Vehicle-mounted illumination intelligent control method and system based on laser radar

Abstract

The invention discloses a vehicle-mounted illumination intelligent control method and system based on a laser radar, and relates to the field of vehicle-mounted illumination, and the method comprises an acquisition module which is used for collecting the three-dimensional point cloud data of a front road detected by the vehicle-mounted laser radar; the processing module is used for receiving the three-dimensional point cloud data, performing denoising, coordinate calibration and multi-dimensional data fusion processing on the three-dimensional point cloud data, and generating a standardized road environment perception data set; three-dimensional point cloud data of a front road is detected through a laser radar, a standardized road environment sensing data set is generated through data processing, road types, traffic participant distribution and relative motion relations are accurately recognized, laser headlamp beam shape, brightness and projection angle control parameters are dynamically output accordingly, and an execution mechanism is driven to adjust the lighting state.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of vehicle lighting, in particular to a vehicle lighting intelligent control method and system based on laser radar. BACKGROUND

[0002] Vehicle lighting is a key configuration to ensure driving safety, covering headlamps (high / low beam), turn signals, brake lights, outline lights, etc. The invention patent application with application number 202411136496.5 discloses an LED vehicle light intelligent control method based on environmental perception, including: obtaining environmental state data in the vehicle driving process at the monitoring time point; sending the current environmental state data to the vehicle light control strategy output model for processing to output the LED vehicle light control strategy; intelligently controlling the LED vehicle light on the vehicle through the LED vehicle light control strategy; at the same time, it also includes real-time training of the vehicle light control strategy output model, the specific steps are as follows: at the monitoring time point, the environmental state data is sent to the target vehicle light control strategy output model for processing to output the target vehicle light control strategy, then the environmental state data and the target vehicle light control strategy are spliced to construct value evaluation data, the value evaluation data is sent to the value evaluation network model for processing to output strategy value data, the strategy value data is calculated for the strategy gradient value of the target vehicle light control strategy, and then the parameters of the vehicle light control strategy output model are adjusted based on the strategy gradient value using the gradient ascent method to realize real-time training of the vehicle light control strategy output model; at the same time, it also includes real-time training of the value evaluation network model, which aims to solve the problem that "traditional LED vehicle light systems usually lack the ability to perceive the surrounding environment and driving conditions, which may not be able to make the most optimized light adjustment in complex driving scenarios, and still need the driver to make manual adjustments, affecting the driving experience".

[0003] However, most of the existing technologies for intelligent control of vehicle headlamps still remain at the level of intelligent control of opening and closing, and do not configure intelligent control means for the beam shape, brightness and projection angle of the vehicle headlamps.

[0004] Therefore, we propose a vehicle lighting intelligent control method and system based on laser radar. SUMMARY

[0005] In view of the above shortcomings of the prior art, the present application provides a vehicle lighting intelligent control method and system based on laser radar, which can effectively solve the problems of the prior art.

[0006] To achieve the above purpose, the present application realizes the following technical solutions; The present application discloses a vehicle lighting intelligent control system based on laser radar, comprising: The collection module is used for collecting three-dimensional point cloud data of a front road detected by a vehicle-mounted laser radar; the processing module is used for receiving the three-dimensional point cloud data, performing denoising, coordinate calibration and multi-dimensional data fusion processing on the three-dimensional point cloud data, and generating a standardized road environment perception data set; the identification module is used for obtaining the standardized road environment perception data set, extracting scene features in the data set and performing classification, identifying a current road type, a traffic participant distribution and a relative motion relationship; the calculation module is used for receiving an identification result in the identification module, outputting dynamic control parameters of a laser headlight beam shape, brightness and projection angle based on the identification result and a preset lighting control logic; the control module is used for driving an execution mechanism of the vehicle-mounted laser headlight to adjust the beam shape, adjust the brightness and adjust the projection angle deflection according to the dynamic control parameters of the laser headlight beam shape, brightness and projection angle; and the optimization module is used for collecting real-time working state data of the control module and environment feedback data detected by the laser radar for a second time, and iteratively optimizing the dynamic control parameters output by the calculation module. The three-dimensional point cloud data of the front road includes a road geometric contour, spatial coordinates of a dynamic target and motion state parameters.

[0007] Further, the motion state parameters of the dynamic target collected in the collection module are obtained through laser radar inter-frame point cloud matching calculation. Inter-frame point cloud matching: two continuous frames of point cloud data of the laser radar are obtained, denoted as the tth frame and the t+1th frame, the frame interval time is set to 0.1s, for a dynamic target point cloud cluster A in the tth frame, a corresponding point cloud cluster A' is found in the t+1th frame through Euclidean distance matching, and the center point of A is determined and the center point of A' . Motion speed calculation of the dynamic target: , denotes the frame interval time. Motion acceleration calculation: the center point of a dynamic target point cloud cluster in the t-1th frame is obtained , the motion speed in the tth frame is calculated, denoted as , and the motion acceleration is , . If the matching degree of the two frames of point cloud clusters is less than 0.8, the motion state parameters of the previous frame are used, and the matching degree is determined by the ratio of the number of coincident point clouds to the total number.

[0008] Each point in the three-dimensional point cloud data is traversed , i=1, 2, …, N, N is the total number of point clouds, and each is taken as the center of a spherical neighborhood with a radius of r, and the number of other point clouds contained in the spherical neighborhood is counted ; Calculate the neighborhood average number of all point clouds , synchronize the denoising threshold , , represents the density adjustment coefficient, the value range is 0.3-0.6, when the laser radar detection distance is far, the overall distribution of the point cloud is sparse, and the density adjustment coefficient should be larger; When the laser radar detection distance is close, the overall distribution of the point cloud is dense, and the density adjustment coefficient should be smaller; ; ; The number of neighborhood points < The points are determined as noise points and removed, and the points ≥ The point cloud after denoising is obtained; , wherein the spherical neighborhood radius r is dynamically determined according to the real-time detection distance d of the laser radar, and the calculation formula is r=d×0.02, the value range of d is 0-200 meters, and 0.02 is an initial preset constant term, and the value range of the preset constant term is 0.01~0.05.

[0009] Conversion of laser radar coordinate system to vehicle coordinate system: Convert the point cloud coordinates in the laser radar coordinate system to the coordinates in the vehicle coordinate system, and the conversion formula is: , , represents the 3×3 rotation matrix of the laser radar coordinate system to the vehicle coordinate system, , represents the 3×1 translation vector of the laser radar coordinate system to the vehicle coordinate system; Determined by the pitch angle α, roll angle β and heading angle γ when the laser radar is installed; ; Determined by the three-dimensional offset of the laser radar installation position relative to the vehicle body origin, that is ; ; Conversion of vehicle coordinate system to world coordinate system: Convert the coordinates in the vehicle coordinate system to the coordinates in the world coordinate system, and the conversion formula is: , , represents the 3×3 rotation matrix of the vehicle coordinate system to the world coordinate system, , represents the 3×1 translation vector of the vehicle coordinate system to the world coordinate system; , represents the current GPS heading angle of the vehicle; , represents the current GPS coordinates of the vehicle; When the converted W-system coordinates are subjected to bias correction, the following is obeyed: ; In the formula: is the corrected coordinate; , and are calculated in the same logic; The coordinate value in the x-axis direction of the jth frame of three-dimensional point cloud data converted from the laser radar coordinate system to the vehicle body coordinate system to the world coordinate system, j is the frame number, Similarly.

[0010] Further, when the identification module extracts scene features and classifies, the continuous operations of extraction, fusion and determination are performed in sequence: Extract the core scene features: Linear fitting is performed on the road edge point cloud in the standardized road environment perception data set to obtain the slope of the road edge, and the absolute value of the slope is taken as the road edge feature S1; The minimum bounding box is identified for each point cloud cluster corresponding to the traffic participant, and the volume of the bounding box is recorded as the traffic participant volume feature S2; The relative speed of the traffic participant relative to the vehicle is calculated, and the absolute value of the relative speed is taken as the relative motion speed feature S3; Feature weighted fusion: , is the weight, The sum of the three is 1, and all are positive numbers, and the value ranges of the three are [0.2, 0.3], [0.4, 0.5], [0.2, 0.3] respectively; Scene classification and determination: When F < F1, it is determined that it is a city branch road and there is no large traffic participant; When F1≤F<F2, it is determined that it is a city main road and there is a small traffic participant; When F≥F2, it is determined that it is a highway and there is a large traffic participant; Wherein, [F1, F2] is a preset threshold, and the unit is defined as .

[0011] Further, when the calculation module outputs the dynamic control parameter based on the identification result, the lighting control logic applied is: Beam shape control includes control of horizontal width W and vertical height H: The horizontal width W is determined according to the standard lane width of the current road type: When it is a city branch road, , represents the standard lane width of the current road type, 0.5 represents the reserved roadside safety distance, and the unit is m; When it is a city main road, ; When it is a highway, ; The vertical height H is determined according to the maximum height of the traffic participant: , represents the maximum height of the traffic participant, 0.3 represents the reserved height safety distance, and H∈[1.8,3.0]m; The brightness control includes the determination of the reference brightness and the control of the actual brightness: According to the environmental light intensity I, the reference brightness is set as 8000cd when I<200lux, corresponding to night, 5000cd when 200lux≤I<500lux, corresponding to dusk or dawn, and 2000cd when I≥500lux, corresponding to daytime; According to the relative distance between the traffic participant and the vehicle, , represents the actual brightness, represents the reference brightness, represents the reference distance, which is initially set as 50m, and ∈[1000,10000]cd; The projection angle control includes the control of the horizontal deflection angle and the vertical deflection angle: determined by the horizontal offset of the traffic participant , , represents the horizontal deflection angle; determined by the vertical height h of the traffic participant and , , represents the vertical deflection angle, and is constrained in range.

[0012] Further, when the control module drives the actuator to adjust the projection angle deflection, the PID closed-loop control logic is used to control the adjustment accuracy synchronously, and the process includes: The projection angle target value is set, which is from the calculation module, the current actual angle of the actuator is collected, and the angle deviation is calculated; The output control quantity of the PID controller is calculated by the following formula: ; wherein: is the control amount of the PID controller output to the projection angle actuator of the vehicle-mounted laser headlamp at the current control cycle time t; is the proportional coefficient; is the angle deviation at the current control cycle time t; is the integral coefficient; is the angle deviation at any time point ; is the differential coefficient; represents the integral value of the angle deviation in the time interval [0, t]; represents the derivative of time t; The actuator drives the angle adjusting motor according to the control amount , and each control cycle is 0.02 seconds long until ; ; wherein, ∈[0.8, 1.2], the greater the value is, the greater the deviation between the actual value and the target value of the projection angle of the vehicle-mounted laser headlamp is, and the faster the deviation needs to be reduced; the smaller the value is, the smaller the deviation between the actual value and the target value of the projection angle is, and the less the adjustment overshoot needs to be avoided; ∈[0.1, 0.3], the greater the value is, the longer the duration of the static deviation between the actual value and the target value of the projection angle is, and the greater the deviation amplitude is; the smaller the value is, the shorter the duration of the static deviation between the actual value and the target value of the projection angle is, and the smaller the deviation amplitude is; ; ∈[0.05, 0.15], the greater the value is, the faster the change rate of the projection angle deviation is, and the more the oscillation in the angle adjustment process needs to be suppressed; the smaller the value is, the slower the change rate of the projection angle deviation is, and the more the adjustment process needs to be smooth.

[0013] Further, when the optimization module iteratively optimizes the dynamic control parameters, the deviation value between the lighting parameters actually output by the actuator and the dynamic control parameters output by the calculation module is taken as the core optimization index, and the process includes: calculating the optimization deviation E to quantify the difference between the actual lighting effect and the theoretical control target: ; wherein: is the actual output brightness, actual projection angle, and actual beam width of the actuator inversely calculated by the secondary detection of the laser radar; is the brightness, angle, and width dynamic control parameters output by the calculation module; set the optimization coefficient λ, which is determined according to the environmental change rate dynamic adjustment; , current light intensity; previous optimization period light intensity; optimization period, =1s; when <50lux / s, λ takes 0.1, when 50lux / s≤ <100lux / s, λ takes 0.3, when ≥100lux / s, λ takes 0; the optimized control parameters are: ; wherein, the optimized brightness, projection angle and beam width are fed back to the calculation module to replace the original parameters, completing an iteration.

[0014] Further, the acquisition module is interactively connected with a processing module and an identification module through a wireless network, the identification module is interactively connected with a calculation module through a wireless network, the calculation module is interactively connected with a control module through a wireless network, and the acquisition module and the control module are interactively connected with an optimization module through a wireless network.

[0015] In another aspect, a laser radar-based vehicle-mounted lighting intelligent control method comprises: acquiring three-dimensional point cloud data of a front road detected by a vehicle-mounted laser radar; sequentially performing denoising, coordinate calibration and multi-dimensional data fusion processing on the acquired three-dimensional point cloud data to generate a standardized road environment perception data set, and the coordinate calibration needs to complete three-level conversion of a laser radar coordinate system, a vehicle body coordinate system and a world coordinate system; extracting scene features from the standardized road environment perception data set and performing weighted fusion classification to identify the current road type, the traffic participant distribution and the relative motion relationship therebetween; based on the identified road type, traffic participant distribution and relative motion relationship, combining a preset lighting control logic, outputting dynamic control parameters of a laser headlight beam shape, brightness and projection angle; driving an execution mechanism of the vehicle-mounted laser headlight to adjust the beam shape, adjust the brightness and deflect the projection angle according to the output dynamic control parameters, and synchronously applying a PID closed-loop control logic in the projection angle deflection adjustment process; collecting real-time working state data of the execution mechanism and environment feedback data detected by the laser radar for the second time, taking the deviation between the actual lighting parameters and the dynamic control parameters as the core index, iteratively optimizing the dynamic control parameters and feeding them back to the parameter calculation link to replace the original parameters.

[0016] ​Compared with the known prior art, the technical scheme provided by the application has the following beneficial effects: The application provides a vehicle-mounted lighting intelligent control method and system based on a laser radar. BRIEF DESCRIPTION OF DRAWINGS

[0017] In order to more clearly illustrate the technical scheme in the embodiments of the present application or the prior art, the drawings needed to be used in the embodiments or the prior art description will be briefly introduced as follows.

[0018] Figure 1 FIG. 1 is a structural schematic diagram of a vehicle-mounted lighting intelligent control system based on a laser radar; Figure 2 FIG. 2 is a flowchart of a vehicle-mounted lighting intelligent control method based on a laser radar. DETAILED DESCRIPTION

[0019] In order to make the purpose, technical scheme and advantages of the embodiments of the present application more clear, the technical scheme in the embodiments of the present application will be clearly and completely described below in combination with the drawings in the embodiments of the present application.

[0020] The present application will be further described below in combination with the embodiments. EMBODIMENT

[0021] The vehicle-mounted lighting intelligent control system based on a laser radar in this embodiment, as shown in FIG. 1, comprises: Figure 1 ​The acquisition module is used to collect the three-dimensional point cloud data of the road ahead detected by the vehicle-mounted laser radar; The dynamic target motion state parameters collected in the acquisition module are obtained through the LiDAR inter-frame point cloud matching calculation: Inter-frame point cloud matching: Obtain two consecutive frames of LiDAR point cloud data, recorded as frame t and frame t+1, with the frame interval set to 0.1s. For the dynamic target point cloud cluster A in frame t, find the corresponding point cloud cluster A′ in frame t+1 through Euclidean distance matching and determine the center point of A. and the center point of A′ ; Calculation of the motion speed of dynamic targets: , Indicates the frame interval time; Calculation of motion acceleration: Get the dynamic target point cloud cluster in the t-1 frame Center point , and then calculate the motion speed of the tth frame, recorded as , then the acceleration of motion is , the unit is ; Among them, if the matching degree of the point cloud clusters of two frames is less than 0.8, the motion state parameters of the previous frame will be used, and the matching degree is determined by the ratio of the number of overlapping point clouds to the total number; The processing module is used to receive 3D point cloud data, perform denoising, coordinate calibration, and multi-dimensional data fusion processing on the 3D point cloud data, and generate a standardized road environment perception data set; When the processing module performs denoising on 3D point cloud data, the following steps are included: Traverse each point in the 3D point cloud data ,i=1,2,…,N, N is the total number of point clouds, each As the center of the sphere, set a spherical neighborhood with a radius of r, and count the number of other point clouds contained in the spherical neighborhood ; Find the average number of neighbors for all point clouds , set the denoising threshold synchronously , Indicates the density adjustment coefficient, with a value range of 0.3-0.6. When the lidar detection distance is far and the point cloud is sparsely distributed, the density adjustment coefficient The larger the value, the higher the density adjustment coefficient. The smaller the value, the lower the value should be; in the laser radar point cloud denoising, due to the low point cloud density in the distant area (n avg The point cloud density in the close-range area is high (n avgLarge) and good signal quality, too high a threshold will mistakenly delete valid data. Therefore, the coefficient k is dynamically adjusted to adapt to the denoising needs of different density areas: when the point cloud is sparse, the k value is increased to increase the denoising threshold T, ensuring effective noise filtering in low-density environments; when the point cloud is dense, the k value is reduced to lower the denoising threshold T, avoiding excessive filtering of valid point cloud data. This adaptive adjustment mechanism maintains relatively stable denoising effects and data integrity at different detection distances. The number of neighborhood point clouds < point Determine as noise point and remove, retain Determine as noise point and remove, retain ≥ The point cloud is denoised to obtain the point cloud data; The spherical neighborhood radius r is dynamically determined based on the real-time detection distance d of the lidar. The calculation formula is r=d×0.02, where the value range of d is 0-200 meters, 0.02 is the initial preset constant term, and the preset constant term ranges from 0.01 to 0.05. When the processing module performs coordinate calibration on 3D point cloud data, the processing goal is to complete the three-level coordinate conversion between the lidar coordinate system, the vehicle body coordinate system, and the world coordinate system. The process is as follows: Conversion from LiDAR coordinate system to vehicle body coordinate system: The point cloud coordinates in the laser radar coordinate system Convert to the coordinates of the vehicle body coordinate system , the conversion formula is: , Represents the 3×3 rotation matrix from the lidar coordinate system to the vehicle body coordinate system, Represents the 3×1 translation vector from the lidar coordinate system to the vehicle body coordinate system; Determined by the pitch angle α, roll angle β and heading angle γ when the lidar is installed; ; The three-dimensional offset of the laser radar installation position relative to the vehicle body origin Determine, that is ; Conversion from body coordinate system to world coordinate system: The coordinates of the vehicle body coordinate system Convert to world coordinates , the conversion formula is: , Represents the 3×3 rotation matrix from the body coordinate system to the world coordinate system, Represents the 3×1 translation vector from the body coordinate system to the world coordinate system; , represents the current GPS heading angle of the vehicle; , represents the current GPS coordinates of the vehicle; When the converted W system coordinates are corrected for deviation, the following is obeyed: ; In the formula: is the corrected coordinate; , and are calculated in the same logic as The coordinate value in the x-axis direction of the jth frame of three-dimensional point cloud data after conversion from the laser radar coordinate system to the vehicle body coordinate system to the world coordinate system, j is the frame number, Similarly; In the above setting, the conversion formula from the laser radar coordinate system to the vehicle body coordinate system is combined by a 3x3 rotation matrix and a 3x1 translation vector, the rotation matrix is determined by the pitch angle α, roll angle β and heading angle γ when the laser radar is installed, which can offset the influence of installation angle deviation on the coordinates, and the translation vector is determined by the three-dimensional offset of the laser radar relative to the vehicle body origin, which ensures that the coordinate conversion can accurately reflect the relative position relationship between the laser radar and the vehicle body; The conversion formula from the vehicle body coordinate system to the world coordinate system is to construct a rotation matrix with the current GPS heading angle of the vehicle and a translation vector with the current GPS coordinates, so as to realize the alignment of the vehicle body coordinates and the global coordinates; the deviation correction formula is corrected by the difference between the average value of the converted coordinates of multiple frames and the converted coordinates of the current frame, which effectively reduces the coordinate deviation caused by GPS positioning error and installation deviation, and ensures the accuracy of the three-dimensional point cloud data in the unified coordinate system; The recognition module is used to obtain a standardized road environment perception data set, extract scene features in the data set and classify them, recognize the current road type, traffic participant distribution and relative motion relationship; When the recognition module extracts scene features and classifies them, the continuous operations of extraction, fusion and determination are performed in sequence: Extract core scene features: Linearly fit the road edge point cloud in the standardized road environment perception data set to obtain the slope of the road edge, take the absolute value of the slope, and mark it as road edge feature S1; Identify the minimum bounding box for each point cloud cluster corresponding to a traffic participant, and mark the volume of the bounding box as the traffic participant volume feature S2; Calculate the relative speed of the traffic participant relative to the vehicle, take the absolute value of the relative speed, and mark it as the relative motion speed feature S3; Feature weighted fusion: , are positive numbers, and the value ranges of the three are [0.2, 0.3], [0.4, 0.5], [0.2, 0.3] respectively; Scene classification determination: When F < F1, it is determined that: urban branch road and no large traffic participants; When F1≤F < F2, it is determined that: urban trunk road and small traffic participants exist; When F≥F2, it is determined that: highway and large traffic participants exist; Wherein, [F1, F2] is a preset threshold, and the unit is defined as ; The calculation module is configured to receive the identification result in the identification module, and output dynamic control parameters of the light beam shape, brightness and projection angle of the laser headlamp based on the identification result and the preset lighting control logic; It should be noted that the laser headlamp of the present application does not limit its specific working mode, which can be a laser direct lighting mode, generating a laser beam directly for road lighting by a laser; it can also be a laser-excited fluorescent powder mode, such as the existing laser headlamp technology, generating white light for lighting by exciting fluorescent powder material with blue laser; it can also be a laser and LED hybrid lighting mode, combining the advantages of laser and LED light source to achieve efficient lighting; no matter which technical route is adopted, the core of the present application is to intelligently control the output parameters of the lighting system based on the perception data of the laser radar; When the calculation module outputs the dynamic control parameters based on the identification result, the lighting control logic applied is: Beam shape control includes control of horizontal width W and vertical height H: The horizontal width W is determined according to the standard lane width of the current road type: When the urban branch road is, , The standard lane width of the current road type is represented by 0.5, which represents the reserved roadside safety distance, and the unit is m; When the urban trunk road is, ; When the highway is, ; The vertical height H is determined according to the maximum height of the traffic participants: , The maximum height of the traffic participants is represented by 0.3, which represents the reserved height safety distance, and H∈[1.8, 3.0]m; Brightness control includes reference brightness determination and actual brightness control: ​According to the ambient light intensity I, the reference brightness is set as 8000cd when I < 200lux, corresponding to night, 5000cd when 200lux≤I < 500lux, corresponding to dusk or dawn, and 2000cd when I≥500lux, corresponding to daytime; According to the relative distance between the traffic participant and the vehicle Correction, , represents the actual brightness, represents the reference brightness, represents the reference distance, which is initially set as 50m, and ∈[1000,10000]cd; The projection angle control includes the control of the horizontal deflection angle and the vertical deflection angle: determined by the horizontal offset of the traffic participant , , represents the horizontal deflection angle; determined by the vertical height h of the traffic participant and , , represents the vertical deflection angle, and is constrained in the range of ; The control module drives the actuator of the vehicle-mounted laser headlamp to adjust the beam shape, the brightness, and the projection angle of the laser headlamp light beam according to the dynamic control parameters of the light beam shape, the brightness, and the projection angle, and to adjust the beam shape, the brightness, and the projection angle deflection; When the control module drives the actuator to adjust the projection angle deflection, the PID closed-loop control logic is used to control the adjustment accuracy, and the process includes: The projection angle target value is set, which is obtained from the calculation module, the current actual angle of the actuator is collected, and the angle deviation is calculated; The output control quantity of the PID controller is calculated by the following formula: ; In the formula: is the control quantity of the PID controller output to the projection angle actuator of the vehicle-mounted laser headlamp at the current control cycle time t; is the proportional coefficient; is the angle deviation at the current control cycle time t; is the integral coefficient; is the angle deviation corresponding to any time point ; is the differential coefficient; represents the integral value of the angle deviation in the time interval [0, t]; represents Differentiation with respect to time t; The actuator is controlled according to the The motor is adjusted by the dynamic angle, and each control cycle is 0.02 seconds long until < Stop adjusting when in, ∈[0.8,1.2], when the actual value of the projection angle of the vehicle laser headlight deviates greatly from the target value and the deviation needs to be quickly reduced, the value is larger; when the actual value of the projection angle deviates slightly from the target value and overshoot needs to be avoided, the value is smaller; ∈[0.1,0.3], when the static deviation between the actual value of the projection angle and the target value lasts for a long time and the deviation amplitude is large, The larger the value, the shorter the duration of the static deviation between the actual value and the target value of the projection angle and the smaller the deviation amplitude. The smaller the value; ∈[0.05,0.15]. When the rate of change of the projection angle deviation of the vehicle laser headlight is fast and the oscillation during the angle adjustment process needs to be suppressed, the larger the value is. When the rate of change of the projection angle deviation is slow and the stability of the adjustment process needs to be ensured, the smaller the value is. The above formula combines the proportional term, the integral term, and the differential term. The proportional coefficient ranges from [0.8, 1.2]. A larger value can quickly reduce the deviation when the deviation is large, while a smaller value can avoid overshoot when the deviation is small, ensuring the response speed and stability of the angle adjustment. The integral coefficient ranges from [0.1, 0.3]. A larger value can effectively eliminate cumulative errors when the static deviation lasts for a long time and has a large amplitude. A smaller value can avoid integral saturation when the deviation is small, ensuring the accuracy of angle control. The differential coefficient ranges from [0.05, 0.15]. A larger value can suppress oscillations during adjustment when the deviation changes rapidly, while a smaller value can ensure smooth adjustment when the deviation changes slowly. A control cycle of 0.02 seconds enables high-frequency adjustment until the angle deviation is less than the set threshold, ensuring that the projection angle of the vehicle laser headlights can quickly and accurately reach the target value. The optimization module is used to collect the real-time working status data of the control module and the environmental feedback data of the secondary detection of the lidar, and iteratively optimize the dynamic control parameters output by the calculation module; It should be noted that the LiDAR secondary detection refers to the complete process of the LiDAR executing the recognition module again after the lighting parameters are adjusted; When the optimization module iteratively optimizes the dynamic control parameters, the deviation between the actual lighting parameters output by the actuator and the dynamic control parameters output by the calculation module is used as the core optimization indicator. The process includes: The optimization deviation E is calculated to quantify the difference between the actual lighting effect and the theoretical control target: ; In the formula: is the actual output brightness, actual projection angle, and actual beam width of the execution mechanism of the laser radar secondary detection inversion; is the brightness, angle, and width dynamic control parameters output by the calculation module; The above formula calculates the sum of the absolute values of the actual brightness, projection angle, and beam width of the laser radar secondary detection inversion and the theoretical parameters output by the calculation module, which can comprehensively quantify the difference between the actual lighting effect and the theoretical control target in the key dimensions, and provide a clear direction for optimization. Here, inversion refers to inferring the actual lighting effect by comparing the changes in environmental data before and after control; Set the optimization coefficient λ, which is based on the environmental change rate Dynamic adjustment; , is the current light intensity; is the light intensity of the previous optimization period; is the optimization period, =1s; When <50lux / s, λ is 0.1, when 50lux / s≤ <100lux / s, λ is 0.3, and when ≥100lux / s, λ is 0; when λ is 0, the parameter optimization process is suspended. Here, it is considered that: first, rapid environmental light changes usually occur in special scenarios such as tunnel entrances and exits, large building shadow areas, etc., at this time the reliability of environmental perception data is reduced, and continuing optimization may introduce errors; second, at the moment of intense light change, the human eye needs a certain adaptation time, at this time maintaining relatively stable lighting parameters is beneficial to the visual adaptation of the driver; finally, to avoid parameter oscillation of the system when the light changes suddenly, and to ensure the smoothness and stability of the lighting control; therefore, when the environmental light tends to be stable, the system will restart the optimization process and continue to fine-tune the parameters; In the above formula, the optimization coefficient λ is calculated according to the difference between the current and the previous optimization period light intensity divided by 1 second optimization period, and when the light change rate is less than 50lux / s, λ is 0.1, when 50~100lux / s, λ is 0.3, and when greater than or equal to 100lux / s, λ is 0, which can dynamically adjust the optimization intensity according to the speed of environmental light change, and increase the optimization amplitude to quickly adapt to the environment when the light changes fast, and reduce the amplitude to maintain stable control when the light changes slowly; The optimized control parameters are: ; Among them, is the optimized brightness, projection angle, and beam width, which is fed back to the calculation module to replace the original parameters, completing an iteration; The front road three-dimensional point cloud data comprises a road geometric contour, spatial coordinates of a dynamic target, and motion state parameters. The control logic of the application can be briefly described as follows: control before state: the environment data obtained by the first identification is used as a reference; control execution: the lighting is adjusted according to the calculated parameters; control after state: the laser radar is detected again to obtain new environment data; comparison and analysis: the change of the environment data before and after (such as target identification definition, point cloud quality, etc.) is compared to indirectly infer whether the lighting system works as expected; parameter correction: if there is a deviation between the actual effect and the expectation, the control parameters are optimized.

[0022] The acquisition module is interactively connected with the processing module and the identification module through a wireless network, the identification module is interactively connected with the calculation module through a wireless network, the calculation module is interactively connected with the control module through a wireless network, and the acquisition module and the control module are interactively connected with the optimization module through a wireless network.

[0023] In the embodiment, the acquisition module runs to acquire the front road three-dimensional point cloud data detected by the vehicle-mounted laser radar, the processing module runs to receive the three-dimensional point cloud data, the three-dimensional point cloud data is processed for denoising, coordinate calibration, and multi-dimensional data fusion, a standardized road environment perception data set is generated, the identification module acquires the standardized road environment perception data set, scene features are extracted in the data set and classified, the current road type, the distribution of traffic participants, and the relative motion relationship are identified, the calculation module is used to receive the identification result in the identification module, based on the identification result and in combination with a preset lighting control logic, dynamic control parameters of the light beam shape, brightness, and projection angle of the laser headlamp are output, the control module synchronously drives the execution mechanism of the vehicle-mounted laser headlamp to adjust the light beam shape, adjust the brightness, and adjust the projection angle deflection according to the dynamic control parameters of the light beam shape, brightness, and projection angle of the laser headlamp, and finally the optimization module acquires the real-time working state data of the control module and the environment feedback data detected by the laser radar for the second time, and iteratively optimizes the dynamic control parameters output by the calculation module.

[0024] In the actual driving scene, the system in the above embodiment can accurately perceive the front road geometry and the state of the traffic participants, and dynamically adjust the light beam shape, brightness, and projection angle of the laser headlamp. For example, at night or under complex lighting, the system can adapt to different road widths, avoid glare to other traffic participants, optimize the brightness according to the target distance, continuously optimize the lighting effect in combination with real-time feedback, improve the definition of the driving field of view and safety, and reduce the driving risk caused by improper lighting. Embodiment

[0025] In the specific implementation level, on the basis of embodiment 1, the embodiment refers to Figure 2Further specific description is made to the laser radar-based vehicle-mounted lighting intelligent control system in Example 1: A laser radar-based vehicle-mounted lighting intelligent control method, comprising: Collecting three-dimensional point cloud data of a front road detected by a vehicle-mounted laser radar; Performing denoising, coordinate calibration and multi-dimensional data fusion processing on the collected three-dimensional point cloud data in sequence to generate a standardized road environment perception data set, and the coordinate calibration needs to complete three-level conversion of a laser radar coordinate system, a vehicle body coordinate system and a world coordinate system; Extracting scene features from the standardized road environment perception data set and performing weighted fusion classification to identify the current road type, the traffic participant distribution and the relative motion relationship therebetween; Based on the identified road type, traffic participant distribution and relative motion relationship, combining a preset lighting control logic, outputting dynamic control parameters of a laser headlight beam shape, brightness and projection angle; Driving an execution mechanism of a vehicle-mounted laser headlight to adjust the beam shape, adjust the brightness and adjust the projection angle deflection according to the output dynamic control parameters, and the projection angle deflection adjustment process synchronously applies a PID closed-loop control logic; Collecting real-time working state data of the execution mechanism and environment feedback data detected by the laser radar for the second time, taking the deviation of the actual lighting parameters and the dynamic control parameters as the core index, iteratively optimizing the dynamic control parameters and feeding them back to the parameter calculation link to replace the original parameters.

[0026] In summary, in the above-mentioned method and system in the execution process, the three-dimensional point cloud data of the front road is detected by the laser radar, the standardized road environment perception data set is generated through data processing, the road type, the traffic participant distribution and the relative motion relationship are accurately identified, the control parameters of the laser headlight beam shape, brightness and projection angle are dynamically outputted, the execution mechanism is driven to adjust the lighting state, and the control parameters are iteratively optimized in combination with the real-time working state and the secondary detection environment feedback, which can adapt to different road scenes and traffic conditions, avoid excessive or insufficient lighting, reduce visual interference to other traffic participants, improve the clarity of the driving field of view at night and under complex lighting conditions, ensure driving safety, and ensure the lighting adjustment accuracy through the PID closed-loop control, optimize the parameters according to the environmental change rate, further enhance the adaptability and stability of the lighting control, and reduce unnecessary energy consumption.

[0027] The above examples are only used to illustrate the technical solutions of the present application, and are not intended to limit the present application; although the present application has been described in detail with reference to the foregoing examples, those skilled in the art should understand that the technical solutions recorded in the foregoing examples can be modified, or some of the technical features can be replaced by equivalents; and these modifications or replacements will not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A vehicle-mounted lighting intelligent control system based on laser radar, characterized in that: include: The acquisition module is used to collect the three-dimensional point cloud data of the road ahead detected by the vehicle-mounted laser radar; The processing module is used to receive 3D point cloud data, perform denoising, coordinate calibration, and multi-dimensional data fusion processing on the 3D point cloud data, and generate a standardized road environment perception data set; The recognition module is used to obtain a standardized road environment perception dataset, extract scene features from the dataset, classify them, and identify the current road type, distribution of traffic participants, and relative motion relationships; A calculation module is used to receive the recognition results from the recognition module and output dynamic control parameters for the laser headlight beam shape, brightness, and projection angle based on the recognition results and preset lighting control logic; A control module is used to drive the actuator of the vehicle-mounted laser headlight to adjust the beam shape, brightness and projection angle deflection based on the dynamic control parameters of the laser headlight beam shape, brightness and projection angle; The optimization module is used to collect the real-time working status data of the control module and the environmental feedback data of the secondary detection of the lidar, and iteratively optimize the dynamic control parameters output by the calculation module; Among them, the three-dimensional point cloud data of the road ahead includes the road geometric outline, the spatial coordinates of dynamic targets and motion state parameters.

2. The vehicle-mounted lighting intelligent control system based on laser radar according to claim 1, characterized in that: The dynamic target motion state parameters collected in the acquisition module are obtained through the laser radar inter-frame point cloud matching calculation: Inter-frame point cloud matching: Obtain two consecutive frames of LiDAR point cloud data, recorded as frame t and frame t+1, with the frame interval set to 0.1s. For the dynamic target point cloud cluster A in frame t, find the corresponding point cloud cluster A′ in frame t+1 through Euclidean distance matching and determine the center point of A. and the center point of A′ ; Calculation of the motion speed of dynamic targets: , Indicates the frame interval time; Calculation of motion acceleration: Get the dynamic target point cloud cluster in the t-1 frame Center point , and then calculate the motion speed of the t-th frame, recorded as , then the acceleration of motion is , the unit is ; Among them, if the matching degree of the point cloud clusters of two frames is less than 0.8, the motion state parameters of the previous frame are used, and the matching degree is determined by the ratio of the number of overlapping point clouds to the total number.

3. The vehicle-mounted lighting intelligent control system based on laser radar according to claim 1, characterized in that: When the processing module performs denoising on the three-dimensional point cloud data, the following steps are included: Traverse each point in the 3D point cloud data , i=1,2,…,N, N is the total number of point clouds, each As the center of the sphere, set a spherical neighborhood with a radius of r, and count the number of other point clouds contained in the spherical neighborhood ; Find the average number of neighbors for all point clouds , set the denoising threshold synchronously , Indicates the density adjustment coefficient, with a value range of 0.3-0.

6. When the lidar detection distance is far and the point cloud is sparsely distributed, the density adjustment coefficient The larger the value, the higher the density adjustment coefficient. The value should be smaller; The number of neighborhood point clouds < point Determine as noise point and remove it, retain ≥ The point cloud is denoised to obtain the point cloud data; The spherical neighborhood radius r is dynamically determined based on the real-time detection distance d of the lidar. The calculation formula is r=d×0.02, where the value range of d is 0-200 meters, 0.02 is the initial preset constant term, and the preset constant term ranges from 0.01 to 0.

05.

4. The vehicle-mounted lighting intelligent control system based on laser radar according to claim 1, characterized in that: When the processing module performs coordinate calibration on the 3D point cloud data, the processing goal is to complete the three-level coordinate conversion of the LiDAR coordinate system, the vehicle body coordinate system, and the world coordinate system. The process is as follows: Conversion from LiDAR coordinate system to vehicle body coordinate system: The point cloud coordinates in the laser radar coordinate system Convert to the coordinates of the vehicle body coordinate system , the conversion formula is: , Represents the 3×3 rotation matrix from the lidar coordinate system to the vehicle body coordinate system, Represents the 3×1 translation vector from the lidar coordinate system to the vehicle body coordinate system; Determined by the pitch angle α, roll angle β and heading angle γ when the lidar is installed; ； The three-dimensional offset of the laser radar installation position relative to the vehicle body origin OK, that is ; Conversion from body coordinate system to world coordinate system: The coordinates of the vehicle body coordinate system Convert to world coordinates , the conversion formula is: , Represents the 3×3 rotation matrix from the body coordinate system to the world coordinate system, Represents the 3×1 translation vector from the body coordinate system to the world coordinate system; , Indicates the vehicle's current GPS heading angle; , Indicates the current GPS coordinates of the vehicle; When performing deviation correction on the converted W coordinates, the following applies: ； Where: is the corrected coordinate; , and The calculation logic is consistent; The j-th frame of 3D point cloud data is converted from the laser radar coordinate system to the vehicle body coordinate system and then to the world coordinate system. The coordinate value of the x-axis in the world coordinate system, j is the frame number. Same thing.

5. The vehicle-mounted lighting intelligent control system based on laser radar according to claim 1, characterized in that: When the recognition module extracts scene features and performs classification, it performs the following operations: extraction, fusion, and judgment: Extract core scene features: Perform linear fitting on the road edge point cloud in the standardized road environment perception dataset to obtain the slope of the road edge, and take the absolute value of the slope as the road edge feature S1; Identify the minimum bounding box for each point cloud cluster corresponding to each traffic participant, and record the volume of the bounding box as the traffic participant volume feature S2; Obtain the relative speed of the traffic participant relative to the vehicle, take the absolute value of the relative speed, and record it as the relative motion speed feature S3; Weighted feature fusion: , is the weight, The sum is 1, and they are all positive numbers, and the ranges of the three are [0.2, 0.3], [0.4, 0.5], and [0.2, 0.3] respectively; Scene classification judgment: When F<F1, it is judged as: urban branch road and there are no large traffic participants; When F1≤F<F2, it is judged as: urban main road and there are small traffic participants; When F ≥ F2, it is determined as: highway with large traffic participants; Among them, [F1, F2] is the preset threshold, and the unit is defined as .

6. The vehicle-mounted lighting intelligent control system based on laser radar according to claim 1, characterized in that: When the calculation module outputs dynamic control parameters based on the recognition results, the lighting control logic applied is: Beam shape control includes the control of horizontal width W and vertical height H: The horizontal width W is determined according to the standard lane width of the current road type: When the city branch road , Indicates the standard lane width of the current road type. 0.5 indicates the reserved roadside safety distance. The unit is m. When the city's main roads ; When driving on the highway, ; The vertical height H is determined according to the maximum height of the traffic participants: , Indicates the maximum height of traffic participants, 0.3 indicates the reserved height safety distance, and H∈[1.8,3.0]m; Brightness control includes reference brightness determination and actual brightness control: The brightness is set according to the ambient light intensity I: when I < 200 lux, the reference brightness is 8000 cd, corresponding to nighttime; when 200 lux ≤ I < 500 lux, the reference brightness is 5000 cd, corresponding to dusk or dawn; when I ≥ 500 lux, the reference brightness is 2000 cd, corresponding to daytime; Based on the relative distance between traffic participants and vehicles correction, , Indicates the actual brightness, Indicates the reference brightness, Indicates the reference distance, which is initially set to 50m, and ∈[1000,10000]cd; Projection angle control includes horizontal deflection angle and vertical deflection angle control: Horizontal offset by traffic participants Sure, , Indicates the horizontal deflection angle; The vertical height h of the traffic participant and Sure, , represents the vertical deflection angle, and Constrained in within the range.

7. The vehicle-mounted lighting intelligent control system based on laser radar according to claim 1, characterized in that: When the control module drives the actuator to adjust the projection angle deflection, the PID closed-loop control logic is used to control the adjustment accuracy. The process includes: Set the target value of the projection angle, which is derived from the calculation module, collect the current actual angle of the actuator, and calculate the angle deviation; The output control quantity of the PID controller is obtained by the following formula: ； Where: is the control quantity output by the PID controller to the vehicle-mounted laser headlight projection angle actuator at the current control cycle time t; is the proportionality coefficient; is the angle deviation at the current control cycle time t; is the integration coefficient; For any time point The corresponding angular deviation; is the differential coefficient; Indicates the integral value of the angle deviation in the time interval [0, t]; express Differentiation with respect to time t; The actuator is controlled according to the Drive angle adjustment motor, each control cycle lasts 0.02 seconds until < Stop adjusting when in, ∈[0.8,1.2], when the actual value of the projection angle of the vehicle laser headlight deviates greatly from the target value and the deviation needs to be quickly reduced, the value is larger; when the actual value of the projection angle deviates slightly from the target value and overshoot needs to be avoided, the value is smaller; ∈[0.1,0.3], when the static deviation between the actual value of the projection angle and the target value lasts for a long time and the deviation amplitude is large, The larger the value, the shorter the duration of the static deviation between the actual value and the target value of the projection angle and the smaller the deviation amplitude. The smaller the value; ∈[0.05,0.15]. When the rate of change of the projection angle deviation of the vehicle-mounted laser headlight is fast and the oscillation during the angle adjustment process needs to be suppressed, the larger the value is. When the rate of change of the projection angle deviation is slow and the stability of the adjustment process needs to be ensured, the smaller the value is.

8. The vehicle-mounted lighting intelligent control system based on laser radar according to claim 1, characterized in that: When the optimization module iteratively optimizes the dynamic control parameters, the deviation between the lighting parameters actually output by the actuator and the dynamic control parameters output by the calculation module is used as the core optimization indicator. The process includes: The optimization deviation E is calculated to quantify the difference between the actual lighting effect and the theoretical control target: ; Where: The actual output brightness, actual projection angle, and actual beam width of the actuator for the secondary detection inversion of the lidar; Dynamically control the brightness, angle, and width parameters output by the calculation module; Set the optimization coefficient λ, which is based on the environmental change rate Dynamic adjustment; , is the current light intensity; Optimize the light intensity for the previous cycle; To optimize the cycle, =1s; when When <50lux / s, λ is 0.1, when 50lux / s≤ When λ is less than 100 lux / s, take 0.

3. When ≥100 lux / s, λ is set to 0; The optimized control parameters are: ; in, To optimize the brightness, projection angle, and beam width, Feedback is sent to the calculation module to replace the original parameters and complete one iteration.

9. The vehicle-mounted lighting intelligent control system based on laser radar according to claim 1, characterized in that: The acquisition module is interactively connected to the processing module and the identification module via a wireless network, the identification module is interactively connected to the calculation module via a wireless network, the calculation module is interactively connected to the control module via a wireless network, and the acquisition module and the control module are interactively connected to the optimization module via a wireless network.

10. A method for intelligent control of vehicle lighting based on laser radar, the method being an implementation of the intelligent control system for vehicle lighting based on laser radar as claimed in any one of claims 1 to 9, characterized in that: include: Collect 3D point cloud data of the road ahead detected by the vehicle-mounted LiDAR; The collected 3D point cloud data is processed in sequence for denoising, coordinate calibration, and multi-dimensional data fusion to generate a standardized road environment perception dataset. Coordinate calibration requires completing three-level conversions among the LiDAR coordinate system, the vehicle body coordinate system, and the world coordinate system. Extract scene features from the standardized road environment perception dataset and perform weighted fusion classification to identify the current road type, the distribution of traffic participants, and the relative motion relationship between the two; Based on the identified road type, traffic participant distribution, and relative motion relationships, combined with preset lighting control logic, it outputs dynamic control parameters for the laser headlight beam shape, brightness, and projection angle. Drive the actuator of the vehicle-mounted laser headlights and adjust the beam shape, brightness, and projection angle deflection based on the output dynamic control parameters. The projection angle deflection adjustment process also uses PID closed-loop control logic. The real-time working status data of the actuator and the environmental feedback data of the secondary detection of the lidar are collected. The deviation between the actual lighting parameters and the dynamic control parameters is used as the core indicator. The dynamic control parameters are iteratively optimized and fed back to the parameter calculation link to replace the original parameters.

Images