It also monitors the driver's pupil diameter in real time and performs closed-loop correction through PID control, making the visual recovery process imperceptible, comfortable and safe.

[0022] (5) This invention constructs a multimodal driver emotional stress assessment and hierarchical control management system. Existing systems only issue alarm prompts after detecting obstruction, without considering the risk of driver error in extreme panic. This invention integrates four modalities: facial movement units, heart rate variability, voice features, and driving operation behavior, to quantify the driver's emotional stress index (ESI) in real time, and executes a hierarchical control strategy from "suggestive reassurance" to "complete vehicle takeover" according to the stress level. This mechanism incorporates the "human" factor into the emergency decision-making closed loop, preventing secondary accidents caused by drivers suddenly braking or swerving due to panic from the source.

[0023] (6) This invention implements multiple safety redundancies in scenarios where the obstruction cannot be removed. When it is determined that the obstruction cannot be removed, three safety channels are activated simultaneously: emotional intervention to prevent panic and misoperation; vehicle-road coordination to request roadside units to compensate for the lack of vision; and rearward warning to calculate the collision time and broadcast a warning to the following vehicle. The three measures work together to plan the safest path to guide the vehicle to pull over, significantly reducing the probability of secondary accidents and filling the safety redundancy gap in the existing technology in this scenario. Attached Figure Description

[0024] Other objects and results of the invention will become more apparent and readily understood with reference to the following description taken in conjunction with the accompanying drawings. In the drawings: Figure 1 The present invention provides a flowchart of a method for removing obstructions from windshields and for emergency vehicle control. Detailed Implementation

[0025] To enable those skilled in the art to better understand the technical solutions and advantages of the present invention, the present application will be described in detail below with reference to the accompanying drawings, but this is not intended to limit the scope of protection of the present invention.

[0026] like Figure 1 As shown, this embodiment provides a method for removing obstructions from windshields and for emergency vehicle control, the method including the following steps: Step S1. Obstruction impact detection and material identification: Four solid-state acoustic sensors are arranged in each of the four corner areas of the inner surface of the windshield. The four solid-state acoustic sensors are located at the lower left corner, lower right corner, upper left corner, and upper right corner of the windshield, respectively, and are attached to the inner side of the glass with optically transparent adhesive to ensure close contact with the glass to effectively transmit vibration signals.

[0027] The rationale behind this arrangement is that impacts at different locations will elicit vibration responses with varying amplitudes and phase distributions. Four solid-state acoustic sensors located at the corners form a distributed acoustic sensing array, which can achieve rough location of the impact and enhanced robustness in material identification by utilizing differences in arrival time and energy distribution.

[0028] In this embodiment, solid-state acoustic signal preprocessing is required:

[0029] in, : No. The original signal amplitude of the solid-state acoustic sensor at each sampling point, in volts (V); : Amplitude of the pre-emphasized filtered signal, in volts (V); Pre-weighting coefficient, ranging from 0.95 to 0.98, dimensionless; : Sampling point number, which is a positive integer.

[0030] The preprocessed signal is then framed and windowed, and the power spectral density is calculated using the Welch method.

[0031] in, :frequency The power spectral density value at the location, in V² / Hz; Vibration signal frequency, unit: Hertz (Hz); : The total number of signal frames used for averaging, a positive integer; The number of signal sampling points per frame, which is a positive integer; : No. The first frame The pre-emphasis signal amplitude at each sampling point, in volts (V); : Window function in the first The value of the point, , The sampling point number; Window function normalization factor Dimensionless; Imaginary unit, =−1.

[0032] Band relative energy extraction:

[0033] in, : No. Normalized relative energy percentage for each frequency band, dimensionless, with a value range of 0 to 100; : No. The lower cutoff frequency of a frequency band, in Hertz (Hz). : No. The upper cutoff frequency of a frequency band, in Hertz (Hz). Frequency band number, with a value range of: to Integers; Total frequency bands, a positive integer; : Integral variable, representing frequency, unit: Hertz (Hz).

[0034] Spectrum centroid calculation:

[0035] in, : Centroid of the spectrum, representing the center frequency of the energy distribution, unit: Hertz (Hz); : No. The center frequency of each frequency band Unit: Hertz (Hz).

[0036] Logarithmic decay rate calculation: Extracting the logarithmic decay rate from the time-domain signal envelope:

[0037] in, The logarithmic decay rate of the signal envelope reflects the rate of vibration decay and is dimensionless. The total number of peak points selected on the signal envelope, which is a positive integer; Time-domain signal envelope The amplitude of each peak, in volts ( The time-domain signal envelope is obtained by applying a pre-emphasis filter to the signal. The result is obtained by performing a Hilbert transform and taking the modulus. The first detected on this envelope sequence The amplitude of each local maximum point; Peak point number, ranging from 1 to .

[0038] Material classification: The aforementioned pre-emphasis filtering, power spectral density calculation, frequency band relative energy extraction, spectral centroid calculation, and logarithmic attenuation rate calculation are performed independently on the signals acquired by each of the four solid-state acoustic sensors, yielding the characteristic parameters of each solid-state acoustic sensor channel. The frequency band relative energy of the four channels is then calculated. Spectral centroid Logarithmic decay rate Concatenate the data according to channel number to construct a fused feature vector. This serves as the input for subsequent material classification.

[0039] Specifically, the fused feature vector constructed in this embodiment ,in For the first Feature vectors of each sensor channel Input the pre-trained Extreme Learning Machine (ELM) classifier:

[0040] in, : Output vector, where each element represents the probability of belonging to the corresponding material category, dimensionless; The ELM output weight matrix is ​​determined by training and is dimensionless. Activation functions, such as Sigmoid or ReLU; The input weight matrix for ELM is determined during training and is dimensionless. ELM bias vector, determined during training, is dimensionless. : Transpose symbol.

[0041] according to The maximum probability value determines the occlusion material type and maps it to the material peeling ease factor. (The range of values ​​is) The higher the value, the easier it is to separate through vehicle attitude control or airflow. For example, plastic bags are 1.0, cardboard is 0.4, and wooden boards are 0.0.

[0042] Step S2. Evaluation of the electrostatic adsorption state of the obstruction: A planar capacitive electrostatic sensor placed on the outer surface of the windshield is used to detect the amount of static charge between the obstruction and the windshield, and to comprehensively assess the ease of removing the obstruction.

[0043] The physical structure of the planar capacitive electrostatic sensor is as follows: It adopts an interdigitated electrode structure, consisting of a pair of interlaced comb-shaped electrodes. The electrode material is an indium tin oxide transparent conductive film, directly deposited on the outer surface or interlayer of the windshield. The electrode width is 50 μm, the electrode spacing is 100 μm, and the effective sensing area size is 10 mm × 10 mm. The sensor surface is covered with a silicon dioxide insulating protective layer with a thickness of 200 nm.

[0044] Placement: Four planar capacitive electrostatic sensors are placed along the four edges of the windshield's outer surface, near the four corners of the glass (avoiding the wiper area), with signals led out via transparent wires. The four sensors are connected in parallel to a differential capacitance detection circuit to collectively detect the accumulation of static charge across the entire windshield surface.

[0045] In this embodiment, the formula for calculating the charge carried by the obstruction is:

[0046] in, : Equivalent electrostatic charge of the obstruction in the c-th sensor region, in coulombs (C); The reference capacitance value of a planar capacitive electrostatic sensor is determined by the sensor's physical structure, and its unit is farad (F). Change in sensor output voltage before and after an obstruction is attached, in volts (V). : Charge coupling efficiency coefficient, characterizing the degree to which the glass dielectric weakens the electric field induction, with a value range of , dimensionless.

[0047] Furthermore, in this embodiment, the reference capacitance value Defined as the intrinsic capacitance of the sensor in a clean, uncharged state. The calibration process is performed in a standard environment (temperature 25℃, relative humidity 40%) before the vehicle leaves the factory: First, the windshield surface is thoroughly cleaned with deionized water and isopropanol and then dried; then, the capacitance value of each sensor is measured using an LCR meter at a frequency of 1kHz, and the measurement is repeated 10 times, with the average value taken as the intrinsic capacitance of that sensor. The calibration values ​​are stored in the non-volatile memory of the vehicle controller. During subsequent use, the system can periodically and automatically perform zero-point calibration while the vehicle is parked to compensate for temperature drift and aging effects.

[0048] In this embodiment, the four planar capacitive electrostatic sensors independently detect the change in electrostatic charge in their respective regions and calculate the equivalent charge in each sensor region. ( The maximum absolute value of the charge measured by the four planar capacitive static sensors is taken as the equivalent static charge of the entire obstruction. ,Right now:

[0049] The basis for this maximum value strategy is that the electrostatic adsorption force is mainly determined by the area with the highest local charge density, and the electrostatic force at this location is a key factor affecting the peeling difficulty.

[0050] Estimation of electrostatic adsorption force:

[0051] in, : The magnitude of the electrostatic attraction between the obstruction and the windshield. A positive value indicates attractive force (resisting peeling), and a negative value indicates repulsive force (facilitating peeling). The unit is Newton (N). Vacuum permittivity ; The relative dielectric constant of windshield glass is typically taken as... Dimensionless; The equivalent static charge pre-existing on the outer surface of the windshield can be obtained through calibration, and the unit is coulomb (C). The equivalent charge distance between the obstruction and the windshield is typically on the order of micrometers. For consistency, the unit in this embodiment is meters (m). Before the vehicle leaves the factory, this parameter is obtained by attaching a standard test object with a known charge to the windshield surface and measuring the change in the output voltage of the electrostatic sensor. This change is then substituted into the electrostatic force calculation formula, and the equivalent charge distance can be calculated by reverse calculation. The average value of multiple test results under different charge amounts and temperature / humidity conditions can be obtained. The calibration value.

[0052] Comprehensive assessment of the feasibility of divestiture:

[0053] in, : Stripping feasibility index, dimensionless, the larger the value, the easier it is to strip; Material peelability factor, dimensionless; : The preset maximum threshold for electrostatic adsorption force, when Values ​​exceeding this are considered to indicate extremely strong electrostatic adsorption. Unit: Newton (N). : Environmental wind field auxiliary factor, the initial value can be set to 0, and it will be updated according to the actual wind conditions. It is dimensionless. Non-electrostatic adhesion factor, based on the proportion of the area covered by the obstruction and correction for ambient humidity, with a value range of... Dimensionless ,in To determine the effective windward area of ​​the obstruction on the windshield, an in-vehicle forward-facing camera captures images of the windshield area. Semantic segmentation algorithms (such as U-Net or DeepLab series networks) are used to perform pixel-level identification of the obstructed area, and the number of obstructed pixels is counted. Total number of pixels in the windshield area Calculate the area covered by the obstruction using the following formula: ,in Total visible area of ​​the windshield. The camera's field-of-view distortion correction coefficients are determined through geometric calibration. The current relative humidity. As the reference relative humidity, it can be taken as... Below this value, the effect of moisture is negligible. The first term in the formula reflects that the larger the shading area, the stronger the adhesion, and the second term reflects that the higher the humidity, the stronger the moisture adsorption. and These are weighting coefficients, determined by experimental calibration. : Weight coefficients, all of which are positive and satisfy The value is dimensionless and can be determined by expert scoring. Multiple experts in vehicle dynamics, aerodynamics, and human factors are invited to independently score the importance of each factor based on their professional knowledge. The score is then normalized and used as a reference value for weighting.

[0054] like (e.g., 0.65), if it is determined that it is worthwhile to attempt stripping, proceed to step S3; if If it is determined that the stripping is not worth attempting, proceed to step S4; where, This is the stripping threshold.

[0055] Step S3. Strippable Strategy: Step S3.1. Pretreatment before static elimination: Specifically, in this embodiment, an ion generator is used to eliminate static electricity, and the total amount of ions to be released is:

[0056] in, The total number of ions that the ion generator needs to release is a positive integer. Elementary charge Coulomb (C); Safety margin factor, range of values It ensures complete neutralization with slight compensation and is dimensionless.

[0057] The working time of the ion generator is:

[0058] in, The duration for which the ion generator needs to operate continuously, in seconds (s); : Nominal ion release rate of the ion generator, unit: ions / second (ions / s).

[0059] Furthermore, in this embodiment, the system activates the ion generator. Over time, ions with opposite charge polarity to the obstruction are released onto the inner surface of the windshield, neutralizing the static charge on the surface area through the glass dielectric polarization effect. The ion generator is installed inside the roof trim panel above the inner side of the vehicle's windshield (i.e., near the base of the rearview mirror), with its ion release port facing the inner surface of the windshield. The release direction is at an angle of 30° to 45° to the windshield plane, ensuring that the released positive and negative ions can evenly cover the entire inner surface area of ​​the windshield.

[0060] Step S3.2. Verification of electrostatic residue: After the static electricity has been eliminated, check the static voltage sensor reading again to confirm that the static electricity has been effectively eliminated.

[0061] Calculate the residual charge:

[0062] in, Equivalent static charge remaining on the glass surface after static electricity elimination, unit: coulomb (C). : The change in sensor output voltage relative to the initial baseline after static electricity is eliminated, in volts (V).

[0063] Static electricity is considered successfully eliminated if the following conditions are met:

[0064] in, Acceptable residual static charge safety threshold, for example Coulomb (C). If not satisfied, the system extends the operating time of the ion generator.

[0065] Furthermore, in this embodiment, after the static electricity is eliminated, the system calls the electrostatic adsorption force estimation formula in step S2 to recalculate the current residual electrostatic adsorption force based on the amount of residual charge after elimination. And use it as the input for the subsequent step S3.4 stripping condition determination.

[0066] Step S3.3. Detection of obstruction shading rate and pre-darkening in front of glass: Before the removal process, the shading rate of the obstruction is measured, and the light transmittance of the windshield with dimming function is reduced in advance to prepare for visual adaptation after removal. Shading rate estimation:

[0067] in, : Light transmittance attenuation coefficient (blocking rate) of the obstruction, with a value range of 0 to 1, dimensionless, where 1 indicates complete opacity; Light intensity measured in the shaded area, in lux (lx); Ambient light intensity measured simultaneously in an unshaded area (such as a side window), in lux (lx).

[0068] Based on the estimated shading rate and ambient light, reduce the light transmittance of the dimming glass in advance:

[0069] in, : The initial light transmittance of the dimming glass (windshield) before the peeling action is performed. It is dimensionless and ranges from 0 to 1. The minimum light transmittance allowed for the dimming glass to prevent it from turning completely black, for example, 0.1, dimensionless; Ambient light intensity, measured by the vehicle's external light sensor, unit: lux (lx); Light intensity threshold that may cause glare to the driver, preset value, unit: lux (lx).

[0070] Those skilled in the art know that the transmittance of smart dimming glass (such as PDLC thin-film type or electrochromic type) can be continuously adjusted by changing the magnitude of the driving voltage applied across its terminals: the higher the voltage, the lower the transmittance; when the driving voltage drops to zero, the transmittance returns to its maximum. In this step, the vehicle electronic control unit calculates the target initial transmittance to be preset before peeling according to the above formula. The drive circuit outputs the corresponding target voltage value to the dimming glass, causing the light transmittance of the dimming glass to gradually decrease within 0.5 seconds before the peeling action is executed. The mapping relationship between the target voltage value and light transmittance has been calibrated and stored in the vehicle's electronic control unit before leaving the factory.

[0071] Step S3.4. Separation of Environmental Perception and Adaptation: In this embodiment, the system first determines the stripping action mode based on the current environmental wind conditions, and then selects the optimal vehicle posture control action based on real-time environmental wind information and vehicle status, using the coupling effect of inertial force and aerodynamic force to strip away the obstruction.

[0072] Table 1 shows the stripping action mode selected by the system based on environmental wind conditions: Environmental wind conditions peeling action Parameter range Smooth Micro-frequency serpentine steering The Electronic Stability Program (ESP) controller sends a sinusoidal wave superimposed steering angle command to the Electric Power Steering (EPS) system via the CAN bus, causing the front wheels to oscillate back and forth at a preset frequency and amplitude within a lateral deviation of ±0.5m from the lane centerline. ESP simultaneously monitors the yaw rate to ensure vehicle stability. against the wind Instantaneous acceleration and deceleration excitation The vehicle control unit (VCU) sends torque pulse commands alternately to the drive motor or electronic throttle and ESP brake actuator via the CAN bus. Within a set time window, braking torque (deceleration of 0.2~0.3g) is applied for 0.3~0.5 seconds, and then the cruising torque is restored, using the change in longitudinal inertial force to generate an excitation effect. crosswind Limited steering towards the windward side The Electric Power Steering (EPS) system actively applies a unidirectional bias torque when the driver does not intervene in steering, causing the vehicle to yaw at a small angle towards the windward side, utilizing crosswinds to help clear obstructions. The magnitude of the bias torque is calculated in real time by the ESP based on the current crosswind speed, and the steering angle is strictly limited within the current lane width. In this embodiment, the parameters (steering angle, deceleration, etc.) in each stripping action mode are preset values. During actual execution, they can be slightly adjusted within the safety constraints so that the vehicle attitude control action can meet the stripping conditions and successfully strip the obstacle.

[0073] Specifically, in this embodiment, the vehicle's velocity vector in the geographic coordinate system (NED) is first obtained. Unit: meters per second (m / s). Construct the rotation matrix from the geographic coordinate system (N) to the vehicle coordinate system (B). :

[0074] The rotation matrices for each axis are:

[0075]

[0076]

[0077] in, Roll angle of vehicle body, unit: radians (rad); Vehicle pitch angle, unit: radians (rad); Vehicle heading angle, unit: radians (rad).

[0078] True wind speed vector in vehicle coordinate system for: ; in, : Relative wind speed vector measured by an ultrasonic anemometer in the vehicle coordinate system, unit: meters per second (m / s); : Vehicle speed relative to the ground scalar Unit: meters per second (m / s).

[0079] Horizontal composite wind speed acting on stripping and relative wind angle for:

[0080]

[0081] in, Horizontal composite wind speed, unit: meters per second (m / s); : The absolute value of the angle between the wind direction and the vehicle's longitudinal axis (X-axis), in radians (rad).

[0082] Update environmental wind field auxiliary factors :

[0083]

[0084] in, : The relative airflow velocity felt on the surface of the obstruction, in meters per second (m / s); Reference wind speed, used for normalization, for example, take 20 m / s, unit: meters per second (m / s).

[0085] Calculate the excitation force required for peeling: Lateral inertial force generated by vehicle attitude control :

[0086] in, Lateral inertial force on an obstruction, unit: Newton (N); Estimated mass of the obstruction, unit: kilograms (kg). , The effective windward area of ​​the obstruction on the windshield; The typical areal density of this material was obtained from a table, as shown in Table 2. Units: kg / m² 2 ); Lateral acceleration of the vehicle (calculated based on the steering angle corresponding to the planned stripping action), unit: meters per second. 2 (m / s) 2 ); : Vehicle speed relative to the ground scalar (current actual vehicle speed; because the stripping action lasts for a very short time, about 0.3~0.5 seconds, the vehicle speed changes very little during this period, so the current value can be directly substituted). Unit: meters per second (m / s); Front wheel equivalent steering angle (steering angle amplitude corresponding to the stripping action planned by the system), unit: radians (rad). Vehicle wheelbase, unit: meters (m); Insufficient vehicle volume: steering gradient, chassis characteristic parameter, unit: seconds. 2 / meter (s) 2 / m).

[0087] Table 2 shows the typical areal density of each material. Material number Material type Ease of stripping factor <![CDATA[Typical areal density (kg / m 2 ).]]> 01 plastic bag 1.0 0.02 02 cardboard 0.4 0.30 03 board 0.0 5.00 airflow shear force :

[0088]

[0089] in, : Lift / shear force generated by airflow on an obstruction, unit: Newton (N); Air density, taken as 1.225 kg / m³ under standard conditions. 3 ; : The relative airflow velocity felt on the surface of the obstruction, in meters per second (m / s); The aerodynamic lift coefficient of the obstruction is determined by its material and attitude, and is dimensionless. The effective windward area of ​​the obstruction on the windshield; The baseline lift coefficient is determined by the material of the obstruction, such as 0.8 for plastic bags, 0.5 for cardboard, and 0.2 for wood. : This is the excitation enhancement factor, related to the excitation intensity of the peeling action; micro-tilt rudder steering mode: ,in This is the current steering angle amplitude. The upper limit of the safe steering angle (corresponding to the lateral deviation) When the steering angle is at its maximum, Maximum Instantaneous force / deceleration excitation mode: ,in Deceleration to the safety limit (take) ), The vehicle's longitudinal deceleration, under instantaneous acceleration and deceleration excitation mode, is set to a preset value (0.2~0.3g, i.e., 1.96~2.94m / s²), unit: meters per second² (m / s²). When the deceleration reaches the upper limit, Maximum If no vibration is required (only airflow), then .

[0090] Longitudinal inertial force:

[0091] in, : The longitudinal inertial force acting on the obstruction, in Newtons (N); Take the preset deceleration value for this action.

[0092] In this embodiment, the peeling condition verification is based on the selected peeling action mode and the corresponding peeling force for determination: When performing micro-frequency serpentine steering, the peeling force is contributed by both lateral inertial force and aerodynamic shear force; during verification, the steering angle in the lateral inertial force is taken as the preset amplitude of the action, and if the conditions are met... If so, it will be executed immediately; among which, : Base non-electrostatic adhesion between the obstruction and the windshield, determined by test, unit: Newton (N). : Current residual electrostatic adsorption force.

[0093] When instantaneous acceleration and deceleration are applied, the peeling force is contributed by both longitudinal inertial force and aerodynamic shear force; if the following conditions are met... If so, it will be executed immediately.

[0094] When turning towards the windward side under crosswind conditions, the peeling force simultaneously includes lateral inertial force, longitudinal inertial force, and aerodynamic shear force, and the vector sum of the three is used for determination.

[0095] If the inequality is not satisfied in the corresponding mode, the action parameters (steering angle magnitude or deceleration value) are fine-tuned within the safety constraints and then recalculated. If the parameters are adjusted to the safety upper limit and still not satisfied, the current stripping attempt is abandoned and the process is switched to the non-stripping branch to execute the safety side-by-side procedure.

[0096] In this embodiment, before performing the action, it is necessary to confirm, through blind spot monitoring radar and V2X, that there are no parallel or rapidly approaching vehicles to the side or rear, and that the deceleration does not exceed 0.3g.

[0097] Step S3.5. Driver's line-of-sight adaptation composite control: Target transmittance calculation:

[0098] in, : The desired transmittance of the dimming glass, dimensionless, range of values. ; Driver's eye comfort illumination, preset value Lux (lx).

[0099] After the peeling process is completed, the light transmittance of the dimming glass first decreases exponentially from... Towards transition:

[0100] in, After peeling Feedforward transmittance per second, dimensionless; Time elapsed from the moment the stripping process was completed, in seconds (s). The human eye's adaptation time constant is typically taken as 2.0 to 5.0 seconds, measured in seconds (s).

[0101] Using in-vehicle cameras to monitor the driver's pupil diameter in real time :

[0102] in, : Deviation between pupil diameter and target comfortable diameter, unit: millimeters (mm); : The diameter of the driver's pupils measured at all times, in millimeters (mm); The target pupil's comfortable diameter is typically 3.0–3.5 mm.

[0103] Calculate the PID feedback correction:

[0104] in, : PID correction amount for the current transmittance, dimensionless; Proportional gain, dimensionless; Integral gain, dimensionless; Differential gain, dimensionless; : Integral variable, representing time, unit: seconds (s).

[0105] PID parameters can be obtained through simulation. A dynamic model of light reflection from the human pupil is established, and a dimming glass-pupil feedback closed-loop control system is built in a simulation environment such as Simulink. The system is initially tuned using Ziegler-Nichols or internal model control methods. , , The parameters are set such that the system's step response overshoot is less than 10% and the settling time is less than 5 seconds.

[0106] Real-time transmittance after output correction:

[0107] in, : Final real-time transmittance after feedback correction, dimensionless; Limiting function to ensure transmittance is within a certain range. Within the range.

[0108] The system continues to execute the above closed-loop control until the driver's pupil diameter stabilizes within the comfortable range and the light transmittance reaches or approaches that level. .

[0109] In this embodiment, once the visual adaptation is complete, the system exits emergency mode, fully returns control of the vehicle, and restores normal driving assistance functions.

[0110] Step S4. Non-detachable strategy: When the obstruction cannot be removed, the system assesses the driver's emotional state and implements graded control interventions based on the level of stress.

[0111] Step S4.1. Driver Emotional Stress Assessment and Control Management:

[0112] in, Driver's emotional stress index, out of 100, ranges from 0 to 100, with higher values ​​indicating greater stress. 、 、 、 The weight coefficients of each sub-item are all positive numbers and their sum is 1. The specific values ​​are determined experimentally and are dimensionless.

[0113] The facial stress score is based on facial action units. Data acquisition method: Real-time acquisition of driver's facial images by a near-infrared camera (sampling frequency ≥30fps, resolution ≥640×480) installed in the A-pillar or above the dashboard inside the vehicle. The infrared light source wavelength is 850nm or 940nm to ensure clear facial features under nighttime and backlight conditions. Processing flow: Facial feature points are located using a pre-trained convolutional neural network model (such as OpenFace, AffectNet architecture), and the intensity of stress-related action units is extracted, including but not limited to: AU4 (frowning), AU5 (upper eyelid tightening), AU6 (orbicularis oculi muscle contraction), AU7 (eyelid tightening), AU17 (chin lifting), and AU23 (lip tightening). Each AU activation intensity is normalized to [0,1]. The intensities of each AU are then summed using preset weights. These weights are determined with reference to the FACS manual for psychology and stress-related research, with typical values ​​being: AU4 (0.25), AU5 (0.15), AU6 (0.20), AU7 (0.15), AU17 (0.10), and AU23 (0.15). Finally, the weighted sum is mapped to a 0-100 percentage system to obtain... 。

[0114] The physiological stress score is based on heart rate variability. Data acquisition method: Real-time acquisition of the driver's ECG signal by a capacitive or photoelectric ECG sensor (sampling frequency ≥250Hz) embedded in the steering wheel rim. When the driver holds the steering wheel with both hands, the sensor automatically contacts the skin to acquire the signal. Processing flow: The raw ECG signal is bandpass filtered from 0.5 to 40Hz to remove motion artifacts and power frequency interference. The R-wave peak point is detected using the Pan-Tompkins algorithm or wavelet transform method. A continuous RR interval sequence is calculated. Within a 5-minute sliding time window, time-domain indices: SDNN (standard deviation of all RR intervals), RMSSD (root mean square of the difference between adjacent RR intervals) and frequency-domain indices: LF / HF (ratio of low-frequency power 0.04~0.15Hz to high-frequency power 0.15~0.4Hz) are calculated using Fast Fourier Transform or Lomb-Scargle periodogram method. Then, the above indices are compared with pre-established baseline values ​​under the driver's calm driving state to calculate the normalized deviation, mapped to 0~100, to obtain... The baseline value is automatically collected and established by the system during the "10-minute calm driving" calibration phase when the driver uses the system for the first time, and is updated regularly.

[0115] This is a speech stress score based on speech features (fundamental frequency variation, etc.). Data acquisition method: Real-time acquisition of driver speech signals by an in-vehicle microphone array (sampling frequency ≥16kHz, at least 2 microphones for noise reduction). The system automatically triggers acquisition when the driver issues voice commands, makes or receives phone calls, or speaks voluntarily. Processing flow: Endpoint detection is performed on the speech signal to remove silent segments; in-vehicle background noise is suppressed using multi-microphone beamforming or spectral subtraction; and the fundamental frequency mean is extracted from effective speech segments using open-source tools such as openSMILE or Librsa. , fundamental frequency standard deviation The acoustic features include speech rate (syllables / second), high-frequency energy proportion (the proportion of energy in the >1000Hz frequency band to the total energy), and the first 12 mean and variance of Mel-frequency cepstral coefficients (MFCC). These feature vectors are then compared with a pre-recorded baseline model (Gaussian mixture model or support vector machine model) under normal driver communication conditions. Mahalanobis distance or likelihood deviation is calculated, and the deviation is mapped to 0-100 to obtain... .

[0116] This is a behavioral stress score based on driving operations (frequency of sudden braking and sharp turns). Data acquisition method: Real-time acquisition of signals such as steering wheel angle, accelerator pedal position, brake pedal pressure, and turn signal status via the vehicle's CAN bus (Controller Area Network), with a sampling frequency ≥10Hz. Processing flow: Within a 30-second sliding time window, the following events are counted: sudden turns: the number of times the steering wheel angle change rate exceeds 200° / s; sudden acceleration: the number of times the accelerator pedal depth change rate exceeds 80% / s; sudden deceleration: the number of times the brake pressure change rate exceeds 10MPa / s; and steering wheel fine-tuning correction frequency: the number of times the steering angle changes frequently within ±3°. These event counts are then normalized and compared with the driver's historical smooth driving style model (a Gaussian distribution model based on long-term driving data) to calculate the likelihood deviation value of the current driving behavior. Finally, the deviation value is mapped to 0~100 to obtain the result. .

[0117] In this embodiment, the hierarchical control management rules are shown in Table 3: Table 3 shows the rules for hierarchical control management: ESI range Control strategy ESI<30 It offers complete control and provides reassuring voice advice. 30≤ESI<60 Reclaim control and initiate proactive soothing (music, ambient lighting, breathing guidance). 60≤ESI<80 Partial control restrictions: Prohibits rapid acceleration and limits single steering range to >15°. ESI≥80 Locking in critical vehicle control: The vehicle controller immediately takes over the throttle and brakes, gradually reducing the speed to below 40km / h with a deceleration of 0.2g. Step S4.2. Vehicle-Road Cooperative Compensation and Backward Early Warning: In this embodiment, once it is determined that the obstruction cannot be removed, this step is initiated immediately without waiting for the driver's emotional stress index to reach a specific threshold. The system sends a compensatory view request to the roadside unit within the communication range via C-V2X, selects the RSU image that meets the conditions, and continuously broadcasts an emergency warning message carrying the semantics "This vehicle's view is obstructed" to vehicles behind until the vehicle comes to a safe stop.

[0118] RSU compensation image viewing angle consistency assessment:

[0119] in, : The pointing deviation angle of the RSU screen relative to the driver's actual view, in radians (rad); :RSU camera optical center three-dimensional coordinate vector in the geodetic coordinate system, unit: meter (m); The driver's eye position in this vehicle is represented by a three-dimensional coordinate vector in the geodetic coordinate system, in meters (m). : The unit vector of the driver's current line of sight, which is dimensionless and is taken by default from the longitudinal axis of the vehicle; Euclidean norm, unit: meter (m).

[0120] when (like When ), the RSU screen is determined to be usable; otherwise, when (like When the RSU screen viewing angle is determined to be too large, the system will execute alternative strategies in the following priority order: Resend the compensation request to other roadside units within the communication range to reassess the viewing angle deviation. Select an RSU image that meets the conditions; if there is still no RSU that meets the viewing angle conditions, request the forward-facing camera image of the vehicle in front to be shared via V2V communication, recalculate the viewing angle deviation and assess availability; if it is still impossible to obtain a compensatory image that meets the viewing angle conditions, downgrade to blind spot monitoring guidance mode.

[0121] Compensation screen end-to-end latency assessment:

[0122] in, Total latency from RSU data acquisition to display on the vehicle screen, in seconds (s); Camera exposure and readout delay, unit: seconds (s); Video encoding latency, in seconds (s); Size of a single frame of compressed video data, in bits. C-V2X communication effective bandwidth, unit: bits per second (bit / s); Electromagnetic wave propagation delay, unit: seconds (s); Video decoding latency, in seconds (s); GPU rendering latency, in seconds (s).

[0123] when When the RSU screen is set to 100ms (e.g., 100ms), it is considered to be available in real-time; otherwise, when... If the delay is 100ms, the RSU screen latency is determined to be too high, and the system executes alternative strategies in the following priority order: Send a degradation request (reduction of image quality request) to the current RSU to reduce the image resolution from 1080p to 720p or 480p in order to reduce the amount of data per frame. This reduces transmission latency and allows for reassessment after quality degradation. If the latency requirement is still not met after degradation, a compensation request is sent to other roadside units within the communication range, prioritizing those with shorter communication distances or stronger coding performance, and the latency is reassessed. If no RSU meets the latency requirement, the forward-facing camera image is requested from the vehicle ahead via V2V communication. Because the vehicle distance is short and the V2V direct connection latency is usually lower than the RSU communication latency, the image obtained via V2V is more likely to meet the real-time requirements. The total latency is also calculated using the aforementioned end-to-end latency evaluation formula. If a compensation image that meets the latency requirement still cannot be obtained, the system is downgraded to blind spot monitoring and guidance mode.

[0124] In this embodiment, the priority of compensation service requests is as follows: This vehicle sent a compensation request to the RSU, with its priority... The calculation is as follows:

[0125] in, Priority of compensation request, ranging from 0 to 1, with higher values ​​indicating greater urgency; The remaining visible area of ​​the windshield when it is not obstructed, in square meters (m²). 2 ); Total visible area of ​​the windshield, unit: square meters (m²) 2 ); Current maximum speed limit on the road, unit: meters per second (m / s); Rear collision time, in seconds (s); : The kurtosis parameter of the Sigmoid function, dimensionless; : The center offset parameter of the Sigmoid function, in seconds (s); : All are weighting coefficients, which sum to 1 and are dimensionless.

[0126] RSU based on each requesting vehicle The system prioritizes providing compensation images to high-priority vehicles.

[0127] Furthermore, in this embodiment, rear collision events are calculated:

[0128] in, The time between this vehicle and the vehicle directly behind in the same lane is the distance and time of collision, in seconds (s). The longitudinal relative distance between the rear of this vehicle and the front of the vehicle behind it, in meters (m). The longitudinal speed of the following vehicle relative to this vehicle. There is a risk of collision only when the value is positive; unit: meters per second (m / s). The longitudinal speed of vehicles (rear vehicles) in the same lane directly behind this vehicle.

[0129] when At 3.5 seconds (e.g.), the system broadcasts a warning with the semantics "This vehicle's visibility is obstructed" to the following vehicle via V2X and controls the taillights to display an emergency pictogram.

[0130] Step S4.3. Guide the vehicle to safely pull over to the side of the road: Based on obtaining the RSU compensation screen (or downgrading to blind spot monitoring guidance mode) and continuously monitoring the driver's emotional state, this step plans an escape route and guides the driver to pull over to the side of the road, taking into account the above information.

[0131] Escape route planning: The system determines the vehicle's current lane and location based on high-precision maps and GPS positioning. Following the principle of "minimizing lane changes and maximizing stopping area," it retrieves information from the map database regarding the area ahead. For all parking areas within the scope (including hard shoulders, emergency parking bays, and service area entrances), calculate the comprehensive score for each candidate area using the following formula and select the optimal target:

[0132] in, : The overall score of the candidate parking area, dimensionless, with a higher value indicating better performance; Distance from the current location of this vehicle to the candidate parking area, in meters (m); : Upper limit of search range, with a value of 2000m; The available width of the candidate parking area is obtained from the shoulder width data provided by the high-precision map, in meters (m). The reference value for the width of the parking area is 3.5m (the width of a standard highway emergency lane). The minimum number of lane changes required for this vehicle to change lanes from its current lane to the candidate parking area; The maximum number of lane changes allowed by the system is capped at 3. The weighting coefficient for the distance term, for example, can be 0.3, and is dimensionless; The weighting coefficient for the width term, for example, can be 0.4, and is dimensionless; : The weighting coefficient for the number of lane changes, for example, a value of 0.3, dimensionless.

[0133] Weight setting basis: The highest weight is applied, prioritizing the safety of the parking area itself (the wider the area and the further away from the driveway, the safer it is). and Equal weighting, in candidate regions of similar width, balances reach distance and operational complexity. The system selects... The largest candidate area will be selected as the target parking spot. If two candidate areas have the same score, the one that is closer will be selected.

[0134] It should be noted that in this embodiment, the escape path consists of the number of lane changes required between the vehicle's current lane and the target parking area, as well as the stable driving segment in the lane after each lane change. The system guides the driver step by step to complete each right lane change through HUD lane line highlighting, voice prompts, etc., until the target parking area is reached.

[0135] Guiding information output: In this embodiment, the driver is guided in three ways for vehicles that have acquired compensation images: (1) Head-up Display (HUD): Based on the acquired compensatory image, the lane dividing line between the vehicle's current lane and the adjacent right lane is highlighted on the windshield. An arrow animation indicates the direction to change lanes to the right, and a digital countdown displays the distance to the target parking area ahead. When the vehicle is already in the rightmost lane, the HUD highlights the boundary markings between that lane and the parking area, guiding the driver to park in the hard shoulder, emergency parking bay, or service area entrance.

[0136] (2) Voice prompts: Announce operating instructions at a calm pace, such as "Please change lanes to the right," "Maintain current speed," and "Pull over 500 meters ahead." The repetition frequency of voice prompts is based on the driver's emotional stress index. Dynamic adjustment: Repeat every 3 seconds to reinforce the guidance; Once every 5 seconds to maintain continuous guidance; It is repeated every 8 seconds to reduce interference.

[0137] (3) Instrument panel auxiliary display: The instrument panel screen displays a simplified relative position diagram of the vehicle and surrounding obstacles (simplified graphical interface). This diagram is generated in real time based on blind spot monitoring radar data, and the threat level of vehicles to the side and rear is marked with red, yellow and green respectively (red indicates high collision risk, yellow indicates need to pay attention, and green indicates safety).

[0138] In blind spot monitoring guidance mode, guidance is provided solely through voice prompts and instrument panel displays. Specifically, the vehicle's millimeter-wave radar and ultrasonic sensors provide distance information to obstacles to the sides and rear. A simplified graphical interface and voice prompts guide the driver to perform the following actions: First, maintain stable driving in the current lane and gradually decelerate; once the blind spot radar confirms the right lane is safe, a voice prompt instructs the driver to change lanes to the right; repeat this process until the vehicle enters the emergency lane or hard shoulder and comes to a stop (target parking area). If there are vehicles continuously in the right lane preventing a lane change, the driver is prompted to maintain the current lane and continue driving slowly until conditions allow for a lane change.

[0139] Furthermore, in this embodiment, a safety fallback mechanism is also provided: If any of the following abnormal situations occur during the guidance process, the system will automatically switch to the corresponding safety fallback strategy, as detailed in Table 4: Table 4 lists the safety fallback trigger conditions and strategies: Triggering conditions bottom-line strategy RSU compensation screen interruption and V2V screen unavailable Relying solely on blind spot radar and voice prompts, the system instructs the driver to maintain their current lane and drive slowly while reducing speed. Once the blind spot radar confirms that the right lane is safe, the system prompts the driver to change lanes to the right. This process is repeated until the vehicle enters the emergency lane or hard shoulder and comes to a stop. If the right lane remains unsuitable for lane changing, the vehicle continues to drive slowly in the current lane while waiting for an opportunity. Blind spot radar detected a vehicle suddenly approaching in the target lane. The lane change guidance is paused, and a voice prompt says "There is a vehicle behind you, please wait." The lane change route will be replanned after the threat is eliminated. The vehicle has arrived at the target parking area, but the driver has not stopped the vehicle. If ESI ≥ 80, the vehicle controller will take over braking and stop the vehicle; if ESI < 80, the voice prompts will be strengthened until the driver executes the command. GPS signal lost Switch to inertial navigation to calculate positioning and maintain short-term navigation, while providing voice prompts for the driver to observe roadside signs and determine the parking area independently. Parking completed confirmation: After the vehicle has come to a complete stop, the system will automatically perform the following operations: (1) Engage the electronic parking brake; (2) Keep the hazard warning lights on continuously; (3) Automatically dial emergency rescue numbers and broadcast vehicle GPS location and event summary via vehicle communication module; (4) Continuously monitor the driver’s emotional stress index. Once the index drops below 60 and the driver’s field of vision returns to normal, gradually return control of the vehicle.

[0140] The present invention also provides an electronic device, comprising: one or more processors and a memory; wherein the memory is used to store one or more programs, and when the one or more programs are executed by the one or more processors, the one or more processors implement the above-described method for removing obstructions from windshields and for emergency vehicle control.

[0141] The present invention also provides a computer-readable medium having a computer program stored thereon, which, when executed by a processor, implements the above-described method for removing obstructions from windshields and for emergency vehicle control.

[0142] Those skilled in the art will understand that all or part of the functions of the various methods / modules in the above embodiments can be implemented by hardware or by computer programs. When all or part of the functions in the above embodiments are implemented by computer programs, the program can be stored in a computer-readable storage medium, which may include: read-only memory, random access memory, disk, optical disk, hard disk, etc., and the above functions can be implemented by executing the program with a computer. For example, the program can be stored in the memory of a device, and when the program in the memory is executed by the processor, all or part of the above functions can be implemented.

[0143] In addition, when all or part of the functions in the above embodiments are implemented by computer programs, the programs can also be stored in storage media such as servers, other computers, disks, optical discs, flash drives, or portable hard drives. They can be downloaded or copied to the memory of the local device, or the system of the local device can be updated. When the program in the memory is executed by the processor, all or part of the functions in the above embodiments can be implemented.

[0144] The above-described specific examples are for illustrative purposes only and are not intended to limit the scope of the invention. Those skilled in the art can make various simple deductions, modifications, or substitutions based on the principles of this invention. Therefore, the scope of protection of this invention should be determined by the scope of the claims.

Claims

1. A method for removing obstructions from windshields and for emergency vehicle control, characterized in that, The method includes the following steps: Step S1. Combine the vibration spectrum of the obstruction hitting the glass collected by the solid-state acoustic sensor with the machine learning model to identify the material type of the obstruction and map it to the material peeling ease factor. Step S2. Use a planar capacitive electrostatic sensor to detect the amount of electrostatic charge between the obstruction and the windshield, estimate the electrostatic attraction force, and assess the feasibility of removing the obstruction by combining the material peeling ease factor, the coverage area of ​​the obstruction, and the current environment; if it is worth trying to remove, proceed to step S3; otherwise, proceed to step S4. Step S3. Use an ion generator to eliminate static electricity and verify static residue; estimate the shading rate of the obstruction, and based on the estimated shading rate and ambient light, obtain the target initial transmittance of the windshield before peeling it off, and reduce the transmittance of the windshield to the target initial transmittance before the peeling action is performed; determine the peeling action mode based on real-time ambient wind information and vehicle status, select the optimal vehicle attitude control action, and use inertial force and aerodynamic force to peel off the obstruction; at the same time, perform driver line-of-sight adaptive composite control, calculate the target transmittance, and after the peeling action is completed, control the windshield transmittance to transition from the target initial transmittance to the target transmittance according to an exponential law; Step S4. Assess the driver's emotional state and implement graded control intervention based on the stress level; at the same time, send a compensatory vision request to the roadside unit within the communication range via C-V2X, select the RSU screen that meets the conditions, issue a rear warning and guide the driver to perform the operation of pulling over.

2. The method for removing obstructions from windshields and controlling vehicles in emergency situations according to claim 1, characterized in that, In step S1, a solid-state acoustic sensor is arranged in each of the four corner areas of the inner surface of the windshield. The signals collected by each solid-state acoustic sensor are pre-emphasized and filtered, then windowed in frames, and the power spectral density is calculated using the Welch method. The relative energy of the frequency band is extracted, the spectral centroid is calculated, and the logarithmic attenuation rate is extracted from the time-domain signal envelope to obtain the feature vector of each solid-state acoustic sensor channel. The time-domain signal envelope is obtained by taking the modulus after performing a Hilbert transform on the pre-emphasized and filtered signal. The feature vectors of the four solid-state acoustic sensors are then concatenated and input into the ELM classifier, which outputs the probability of the corresponding material category. The occlusion material type is determined based on the maximum probability value and mapped to the material peeling ease factor.

3. The method for removing obstructions from windshields and emergency vehicle control according to claim 1, characterized in that, In step S2, four planar capacitive electrostatic sensors are arranged along the four edges of the outer surface of the windshield. Each of the four planar capacitive electrostatic sensors independently detects the change in static charge in its area, calculates the equivalent charge in the area where each planar capacitive electrostatic sensor is located, and takes the largest absolute value of the four equivalent charge as the equivalent static charge of the entire shield, which is used to estimate the electrostatic adsorption force.

4. The method for removing obstructions and controlling emergency vehicle operation in response to windshield obstructions according to claim 1, characterized in that, The method for determining the feasibility of stripping in step S2 is as follows: ； in, Feasibility index for divestiture; Material peeling ease factor; : The preset maximum threshold for electrostatic adsorption force; The magnitude of the electrostatic attraction between the obstruction and the windshield. : Environmental wind field auxiliary factor, initially set to 0, and updated according to actual wind conditions; Non-electrostatic adhesion factor, based on the proportion of the area covered by the obstruction and correction for ambient humidity, with a value range of... ; All are weighting coefficients.

5. The method for removing obstructions from windshield obstructions and controlling vehicles in emergency situations according to claim 1, characterized in that, In step S3, the ion generator is positioned near the base of the rearview mirror inside the vehicle, with its ion release port facing the inner surface of the windshield and the release direction forming an angle of 30° to 45° with the windshield plane. When using the ion generator to eliminate static electricity, the total number of ions that the ion generator needs to release is first calculated based on the equivalent static charge of the entire obstruction, thereby determining the working time of the ion generator. Then, the ion generator is activated to release ions with opposite charge polarity to the obstruction onto the inner surface of the windshield, neutralizing the static charge on the outer surface.

6. The method for removing obstructions from windshield obstructions and controlling vehicle emergency response according to claim 5, characterized in that, In step S3, the stripping action mode is first determined based on the current environmental wind conditions. The stripping action mode is divided into three types according to tailwind, headwind, and crosswind. Then, based on real-time environmental wind information and vehicle status, the lateral inertial force on the obstruction under vehicle attitude control is calculated. Lift / shear force generated by airflow on obstructions The system measures the longitudinal inertial force on the obstruction and determines whether the current peeling action can be successfully performed. If it can, the peeling action is executed immediately. If it cannot, the attitude control parameters are adjusted within the safety constraints, and the system is recalculated and re-verified. If the parameters have been adjusted to the preset safety limit and the stripping conditions still cannot be met, the stripping attempt is deemed to have failed, the active stripping is abandoned, and the process proceeds to step S4 and the safe side-by-side procedure is executed.

7. The method for removing obstructions from windshield obstructions and controlling vehicles in emergency situations according to claim 6, characterized in that, In step S3, the post-peeling transmittance is first calculated based on the target transmittance and the initial target transmittance. The system measures the feedforward transmittance per second; simultaneously, it uses an in-vehicle camera to monitor the driver's pupil diameter in real time, and uses PID control to perform closed-loop correction of the feedforward transmittance, so that the driver's pupil diameter is eventually stabilized in a comfortable range and the transmittance reaches the target transmittance.

8. The method for removing obstructions from windshield obstructions and controlling vehicle emergency response according to claim 7, characterized in that, In step S4, the driver's facial stress score, physiological stress score, verbal stress score, and behavioral stress score are first obtained. Then, the four scores are weighted and fused to obtain the driver's emotional stress index. ,like If so, then complete control is granted; if Then relinquish control and initiate proactive appeasement; like Rapid acceleration is prohibited, and single steering angles are limited to >15°; if If the vehicle controller immediately takes over the throttle and brakes, the speed will be reduced to below 40 km / h with a deceleration of 0.2g.

9. The method for removing obstructions from windshield obstructions and controlling vehicle emergency response according to claim 8, characterized in that, In step S4, the RSU image must simultaneously meet the view consistency assessment and latency assessment. If the latency assessment is not met, a request to reduce the image quality is first sent to the current RSU. If the image quality is still not met after reducing the image quality, a compensation request is resent to other roadside units. If a satisfactory RSU image is still not obtained, the forward-looking camera image of the vehicle ahead is requested to be shared via V2V communication, and the view deviation and total latency are recalculated to assess availability. If a satisfactory compensation image is still not obtained, the system is downgraded to blind spot monitoring guidance mode.

10. The method for removing obstructions from windshield obstructions and controlling vehicle emergency control according to claim 9, characterized in that, In step S4, when guiding the vehicle to safely pull over, first determine the vehicle's current lane and position. Then, retrieve all parking areas within a specified range from the map database, calculate the comprehensive score of each candidate area, and select the candidate area with the highest comprehensive score as the target parking point. For vehicles that have acquired compensation images, guide the driver through the head-up display, voice prompts, and instrument panel auxiliary display until they reach the target parking area. If the current mode is blind spot monitoring guidance, guide the driver to perform the following operations only through voice prompts and instrument panel auxiliary display: First, maintain stable driving in the current lane and gradually decelerate; when the blind spot radar confirms that the right lane is safe, the driver is prompted to change lanes to the right; repeat this process until the vehicle enters the target parking area. If there are vehicles in the right lane that prevent the driver from changing lanes, the driver will be prompted to stay in the current lane and change lanes when conditions permit.

Images