A vehicle headlamp adaptive control method for a continuous tunnel group scenario

Abstract

The present application belongs to the field of control of lighting, and relates to a vehicle headlamp adaptive control method for continuous tunnel group scenes. Whether the adjacent tunnel is a continuous tunnel is determined to determine the current driving stage of the vehicle. Residual visual adaptation state characteristic quantity is constructed, and the required features are extracted. A residual visual adaptation state model is constructed, the residual visual adaptation state intensity is determined based on the obtained data, and when it is judged as a continuous tunnel, the residual visual adaptation state propagation and inheritance between tunnel groups are completed. The visual effective detection distance and the safe disposal distance are calculated, the visibility safety gap is determined, which is mapped as an equivalent lighting demand increment, the target value of each component of the headlamp control vector is calculated, and finally the final control vector is output based on the constraint condition. The method integrates the accumulation of the visual adaptation residual which has not been recovered after the exit of the last tunnel into the lighting control, and significantly improves the stability of the forward visibility and the driving safety under the condition of continuous entry and exit of the hole.

Description

Technical Field

[0001] This invention belongs to the field of lighting control, and relates to the control of vehicle lighting in tunnels, and particularly to an adaptive control method for vehicle headlights in continuous tunnel scenarios. Background Technology

[0002] In sections of highways with continuous tunnel clusters, multiple tunnels are distributed consecutively at short intervals along the route. Vehicles repeatedly experience rapid transitions between the natural light environment outside the tunnels and the artificial lighting environment inside. Compared to a single tunnel scenario, the driver's visual adaptation process in continuous tunnel cluster conditions exhibits more pronounced continuity, lag, and cumulative effects. After exiting one tunnel, the driver's visual system often has not yet completed its recovery and adaptation from the darker environment inside to the brighter environment outside, and the vehicle approaches and enters the next tunnel in a short period of time, thus facing another abrupt change in lighting environment from bright to dark or from dark to bright. This process can easily lead to problems such as the black hole effect, white hole effect, delayed recognition of objects ahead, short-term reduction in effective visibility distance, and increased handling fluctuations, thereby increasing the risk of rear-end collisions, lane departures, and other driving risks.

[0003] Existing technologies have conducted considerable research on tunnel visual environment and lighting control, resulting in several representative technical approaches. For example, Chinese patent CN106332345A discloses a "Smart Lighting System for Continuous Tunnel Groups," which can coordinate and adjust the lighting of tunnel fixtures and adjacent approach road lighting based on ambient brightness, traffic flow, and other information, making the transition in lighting brightness from the exit area of ​​one tunnel to the entrance area of ​​the next tunnel smoother. However, continuous tunnel group lighting systems primarily address the brightness, color temperature, and spatial transition relationships of roadside lighting fixtures, focusing on the continuity of external lighting output. It is important to clarify that a continuous transition in roadside lighting does not equate to a full recovery of the driver's visual state. When vehicles are traveling through continuous tunnel groups, what truly affects vehicle decision-making is not merely the smooth transition of external lighting, but more importantly, the residual visual adaptation load formed at the exit of the previous tunnel. Whether this residual state will continue to affect target recognition and forward visibility when entering the next tunnel is crucial. Therefore, this system is not suitable for vehicle headlight control.

[0004] Chinese patent CN117445794A discloses "a method, device, and storage medium for vehicle headlight control in a tunnel scenario." This method determines the tunnel entrance based on vehicle and tunnel entrance / exit image features, and then uses distance features to determine the timing for pre-activating headlights before entering or exiting the tunnel, thus reducing visual stimulation during a single entry or exit. However, this method treats each tunnel entrance or exit as a separate event, and its basic logic remains "identifying the current entry or exit scenario—triggering the headlight action based on current distance, vehicle speed, or brightness information." While this technology can achieve good pre-response in a single tunnel, its control mode is essentially memoryless, independent event processing—one event corresponds to one control, with no state transfer between events. Furthermore, this mode does not consider the residual visual adaptation after the previous tunnel has been traversed, and whether it needs to continue participating in headlight control at the entrance of the next tunnel. Therefore, in continuous tunnel scenarios, this type of technology typically only manifests as repeated triggering of multiple tunnel entrances, making it difficult to form a state transfer control with continuous dependency between adjacent tunnels.

[0005] Chinese patent CN121510434A discloses "a tunnel lighting control method considering driver visual adaptation". This method optimizes and adjusts tunnel lighting parameters based on relevant visual adaptation physiological indicators such as pupil changes and scanning speed. Although this method recognizes the important influence of the visual adaptation process on lighting control, it mainly establishes general visual adaptation evaluation models, prediction models or comfort optimization models, which are mainly used for lighting parameter optimization or visual state assessment. Such models usually lack direct coupling with the topology of continuous tunnel groups, the interval characteristics of adjacent tunnels and the vehicle operation process, and do not solve the following problems: (1) how to construct the visual adaptation residual state index at the exit of the previous tunnel when the recovery time between adjacent tunnels is insufficient; (2) how to integrate the cross-tunnel section and continue to enter the control decision at the next tunnel entrance. It can be seen that the existing visual adaptation model can explain that "visual state is worth considering", but it does not form a complete control chain from the residual state of the previous exit to the event triggering of the next entrance, and finally realize the headlight output of the vehicle.

[0006] In summary, although existing technologies have made some progress in areas such as lighting transition in continuous tunnel groups, forward-looking control of headlights in single tunnels, and visual adaptation status assessment, the following shortcomings still exist: (1) There is a lack of a continuous tunnel identification mechanism based on an effective recovery window between adjacent tunnels, making it difficult to determine which adjacent tunnels belong to a continuously triggered scenario in terms of visual recovery; (2) There is a lack of vehicle-side control state variables that can characterize the residual visual adaptation state that has not yet recovered after the exit of the previous tunnel; (3) There is a lack of an event-triggered inheritance mechanism that reintroduces the residual state of the previous tunnel exit into the current headlight control decision when the next tunnel entrance event occurs. Therefore, it is necessary to provide a vehicle headlight adaptive control method for continuous tunnel group scenarios to improve the stability of forward visibility and driving safety under continuous tunnel entry and exit conditions. Summary of the Invention

[0007] In view of the shortcomings and deficiencies of the existing technology, the purpose of this invention is to provide an adaptive control method for vehicle headlights in continuous tunnel scenarios. This method not only focuses on hysteresis, but also incorporates the accumulation of visual adaptation residues that have not yet recovered after the exit of the previous tunnel into the lighting control, which significantly improves the stability of forward visibility and driving safety under continuous tunnel entry and exit conditions.

[0008] To achieve the above objectives, the present invention adopts the following technical solution: An adaptive control method for vehicle headlights in a continuous tunnel group scenario, the method includes the following steps: Step S1. Obtain the topology of the tunnel group, determine whether adjacent tunnels are continuous based on the visual effective recovery window, and determine the current driving stage of the vehicle based on the distance-time quantity and brightness change trend; Step S2. Construct residual visual adaptation state features and extract vehicle motion and road geometry features, glare-related target features; the residual visual adaptation state features include lighting environment features, occlusion features, and visibility features; Step S3. Construct a residual visual adaptation state model, determine the residual visual adaptation state intensity based on the data obtained in steps S1 and S2; and when it is determined to be a continuous tunnel in step S1, complete the propagation and inheritance of residual visual adaptation state between tunnel groups; at the same time, correct the residual visual adaptation state intensity based on vehicle speed to obtain the equivalent residual visual adaptation state corrected by vehicle speed. Step S4. Construct an effective illuminance model of the headlight output and the target area. Calculate the effective visual detection distance based on illuminance, visibility, occlusion, and equivalent residual visual adaptation state. Correct the effective detection distance based on road curvature to obtain a geometrically corrected effective detection distance. Then calculate the safe handling distance based on vehicle speed and equivalent residual visual adaptation state. Finally, establish the constraints for headlight control in the tunnel group. Step S5. Calculate the visibility safety gap based on the effective detection distance and safe handling distance with geometric correction, and map it to the equivalent lighting demand increment. Then calculate the target value of each component of the headlight control vector, and finally output the final control vector based on the constraints.

[0009] In a preferred embodiment of the present invention, step S1 obtains the arc length coordinates corresponding to each tunnel entrance and exit, and calculates the arc length coordinates of the vehicle based on the positional relationship between the vehicle and the previous tunnel exit and the next tunnel entrance. Location in the section between tunnels Effective contribution weight of visual restoration Thus determining the first Effective visual recovery time between tunnel sections The expression is: ; in, and The vehicles traveled to the first The tunnel exit and the first The moment of the tunnel entrance To obtain the average speed of the vehicle 200m before the exit of the kth tunnel, as the predicted speed of the road section between tunnels; When the Effective visual recovery time between tunnel sections Less than the minimum recovery time threshold At that time, the first The tunnel and the first The tunnels form a continuous tunnel pair in the sense of visual restoration.

[0010] As a preferred embodiment of the present invention, in step S1, the vehicle relative to the first Signed distance between the tunnel entrance and exit Then calculate the vehicle relative to the first Time distance between tunnel entrance and exit Then based on the real-time ambient brightness sequence Calculate the rate of change of brightness Finally, based on the acquired data, the current driving stage of the vehicle is determined. The driving stage is divided into the entrance pre-adaptation stage, the tunnel stabilization stage, the exit near transition stage, and the exit slow adaptation stage.

[0011] As a preferred embodiment of the present invention, the light environment characteristic in step S2 is a light intensity abrupt change index, expressed as: ; in, Indicates time The intensity of sudden changes in light intensity; This represents the difference between the current brightness and the reference brightness, i.e., the sudden change in brightness due to the static difference between light and dark. Indicates the rate of change in brightness; The reference luminance constant is pre-calibrated. and , representing the weights of static brightness difference and dynamic change term in the comprehensive index, respectively; The occlusion feature is the occlusion rate. The visibility feature is visibility. ; The glare-related target features are glare risk indicators. The expression is: ; in, , These represent the relative distance and relative azimuth angle of oncoming vehicles, respectively. , These represent the relative distance and relative azimuth angle of the vehicle in front, respectively. Angle-sensitive weighting function, The relative azimuth angle. These are the weighting coefficients. To prevent positive numbers with a denominator of zero.

[0012] As a preferred embodiment of the present invention, the expression for the residual visual adaptation state model in step S3 is: ; in, Indicates time The intensity of residual visual adaptation state, This indicates that the visual adaptation is sufficient. This indicates that the debt has reached saturation. For saturated projection operators, For a moment The residual visual adaptation state increment, i.e., the debt injection increment. For a moment The amount of residual visual adaptation state attenuation.

[0013] As a preferred embodiment of the present invention, in step S3, during state propagation and inheritance, the visual deficit that has not yet faded from the previous tunnel exit is explicitly carried into the control time of the next tunnel entrance, and the initial adaptive state at the entrance is inherited and updated: ; in, Indicates the number of inheritance updates. The residual visual adaptation state at the tunnel entrance, the vehicle arrives at the first The time of the tunnel entrance is , This means that the effective visual recovery time between tunnels is less than the minimum recovery time threshold, indicating that the visual residue of the preceding tunnel still has an impact on the subsequent tunnel. This indicates that the residual visual adaptation between tunnel sections has been fully restored and is unaffected by the preceding tunnel. No. The equivalent residual visual adaptation state inside the tunnel is as follows:

[0014] in, The equivalent residual visual adaptation state after vehicle speed correction. For the vehicle's speed, The value is taken as a reference speed.

[0015] As a preferred embodiment of the present invention, the expression for the headlight output and the effective illuminance model of the target area in step S4 is as follows: ; in, Indicates time Effective illuminance in the target area; This represents the base illuminance conversion factor after considering the combined optical characteristics and installation parameters of the luminaire. Indicates the basic lighting intensity; This represents the illuminance gain coefficient corresponding to the specular enhancement component. Indicates the highlight enhancement component. This represents the combined effect of each zone of the matrix headlight on the target area, i.e., the equivalent zone contribution. Indicates the curvature of the road; Indicates the road slope; and These represent the attenuation coefficients of road curvature and slope on the effectiveness of lighting projection, respectively.

[0016] As a preferred embodiment of the present invention, the effective visual detection distance in step S4 To reliably identify the furthest distance of a target ahead under the current complex environment and driver status, the expression is: ; In the formula, The reference distance calibration value, , , , All are functions, representing the effects of illuminance, visibility, occlusion, and equivalent residual visual adaptation state on detection distance, respectively. ;in, It is the illuminance saturation characteristic constant of the human eye's photosensitive properties. , For visibility degradation sensitivity index, For reference visibility; ;in, This is the magnification factor for occlusion. ;in, This represents the driver's visual recovery time constant.

[0017] As a preferred embodiment of the present invention, the safe handling distance in step S4 The expression is: ; in, For vehicles at any time The speed of travel; The baseline reaction time; The sensitivity coefficient of residual visual adaptation to reaction time amplification; For a moment The equivalent residual visual adaptation state of vehicle speed correction; For a moment The road surface adhesion coefficient; is the gravitational acceleration constant.

[0018] As a preferred embodiment of the present invention, the constraint condition for controlling the tunnel group headlights in step S4 is: at any time, the geometrically corrected effective visual detection distance must not be less than the safe handling distance; and a dynamic allowable upper limit is set for the highlight enhancement component. ;in, This indicates the maximum permissible value of the highlight enhancement component under conditions where there is no risk of glare. Indicates the glare risk suppression coefficient; To assess glare risk, an upper limit is set for the rate of change of the control vector within a single sampling period, the upper limit being based on the current vehicle driving stage. With continuous tunnel status signs Adaptive adjustment.

[0019] As a preferred embodiment of the present invention, the target value of the basic lighting component in step S5 The expression is: ; in, This represents an interval projection operator that limits the target value of basic lighting to an allowable range. Maximum basic lighting intensity; This represents the reference value for basic lighting under normal driving conditions outside the tunnel; Reference values ​​for basic lighting during the stable phase inside the cave; This represents the rate of decline during the export adjustment phase. The entrance look-ahead weight is calculated based on the time distance from the vehicle to the current tunnel entrance. The exit drop weight is calculated based on the time distance from the vehicle to the current tunnel exit. For visual persistence compensation weights, This is the headlight control vector for the vehicle at the previous moment. This is the equivalent increase in lighting demand; Target value of specular enhancement component The expression is: ; in, This indicates the gain coefficient required to enhance highlights; This represents the dynamic allowable upper limit determined by the glare risk constraint, i.e., the dynamic allowable upper limit of the specular enhancement component; Matrix headlights Each partition at time Target control quantity The expression is: ; in, and They represent the first Minimum and maximum values ​​of each zone control quantity; Indicates the first Reference light distribution values ​​for each zone under normal operating conditions; Indicates the road curvature with respect to the first Gain coefficient of target light distribution in each zone; Indicates the yaw rate with respect to the first Gain coefficient of target light distribution in each zone; Indicates the risk of glare to the first The suppression coefficient of target light distribution in each zone; Indicates the curvature of the road; Indicates the vehicle's yaw rate; This indicates a risk indicator of glare.

[0020] As a further preferred embodiment of the present invention, in step S3, the time... Residual visual adaptation state increment The expression is: ; in, The injection gate function takes a value of 1 during the inlet pre-adaptation and outlet soft-adaptation phases, and a value of 0 otherwise. Control the injection amplitude, The adjustment parameter is used to regulate the rate at which the injection amount increases with the mutation intensity; For a moment The intensity of sudden changes in light intensity, The visibility degradation factor is calculated based on occlusion features and visibility features. time Residual visual adaptation state attenuation The expression is: ; in, To restore the gate function, the value is 1 during the stable section inside the tunnel and the recovery section between tunnels, and 0 otherwise; The recovery time constant is a time-varying value, calculated based on occlusion and visibility features; This represents the sampling time interval.

[0021] Advantages and beneficial effects of the present invention: (1) In a continuous tunnel group section, vehicles experience the switching between the natural light environment outside the tunnel and the artificial lighting environment inside the tunnel in a short period of time. The driver's visual adaptation process has obvious lag and accumulation. The adaptive control method of vehicle headlights in continuous tunnel groups based on the inheritance of residual visual adaptation state provided by the present invention not only focuses on the lag, but also incorporates the accumulation of visual adaptation residues that have not yet recovered after the exit of the previous tunnel into the lighting control, which significantly improves the stability of forward visibility and driving safety under continuous entry and exit conditions.

[0022] (2) This invention proposes a continuous tunnel identification method based on the effective recovery window of adjacent tunnels, which is used to determine whether adjacent tunnels constitute a continuous triggering scene in the sense of visual recovery, so as to facilitate subsequent separate control for different scenes.

[0023] (3) This invention proposes a method for describing and updating residual visual adaptation state, which abstracts the visual load that has not yet recovered after the exit of the previous tunnel into a calculable and updatable control state quantity, and injects, recovers and attenuates it according to different stage rules in the entrance pre-adaptation section, the tunnel stabilization section, the exit slow adaptation section and the inter-tunnel recovery section, so as to reflect the dynamic evolution process of the driver's visual adaptation state under the condition of continuous tunnel group.

[0024] (4) This invention proposes a dynamic evolution mechanism for residual visual adaptation state that is adaptive during driving phases. By forcibly associating the visual adaptation process with the specific driving phase of the vehicle, it accurately characterizes the differences in the injection and decay rates of residual visual state in each segment, thereby improving the spatiotemporal accuracy of driver visual load state estimation.

[0025] (5) This invention proposes an adaptive adjustment method for visual recovery rate in occlusion and low visibility environments. By sensing the degree of environmental attenuation and dynamically correcting the fading rate parameter of the visual residual state, the control logic can predict the delay effect of actual visual recovery, thereby triggering lighting compensation or extending the high illumination maintenance time in advance to prevent safety misjudgment caused by recovery lag.

[0026] (6) This invention proposes a nonlinear equivalent correction method for residual visual adaptation state based on vehicle speed. By constructing a quantitative mapping relationship between vehicle speed and visual residual risk, the compression effect of high-speed conditions on effective recovery time is transformed into an equivalent visual residual increment, ensuring that the headlight control intensity and the real-time operating risk of the vehicle remain dynamically consistent.

[0027] (7) The present invention proposes a headlight control method that inherits and updates the residual visual adaptation state of the previous tunnel exit based on the continuous determination result of the adjacent tunnel when the next tunnel entrance event occurs, so that the headlight control is transformed from a single event response to a cross-tunnel continuous state control with a previous-to-next dependency relationship, thereby improving the continuity, foresight and pertinence of headlight regulation in continuous tunnel group scenarios.

[0028] (8) This invention proposes an effective illuminance real-time compensation mechanism that takes into account road geometric attenuation. By calculating the geometric attenuation of the effective coverage area of ​​the headlights due to road curvature and slope, the output intensity of the headlights is dynamically compensated to maintain the stability of the visual effective detection distance under complex linear conditions and eliminate the transient lighting blind spot caused by road geometric changes.

[0029] (9) This invention proposes a dynamic lighting demand mapping strategy based on "visibility safety gap". By constructing a gap quantification model that integrates vehicle speed, road surface adhesion, visibility, occlusion and residual visual adaptation state, the minimum lighting compensation intensity required is deduced from the real-time calculated safety gap value, so as to achieve the optimal balance control between safety margin and energy consumption glare. Attached Figure Description

[0030] The present invention will be described and illustrated in detail below with reference to the accompanying drawings and through a detailed description of the embodiments.

[0031] Figure 1 A flowchart of an adaptive control method for vehicle headlights in a continuous tunnel group scenario provided by the present invention; Figure 2 This is a schematic diagram showing the division of the tunnel into different stages. Detailed Implementation

[0032] The present invention will be further described in detail below with reference to examples and accompanying drawings, but the embodiments of the present invention are not limited thereto.

[0033] like Figure 1 , Figure 2 As shown in the figure, this embodiment provides an adaptive control method for vehicle headlights in a continuous tunnel group scenario. The method includes the following steps: Step S1. Tunnel group scene recognition and driving stage division: Step S1.1. Obtaining and representing the tunnel group topology: A tunnel group consists of multiple tunnels, and the set of tunnels is represented as follows: .in, The total number of tunnels, Indicates the first The tunnel entity.

[0034] To standardize the representation of tunnel entrances and exits on roads, the arc length coordinate of the road centerline is introduced, which represents the position of the tunnel entrance on the road centerline, and the first... The arc length coordinates corresponding to the tunnel entrance are The arc length coordinates corresponding to the exit are: ,and Therefore, the first... The tunnel is [length] Distance between adjacent tunnels .

[0035] Step S1.2. Determination of continuous tunnels based on the effective recovery window: In existing technologies, the identification of continuous tunnel groups is mostly based on the geometric distance between tunnels or short intervals in a general sense, which is difficult to directly reflect whether the driver has sufficient visual recovery conditions between adjacent tunnels. Therefore, this invention no longer uses a fixed length deduction method between tunnel sections to calculate the effective recovery distance, but instead uses the first... The tunnel exit and the first The road section between the tunnel entrances is regarded as a continuous distribution interval of recovery effectiveness, and the contribution of different locations within the interval to visual recovery is weighted and integrated to obtain an effective recovery amount that is more in line with the driver's visual adaptation mechanism.

[0036] In this invention, the vehicle's online position is obtained by fusing multi-source positioning data, including GNSS (Global Navigation Satellite System), IMU (Inertial Measurement Unit), and wheel speed-odometer data. Furthermore, forward-looking sensors (such as cameras, millimeter-wave radar, or lidar) are used to perform online calibration of redundant detection events at the tunnel entrance to correct positioning drift and ensure the accuracy of distance calculation. The vehicle's position in the road centerline arc-length coordinate system is represented as follows: .

[0037] Considering that near the exit of the previous tunnel, the driver is still under the influence of exit adaptation, resulting in lower visual recovery effectiveness; and near the entrance of the next tunnel, the driver gradually enters the entry pre-adaptation state, further reducing visual recovery effectiveness, the exit-side recovery activation function and the entrance-side recovery inhibition function are defined as follows:

[0038]

[0039] in, and These are the recovery impact scale parameters for the exit and inlet sides, respectively, and both are positive. Therefore, the first... Location in the section between tunnels The effective contribution weight of visual restoration is:

[0040] Based on this, the first The effective visual recovery time for road sections between tunnels is defined by a clear upper and lower time limit:

[0041] in, and The vehicles traveled to the first The tunnel exit and the first The moment of the tunnel entrance The average speed of the vehicle 200m before the exit of the kth tunnel is taken as the predicted speed for the tunnel section.

[0042] Definition of the first The criteria for determining consecutive tunnels between adjacent tunnels are:

[0043] Among them, the minimum recovery time threshold is , This is an indicator function that takes the value 1 when the condition inside the parentheses is true, and 0 otherwise. Indicates the first The tunnel and the first The tunnels constitute a continuous tunnel pair in the sense of visual restoration; This indicates that there is a relatively sufficient recovery window between the two, and they do not constitute a continuous tunnel relationship that requires cross-tunnel inheritance control.

[0044] Step S1.3. Division of tunnel travel stages: Based on the entrance and exit locations of the tunnel, the vehicle's position relative to the first... The signed distance between the tunnel entrance and exit is:

[0045]

[0046] in, Indicates the first The location of the tunnel entrance in the road centerline arc length coordinate system. Indicates the first The location of the tunnel exit in the road centerline arc length coordinate system. When At that time, it indicates that the vehicle has not yet arrived at the designated location. Tunnel entrance; when At that time, it indicates that the vehicle has entered the [stage / phase]. A tunnel. When At that time, it indicates that the vehicle has not yet left the first... Tunnel exit; when At that time, it indicates that the vehicle has left the first... The tunnel leads to the section of road outside the tunnel.

[0047] To translate spatial proximity into a controllable time scale, vehicle longitudinal velocity (driving speed) is introduced. With minimum speed threshold Define the vehicle relative to the first The time distance between the tunnel entrance and exit is:

[0048]

[0049] in, Indicates the vehicle relative to the first The time distance to the tunnel entrance is represented as the estimated time required to reach the entrance before entering the tunnel, and as the time already elapsed after entering the tunnel. Indicates the vehicle relative to the first The time distance to the tunnel exit is indicated by the estimated time required to reach the exit during the tunnel phase and by the time elapsed since leaving the exit during the post-exit phase.

[0050] Considering that relying solely on positioning distance may introduce errors due to signal jitter, this invention also incorporates the trend of light environment changes as an auxiliary criterion. The system performs online estimation of the ambient brightness in front of the vehicle, obtaining a real-time ambient brightness sequence. .

[0051] Specifically, in this embodiment, the ambient brightness sequence The brightness channel can be obtained from grayscale statistics of the forward-facing camera image, HDR brightness estimation, or light sensor measurement. The effective forward field of view (ROI) of the vehicle is selected from the forward-facing camera image (excluding high-brightness areas of the vehicle's headlights, sky areas, and instrument panel reflection areas), and the brightness channel is extracted. Alternatively, linear luminance can be used, and median robust statistics can be employed to estimate ambient luminance.

[0052] set up Given a sampling time interval, the rate of change of brightness is defined as:

[0053] Based on the combined distance-time and brightness variation trends, the system divides the driving process within the tunnel complex into four stages with clear physical meaning, and uses tags to identify the stage when the vehicle is in tunnel K. .

[0054]

[0055] in, It represents the logical "AND". Indicates the inlet pre-adaptation section. Indicates the stable section inside the tunnel. This indicates that the export is approaching the transition zone. This indicates a gradual adaptation period for exports.

[0056] In this embodiment, when a vehicle meets the stage conditions based on location and time, but the brightness change rate does not reach the corresponding threshold, the aforementioned location-time determination result is still used as the stage label, and the brightness change rate is only used as the basis for stage confidence correction.

[0057] Step S2. Construction of residual visual adaptation state features: Step S2.1. Light environment feature extraction: The system utilizes stage labels The luminance sequence is segmented to construct reference luminances representing the external and internal lighting environments of the cave, respectively. The external and internal stage indicators are defined as follows:

[0058] in, This is an indicator function; it takes the value 1 when the condition within the parentheses is true, and 0 otherwise. Based on the above indicator, the system adopts a phase-driven reference update—a non-phase hold strategy—to update the reference brightness, ensuring the accuracy of the external reference brightness. Compared with the reference brightness inside the cave Strictly corresponds to the driving stage:

[0059]

[0060] in, To update the coefficients, balancing real-time performance and noise resistance, The sampling period.

[0061] When the vehicle is in the recovery phase outside or between tunnels, only the reference brightness outside the tunnel is updated. When the vehicle is in the relevant stage inside the tunnel, only the reference brightness inside the tunnel is updated. The remaining time frame maintains the reference value from the previous moment, thus avoiding mutual contamination of brightness references inside and outside the tunnel and enhancing the interpretability of subsequent features. When the vehicle first enters the tunnel complex at a preset distance, the reference value is taken as... Frame hole out-of-hole ROI brightness mean initialization After entering the stable section inside the tunnel, take continuous... Frame hole ROI brightness mean initialization .

[0062] Based on the reference brightness inside and outside the cave, the brightness abrupt change quantity, which characterizes the static difference between light and dark, is defined as:

[0063] in, It directly reflects the magnitude of the difference in brightness between the inside and outside of the cave. The larger the absolute value, the stronger the static contrast between light and dark that the visual adaptation needs to overcome.

[0064] By combining static differences and dynamic trends, an index for sudden changes in light intensity is constructed:

[0065] in, Indicates time The intensity of sudden changes in light intensity; This represents the difference between the current brightness and the reference brightness, i.e., the sudden change in brightness due to the static difference between light and dark. Indicates the rate of change in brightness; The pre-calibrated reference brightness constant can be determined based on the statistics of typical tunnel brightness difference samples. Under normal weather conditions, the statistical median is taken as 3000 lux. and , representing the weights of static brightness difference and dynamic change term in the comprehensive index, respectively.

[0066] Step S2.2. Extraction of occlusion features and visibility features: Occlusion features are used to quantify the degree to which effective visual information is blocked in the forward field of view. The occlusion rate is defined as:

[0067] in, Indicates time The area of ​​pixels occluded by obstacles within the region of interest of the road in the downward direction; Indicates time The pixel area corresponding to the region of interest in the forward road. A higher occlusion rate indicates less effective visible information in the forward road.

[0068] Specifically, in this embodiment, to obtain the pixel area corresponding to the region of interest of the forward road, a mask for the region of interest of the forward road is first constructed. When pixels Located at time When within the area of ​​interest of the forward road, take Otherwise, take The forward road region of interest (ROI) can be constructed as a lane polygon by combining lane line detection results with camera calibration parameters, and then truncated along the road depth to a preset distance range. When lane line markings are unclear, it can be degenerated into a fixed trapezoidal ROI, or obtained by projecting lane geometry information from a high-precision map. Therefore, the pixel area corresponding to the forward road ROI is represented as:

[0069] To obtain the shading area The system is at all times Obstacle identification and distance constraint filtering are performed on the forward sensing data. Specifically, firstly, the distance to obstacles is directly obtained using LiDAR or millimeter-wave radar to form a depth map. The ranging results are projected onto the image coordinate system to assist in the verification or compensation of the depth information. Simultaneously, the system employs a target detection network to identify dynamic or static obstacles in the forward scene and constructs an obstacle mask. When pixels When the target area includes vehicles, pedestrians, traffic facilities, construction obstacles, or other objects that may obstruct effective visibility of the road ahead, take... Otherwise, take .

[0070] To avoid including distant, irrelevant targets or targets not related to this lane in the occlusion area, a depth-constrained mask is further constructed. When pixels The corresponding depth satisfies At that time, take Otherwise, take .in, and These represent the lower and upper limits of the preset effective forward distance range, respectively, which can be pre-calibrated and determined based on vehicle speed, the installation position of the forward-looking sensor, the field of view, and the headlight control distance. Specifically, Desirable This is used to exclude irrelevant near-field areas around the front of the vehicle; Desirable This is used to cover the main range of forward visibility assessment and headlight control in tunnel complex scenarios. Preferably, it can be used for low-speed or short tunnel sections. , High-speed or long tunnel sections can be taken , .

[0071] Based on the forward road region of interest mask, obstacle mask, and depth constraint mask, the occlusion area is defined as:

[0072] The occlusion area is only included if a pixel is simultaneously located within the forward road region of interest, belongs to the obstacle region, and meets the effective forward distance constraint. .

[0073] Image from the front-facing camera Select regions of interest (ROIs) along the lane direction, both near (5m for low-speed or short tunnel sections, 10m for high-speed or long tunnel sections) and far (80m for low-speed or short tunnel sections, 120m for high-speed or long tunnel sections). Calculate the brightness contrast between the lane line pixels and the road surface background pixels within each of the two regions.

[0074] in, , The first The average brightness of lane lines and background within each area , The brightness contrast ratios for near and far ROIs are shown respectively. To prevent small quantities from being zero, 10 is usually used. -3 .

[0075] Assuming uniform atmospheric attenuation, brightness and contrast decrease exponentially with distance. The attenuation coefficient can then be estimated using the effective viewing distance contrast ratio. :

[0076] in, , These are the preset lower and upper bounds of the effective forward distance range. To suppress noise, a first-order low-pass filter is performed.

[0077] visibility Defined as: brightness contrast decreasing to the perception threshold. The distance (usually taken as 0.05) is expressed as:

[0078] Step S2.3. Vehicle motion and road geometry feature extraction: To characterize the impact of longitudinal operating conditions on safe handling distance and lighting requirements, the longitudinal motion state of the vehicle is measured through vehicle speed. Characterization; Lateral maneuvering state via yaw rate Reflection. Road geometry features include road curvature. and road slope It is used to correct the effective line-of-sight model and adjust the boundary conditions for lighting control.

[0079] Step S2.4. Dazzle-related target feature extraction: To establish effective anti-glare constraints, the system identifies and quantifies the glare risk caused by surrounding vehicles. , These represent the relative distance and relative azimuth angle of oncoming vehicles, respectively. , These represent the relative distance and relative azimuth angle of the vehicle in front, respectively. These relative quantities can be obtained through camera target detection and tracking or radar ranging. Considering that the glare effect is related to both relative distance and azimuth angle, an angle-sensitive weighting function is constructed:

[0080] in, =3° is the threshold for the strong glare-sensitive angle region. =10° is the threshold for the glare reduction angle. This refers to the relative azimuth angle.

[0081] To characterize the intensity of glare risk and facilitate subsequent adjustment of constraint weights, the glare risk index is defined as follows:

[0082] in, As a glare risk indicator, These are the weighting coefficients. To prevent tiny positive numbers with a denominator of zero.

[0083] Step S3. Residual visual adaptation state modeling and tunnel group cumulative propagation update: Step S3.1. Definition of residual visual adaptation state and discrete update mechanism: During continuous entry and exit from tunnels, the driver's visual system does not adapt to sudden changes in the external light environment instantaneously. When external brightness, visibility, and occlusion conditions change rapidly, a temporary deviation occurs between the driver's current visual adaptation level and the ideal stable adaptation level. This deviation leads to a decrease in the ability to identify targets ahead, a shortened effective viewing distance, and a lag in reaction time. To describe this adaptation lag effect caused by continuous changes in brightness and darkness, which accumulates, recovers, and is transmitted across tunnels over time, this invention defines the residual visual adaptation state. The residual visual adaptation state is a normalized scalar. . Indicates time The intensity of residual visual adaptation state. This indicates that visual adaptation is adequate and there are no significant shortcomings. This indicates that the visual system is under immense pressure due to saturation. This state is independent of individual physiological sensors and is calculated entirely from observable external features of the vehicle, such as the lighting environment, visibility, and driving stage. Specifically, the dynamic evolution of this state is described using a discrete-time model, the core of which consists of two opposing processes: "injection" and "recovery."

[0084] In the formula, For saturated projection operators, ensure that the state value always falls within the range of saturated projection operators. Interval. For a moment The residual visual adaptation state increment, i.e., the debt injection increment. For a moment The amount of residual visual adaptation state attenuation.

[0085] Step S3.2. Construction of a scenario-based debt injection and recovery model: The process of injecting and restoring debts is comprehensively regulated by the driving stage, the intensity of sudden changes in light intensity, and environmental visibility conditions.

[0086] The debt injection mechanism mainly occurs in the ingress pre-adaptation phase where there are drastic changes in illumination. ) and export adaptation phase ( The injection gate function is:

[0087] The injection intensity is determined by the light mutation intensity index extracted from S2. Driven by the fact that harsh visual environments (low visibility, strong occlusion) amplify the impact of sudden changes in illumination on the visual system, a visibility degradation factor is introduced for correction:

[0088] in, For reference visibility, it is obtained from the quantiles of the set of unobstructed, clear samples; For visibility degradation sensitivity index, The occlusion amplification factor can be determined by least squares or grid search using the calibration data of abrupt events at the entrance and exit, and stored and retrieved as a parameter table for the tunnel group.

[0089] make The equivalent stimulus intensity characterizes the sudden change in illumination. The mutation window for each entry or exit event is extracted based on calibration data. Calculate the peak value within the event. and the statistics of its set as a reference scale. .

[0090] Taking all the above factors into account, the incremental debt injection is calculated using a non-linear saturation function to avoid extreme values:

[0091] in, Control the injection amplitude, To adjust the parameters, by adjusting the typical mutation strength The corresponding injection amount has reached the upper limit. The dosage is calibrated to control sensitivity to stimuli, which is used to adjust the rate at which the injection dose increases with the intensity of the mutation. Indicates time The intensity of sudden changes in light intensity, Indicates the visibility degradation factor. The function ensures that there is an upper limit to the amount injected.

[0092] The debt recovery mechanism process mainly occurs in the stable section inside the tunnel. ) and inter-tunnel recovery section ( The recovery gate function is:

[0093] The recovery speed of visual adaptation is not constant but is affected by environmental conditions. We define a time-varying recovery time constant:

[0094] in, As the baseline recovery time, and These represent the mitigation coefficients of recovery due to obstruction and low visibility, respectively. This means that the worse the visual field conditions, the slower the visual system recovers from the adaptation lag.

[0095] Based on this time-varying constant, the recovery decay adopts an exponential decay form:

[0096] This formula indicates that the recovery amount is proportional to the current level of debt, and the recovery process is smooth and continuous.

[0097] Step S3.3. State propagation and inheritance between tunnel groups: When the determination is made in step S1 The tunnel and the first The tunnels constitute a continuous tunnel pair in the sense of visual restoration, that is At that time, it indicates that the vehicle has left the first... After passing through the tunnel, the effective recovery window obtained by the driver in the tunnel section is insufficient, at this time the first The residual visual adaptation formed at the tunnel exit should not be considered as having completely faded, but should be considered in the [missing information]. They continued to maintain control at the tunnel entrance.

[0098] In this embodiment, the inheritance update does not control the two tunnel events independently. Instead, when it is confirmed that the effective recovery window for the visual between tunnels is insufficient, the visual deficit that has not yet faded at the exit of the previous tunnel is explicitly brought into the control time of the entrance of the next tunnel, thereby avoiding the continuous tunnel scene being mistakenly treated as multiple isolated single tunnel events.

[0099] Assume the vehicle arrives at the... The time of the tunnel entrance is At this time, the real-time residual visual adaptation state The value is This value is obtained by continuously restoring and evolving the road through the tunnel sections, starting from the previous tunnel exit.

[0100] According to the continuous tunnel determination flag defined in step S1 ,right Update the settings to reflect the initial state upon entering the new tunnel:

[0101] in, Indicates the number of inheritance updates. The residual visual adaptation state at the tunnel entrance Pick This means that the effective visual recovery time between tunnels is less than the preset threshold, indicating that the visual residue of the preceding tunnel still has an impact on the subsequent tunnel. Pick This indicates that the residual visual adaptation between tunnel sections has been fully restored and is unaffected by the preceding tunnel.

[0102] Step S3.4. Equivalent residual visual adaptation state of vehicle speed correction: Considering that vehicle speed directly affects safe following distance and driver reaction time, the same visual deficit poses a greater risk at high speeds. Therefore, The basic residual state control decision for the tunnel is determined by the equivalent debt state upon initial entry into the tunnel, therefore... The equivalent residual visual adaptation state inside the tunnel is as follows:

[0103] in, The equivalent residual visual adaptation state after vehicle speed correction. For the vehicle's speed, The reference speed is 16.67 m / s.

[0104] Step S4. Visibility constraint index construction and control target determination: Step S4.1. Headlight output and effective illuminance model: Define time The headlight control vector is .

[0105] in, Indicates the basic lighting intensity; Indicates the highlight enhancement component; Represents the matrix headlight. Each partition at time Light distribution control amount.

[0106] In this embodiment, the combined effect of each zone of the matrix headlight on the target area is represented as the equivalent zone contribution. The expression is:

[0107] in, This indicates the total number of zones in a matrix headlight system; For the first Electro-optical conversion coefficient of the partition Indicates the first Each partition at time The amount of light distribution control; Indicates the first The contribution weight of each zone to the effective illuminance of the target area can be determined by the luminaire configuration calibration results, installation parameters, and the location of the target area.

[0108] The actual lighting effect of the road ahead depends not only on the output of the luminaires but is also significantly affected by the road's geometry. Therefore, the effective illuminance of the target area is defined as... Therefore, the mapping relationship from the control vector to the effective illuminance of the target area can be expressed as integrating basic lighting, highlight enhancement, and zoned lighting; specifically, the effective illuminance of the target area is:

[0109] in, Indicates time Effective illuminance in the target area; The base illuminance conversion factor represents the overall optical characteristics of the luminaire and its installation parameters. This represents the illuminance gain coefficient corresponding to the specular enhancement component; Indicates the curvature of the road; Indicates the road slope; and These represent the attenuation coefficients of road curvature and slope on the effectiveness of lighting projection, respectively.

[0110] In this embodiment, the effective illuminance of the target area is jointly determined by the base lighting intensity, the highlight enhancement component, and the zonal beam distribution of the matrix headlights, and is also affected by the correction of the lighting projection geometry by the road curvature and slope. When the absolute value of the road curvature or the absolute value of the slope increases, the effective coverage of the headlight beam in the target area decreases, resulting in a reduction in effective illuminance.

[0111] Step S4.2. Composite modeling of visually effective detection range: Constructing an "effective visual detection range" It quantifies the farthest distance at which a target ahead can be reliably identified under the current complex environment and driver status.

[0112]

[0113] In the formula, As the reference distance calibration value, each The functions characterize the effects of illumination, visibility, occlusion, and equivalent residual visual adaptation on detection distance, respectively. Specifically, the following functions can be used: The effect of illuminance is manifested in the saturation gain characteristics: ;in, It is the illuminance saturation characteristic constant of the human eye's photosensitive properties.

[0114] The visibility effect is represented by a normalized power function: , This is a sensitivity index for visibility degradation.

[0115] Line-of-sight occlusion affects modeling: ;in, This is the magnification factor for occlusion.

[0116] The equivalent residual visual adaptation state after vehicle speed correction is expressed by exponential decay: ;in, This represents the driver's visual recovery time constant.

[0117] Step S4.3. Geometric Correction and Safety Requirement Distance: On curved roads, there is a difference between the detection distance in the straight direction and the actual usable road arc length. Furthermore, there is a deviation between the road centerline direction and the headlight projection direction under curved conditions, resulting in a lower actual usable detection distance compared to straight-line conditions.

[0118] To reflect the geometric limitation effect of road curvature on the actual usable detection distance, a curvature correction factor is introduced:

[0119] in, Indicates time The curvature correction factor; Indicates the curvature influence coefficient; Indicates the curvature of the road; Indicates time The effective visual detection range. When the road is straight. ,at this time When the road curvature increases, Gradually decrease.

[0120] The effective detection range after considering the influence of road curvature is:

[0121] in, Indicates time Effective detection range after road geometry correction.

[0122] In contrast, the minimum longitudinal distance required for a vehicle to meet safety handling requirements is the safety handling distance. It consists of reaction distance and braking distance:

[0123] in, For vehicles at any time The speed of travel; The baseline reaction time; The sensitivity coefficient of residual visual adaptation to reaction time amplification is determined through calibration experiments, and is usually taken as 0.2~0.5 to ensure that the reaction time does not exceed 2.5 seconds; For a moment The equivalent residual visual adaptation state of vehicle speed correction; For a moment The road surface adhesion coefficient; is the gravitational acceleration constant.

[0124] Step S4.4. Comprehensive control constraint system: Based on the two distance parameters mentioned above, the core constraint for the control of tunnel group headlights can be established: at any time, the geometrically corrected effective visual detection distance must not be less than the safe handling distance.

[0125]

[0126] To reflect the anti-glare requirements, the glare risk index extracted in step S2 is used. The dynamic allowable upper limit for the specular enhancement component is defined as follows:

[0127] in, This indicates the maximum permissible value of the highlight enhancement component under conditions where there is no risk of glare. This represents the glare risk suppression coefficient. As the glare risk increases, the upper limit of permissible highlight enhancement automatically decreases, thereby avoiding causing uncomfortable glare to oncoming or preceding vehicles.

[0128] Furthermore, to suppress drastic fluctuations in light output, a rate-of-change constraint is constructed. The upper limit of the rate of change of the control vector within a single sampling period is defined as:

[0129] in, Represents the infinite norm; Indicates the label based on the current driving stage. and continuous tunnel status indicators An upper limit for the adaptive control rate of change is determined. In the inlet pre-adaptation section and in a continuous tunnel scenario, a faster acceleration is allowed; in the outlet gradual adaptation section, a more gradual descent rate is forced to ensure comfort and continuity.

[0130] Step S5. Headlight adaptive control command generation and constraint smooth output: Step S5.1. Target generation based on visibility security gaps: The control system first quantifies the gap between the current level of visibility and security requirements. This defines the visibility-security gap. for:

[0131] when When this occurs, it indicates that the current effective visual detection range is insufficient and enhanced illumination is needed; when When this occurs, it indicates that the safety margin has been met.

[0132] To drive control decisions, this distance gap is mapped to an equivalent increment in lighting demand:

[0133] in, This is a mapping coefficient from distance requirements to lighting intensity requirements, used to address visibility safety gaps. Converted into equivalent incremental lighting demand Preferably, The headlight output can be gradually adjusted until it meets the requirements through calibration tests under several typical vehicle speeds, visibility conditions, and road surface adhesion conditions. Record the corresponding Sample pairs were obtained, and least squares fitting was used to obtain... ;in, For the first Visible safety gaps recorded in the calibration test For the first The incremental lighting requirement applied during the calibration test to ensure that the line-of-sight distance meets safety requirements.

[0134] Step S5.2. Stage-adaptive target generation of basic lighting components: Considering the different lighting requirements of vehicles at different stages in the tunnel complex, the basic lighting components... The target value is generated in stages.

[0135] The estimated remaining time (time distance) for the vehicle to reach the current tunnel entrance, calculated based on step S1, is as follows: The estimated remaining time to the current tunnel exit is Define the entry lookup weight. The weighting of the decline in exports is for:

[0136] To compensate for the inhibitory effect of visual residual states on recognition ability, the visual residual compensation weight is defined as follows:

[0137] in, This represents the residual visual adaptation state compensation gain coefficient.

[0138] Target values ​​of basic lighting components The target basic lighting intensity is defined as: ;

[0139] in, This represents an interval projection operator that limits the target value of basic lighting to an allowable range. Maximum basic lighting intensity; This represents the reference value for basic lighting under normal driving conditions outside the tunnel; Reference values ​​for basic lighting during the stable phase inside the cave; This represents the rate of decline during the export recovery phase.

[0140] Step S5.3. Synergistic target generation of specular enhancement and matrix-based zoned light distribution: In addition to the basic lighting component, to further improve the effective lighting capability of the forward key area, a coordinated control objective is constructed for the high-brightness enhancement component and the zonal light distribution of the matrix headlights.

[0141] Target value of specular enhancement component The target highlight enhancement component is defined as follows:

[0142] in, This indicates the gain coefficient required to enhance highlights; This represents the dynamic permissible upper limit determined by the glare risk constraint. The formula shows that the activation intensity of highlight enhancement is primarily driven by visibility safety gaps and visual residual compensation requirements, while also being strictly constrained by the glare risk upper limit.

[0143] For advanced lighting systems, the target light distribution control parameters for each zone of a matrix headlight are determined jointly by road geometry and glare risk. The matrix headlight's first... Each partition at time Target control quantity That is, the first Each partition at time The target light distribution control quantity is:

[0144] in, and They represent the first Minimum and maximum values ​​of each zone control quantity; Indicates the first Reference light distribution values ​​for each zone under normal operating conditions; Indicates the road curvature with respect to the first Gain coefficient of target light distribution in each zone; Indicates the yaw rate with respect to the first Gain coefficient of target light distribution in each zone; Indicates the risk of glare to the first The suppression coefficient of target light distribution in each zone; Indicates the curvature of the road; Indicates the vehicle's yaw rate; This indicates a risk indicator of glare.

[0145] Step S5.4. Output Smoothing: The target control quantities of each zone of the integrated basic lighting, high beam enhancement, and matrix headlights are used to construct the timing. The target control vector is:

[0146] To ensure the continuity of control output and driving comfort, the rate of change of the target control vector needs to be limited. (Note: The last part, "based on the current driving stage label," appears to be an error and doesn't need a direct translation.) The maximum allowable step size vector for each control component adaptively determined by the continuous tunnel state flag is:

[0147] in, These represent the maximum allowable step size vectors for the base illumination intensity and the highlight enhancement component, respectively. represent Each partition at time The maximum allowable step size vector of the light distribution control quantity. Represents the transpose symbol.

[0148] The intermediate control vector after the rate of change constraint is defined as follows:

[0149] in, This represents the intermediate control vector after the rate of change has been limited; This represents the element-wise limiting operator, used to limit the change of each control component within a single sampling period to the corresponding allowable range.

[0150] To further suppress high-frequency jumps, a first-order low-pass filter is applied to the intermediate control vector to obtain the final output control vector:

[0151] in, Represents the filter coefficients. The smaller the value, the smoother the output.

[0152] 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 adaptive control method for vehicle headlights in a continuous tunnel group scenario.

[0153] The present invention also provides a computer-readable medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the above-described adaptive control method for vehicle headlights in a continuous tunnel group scenario.

[0154] 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 program is executed by a computer to achieve the above functions. 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 achieved.

[0155] 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.

[0156] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of protection described in the claims.

Claims

1. An adaptive control method for vehicle headlights in a continuous tunnel group scenario, characterized in that, The method includes the following steps: Step S1. Obtain the topology of the tunnel group, determine whether adjacent tunnels are continuous based on the visual effective recovery window, and determine the current driving stage of the vehicle based on the distance-time quantity and brightness change trend; Step S2. Construct residual visual adaptation state features and extract vehicle motion and road geometry features, glare-related target features; the residual visual adaptation state features include lighting environment features, occlusion features, and visibility features; Step S3. Construct a residual visual adaptation state model, determine the residual visual adaptation state intensity based on the data obtained in steps S1 and S2; and when it is determined to be a continuous tunnel in step S1, complete the propagation and inheritance of residual visual adaptation state between tunnel groups; at the same time, correct the residual visual adaptation state intensity based on vehicle speed to obtain the equivalent residual visual adaptation state corrected by vehicle speed. Step S4. Construct an effective illuminance model of the headlight output and the target area. Calculate the effective visual detection distance based on illuminance, visibility, occlusion, and equivalent residual visual adaptation state. Correct the effective detection distance based on road curvature to obtain a geometrically corrected effective detection distance. Then calculate the safe handling distance based on vehicle speed and equivalent residual visual adaptation state. Finally, establish the constraints for headlight control in the tunnel group. Step S5. Calculate the visibility safety gap based on the effective detection distance and safe handling distance with geometric correction, and map it to the equivalent lighting demand increment. Then calculate the target value of each component of the headlight control vector, and finally output the final control vector based on the constraints.

2. The adaptive control method for vehicle headlights in a continuous tunnel group scenario according to claim 1, characterized in that, Step S1: Obtain the arc length coordinates corresponding to each tunnel entrance and exit. Based on the positional relationship between the vehicle and the previous tunnel exit and the next tunnel entrance, calculate the arc length coordinates of the first tunnel entrance and exit. Location in the section between tunnels Effective contribution weight of visual restoration Thus determining the first Effective visual recovery time between tunnel sections The expression is: ； in, and The vehicles traveled to the first The tunnel exit and the first The moment of the tunnel entrance To obtain the average speed of the vehicle within a set range before the exit of the kth tunnel, as the predicted speed of the road section between tunnels; When the Effective visual recovery time between tunnel sections Less than the minimum recovery time threshold At that time, the first The tunnel and the first The tunnels form a continuous tunnel pair in the sense of visual restoration.

3. The adaptive control method for vehicle headlights in a continuous tunnel group scenario according to claim 2, characterized in that, In step S1, first determine the vehicle relative to the first... Signed distance between the tunnel entrance and exit Then calculate the vehicle relative to the first Time distance between tunnel entrance and exit Then based on the real-time ambient brightness sequence Calculate the rate of change of brightness Finally, based on the acquired data, the current driving stage of the vehicle is determined, which is divided into four segments.

4. The adaptive control method for vehicle headlights in a continuous tunnel group scenario according to claim 3, characterized in that, In step S2, the light environment characteristic is the light intensity abrupt change index, expressed as: ； in, Indicates time The intensity of sudden changes in light intensity; This represents the difference between the current brightness and the reference brightness, i.e., the sudden change in brightness due to the static difference between light and dark. Indicates the rate of change in brightness; The reference luminance constant is pre-calibrated. and , representing the weights of static brightness difference and dynamic change term in the comprehensive index, respectively; The occlusion feature is the occlusion rate. The visibility feature is visibility. ; The glare-related target features are glare risk indicators. The expression is: ； in, , These represent the relative distance and relative azimuth angle of oncoming vehicles, respectively. , These represent the relative distance and relative azimuth angle of the vehicle in front, respectively. Angle-sensitive weighting function, The relative azimuth angle. These are the weighting coefficients. To prevent positive numbers with a denominator of zero.

5. The adaptive control method for vehicle headlights in a continuous tunnel group scenario according to claim 4, characterized in that, The expression for the residual visual adaptation state model in step S3 is: ； in, Indicates time The intensity of residual visual adaptation state, This indicates that the visual adaptation is sufficient. This indicates that the debt has reached saturation. For saturated projection operators, For a moment The residual visual adaptation state increment, i.e., the debt injection increment. For a moment The amount of residual visual adaptation state attenuation; The sampling time interval; During state propagation and inheritance, the visual deficit that has not yet faded from the previous tunnel exit is explicitly carried over to the control time of the next tunnel entrance, and the initial adaptive state at the entrance is inherited and updated: ； in, Indicates the number of inheritance updates. The residual visual adaptation state at the tunnel entrance, the vehicle arrives at the first The time of the tunnel entrance is , This means that the effective visual recovery time between tunnels is less than the minimum recovery time threshold, indicating that the visual residue of the preceding tunnel still has an impact on the subsequent tunnel. This indicates that the residual visual adaptation between tunnel sections has been fully restored and is unaffected by the preceding tunnel. No. The equivalent residual visual adaptation state inside the tunnel is as follows: ； in, The equivalent residual visual adaptation state after vehicle speed correction. For the vehicle's speed, The value is taken as a reference speed.

6. The adaptive control method for vehicle headlights in a continuous tunnel group scenario according to claim 5, characterized in that, The expression for the headlight output and the effective illuminance model of the target area in step S4 is as follows: ； in, Indicates time Effective illuminance in the target area; This represents the base illuminance conversion factor after considering the combined optical characteristics and installation parameters of the luminaire. Indicates the basic lighting intensity; This represents the illuminance gain coefficient corresponding to the specular enhancement component. Indicates the highlight enhancement component. This represents the combined effect of each zone of the matrix headlight on the target area, i.e., the equivalent zone contribution. Indicates the curvature of the road; Indicates the road slope; and These represent the attenuation coefficients of road curvature and slope on the effectiveness of lighting projection, respectively.

7. The adaptive control method for vehicle headlights in a continuous tunnel group scenario according to claim 6, characterized in that, Effective visual detection distance in step S4 To reliably identify the furthest distance of a target ahead under the current complex environment and driver status, the expression is: ； In the formula, The reference distance calibration value, , , , All are functions, representing the effects of illuminance, visibility, occlusion, and equivalent residual visual adaptation state on detection distance, respectively. ;in, It is the illuminance saturation characteristic constant of the human eye's photosensitive properties. , For visibility degradation sensitivity index, For reference visibility; ;in, This is the magnification factor for occlusion. ;in, The driver's visual recovery time constant; Safe handling distance in step S4 The expression is: ； in, For vehicles at any time The speed of travel; The baseline reaction time; The sensitivity coefficient of residual visual adaptation to reaction time amplification; For a moment The equivalent residual visual adaptation state of vehicle speed correction; For a moment The road surface adhesion coefficient; is the gravitational acceleration constant.

8. The adaptive control method for vehicle headlights in a continuous tunnel group scenario according to claim 7, characterized in that, The constraint condition for controlling the tunnel group headlights in step S4 is: at any time, the geometrically corrected effective visual detection distance must not be less than the safe handling distance; Set the dynamic allowable upper limit for the specular enhancement component. ;in, This indicates the maximum permissible value of the highlight enhancement component under conditions where there is no risk of glare. Indicates the glare risk suppression coefficient; To assess glare risk, an upper limit is set for the rate of change of the control vector within a single sampling period, the upper limit being based on the current vehicle driving stage. With continuous tunnel status signs Adaptive adjustment.

9. The adaptive control method for vehicle headlights in a continuous tunnel group scenario according to claim 8, characterized in that, Target value of the basic lighting component in step S5 The expression is: ； in, This represents an interval projection operator that limits the target value of basic lighting to an allowable range. Maximum basic lighting intensity; This represents the reference value for basic lighting under normal driving conditions outside the tunnel; Reference values ​​for basic lighting during the stable phase inside the cave; This represents the rate of decline during the export adjustment phase. The entrance look-ahead weight is calculated based on the time distance from the vehicle to the current tunnel entrance. The exit drop weight is calculated based on the time distance from the vehicle to the current tunnel exit. For visual persistence compensation weights, This is the headlight control vector for the vehicle at the previous moment. This is the equivalent increase in lighting demand; Target value of specular enhancement component The expression is: ; in, This indicates the gain coefficient required to enhance highlights; This represents the dynamic allowable upper limit determined by the glare risk constraint, i.e., the dynamic allowable upper limit of the specular enhancement component; Matrix headlights Each partition at time Target control quantity The expression is: ; in, and They represent the first Minimum and maximum values ​​of each zone control quantity; Indicates the first Reference light distribution values ​​for each zone under normal operating conditions; Indicates the road curvature with respect to the first Gain coefficient of target light distribution in each zone; Indicates the yaw rate with respect to the first Gain coefficient of target light distribution in each zone; Indicates the risk of glare to the first The suppression coefficient of target light distribution in each zone; Indicates the curvature of the road; Indicates the vehicle's yaw rate; This indicates a risk indicator of glare.

10. A vehicle headlight adaptive control method for continuous tunnel group scenarios according to any one of claims 5 to 9, characterized in that, In step S3, time Residual visual adaptation state increment The expression is: ； in, The injection gate function takes a value of 1 during the inlet pre-adaptation and outlet soft-adaptation phases, and a value of 0 otherwise. Control the injection amplitude, The adjustment parameter is used to regulate the rate at which the injection amount increases with the mutation intensity; For a moment The intensity of sudden changes in light intensity, The visibility degradation factor is calculated based on occlusion features and visibility features. time Residual visual adaptation state attenuation The expression is: ； in, To restore the gate function, the value is 1 during the stable section inside the tunnel and the recovery section between tunnels, and 0 otherwise; The recovery time constant is a time-varying value, calculated based on occlusion and visibility features.

Images