Intelligent vehicle lamp dynamic control system and method based on support vector regression

Abstract

The invention provides an intelligent vehicle lamp dynamic control system and method based on support vector regression, and belongs to the technical field of vehicle lamp intelligent control. The intelligent vehicle lamp dynamic control system comprises a data acquisition module used for acquiring dynamic data of a vehicle in real time; the data processing module is used for processing the dynamic data of the vehicle to generate feature data; the SVR module is used for generating driving path prediction data and vehicle state prediction data of the vehicle in several seconds in the future based on the feature data; the vehicle lamp dynamic control module is used for generating a vehicle lamp control instruction corresponding to the driving path prediction data and the feature data based on the driving path prediction data and the feature data; the motor module is used for driving the vehicle lamp system to execute corresponding operation based on the light beam angle control instruction of the vehicle lamp; and the vehicle lamp projection subsystem is used for projecting the received image data to a front road surface. By combining the prediction capability of the support vector regression model, the irradiation area and brightness of the vehicle lamp are dynamically adjusted, and the safety and driving comfort of night driving are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent control of vehicle lights, and in particular to an intelligent vehicle light dynamic control system and method based on support vector regression. Background Art

[0002] With the development of vehicle automation and intelligence, intelligent headlight systems have gradually become one of the key technologies in the automotive industry. Traditional headlight systems are controlled by simple switches or by adjusting the light intensity, which cannot adapt to real-time changes in complex road conditions, especially at night or in bad weather conditions. In order to improve driving safety and driving experience, some existing technologies have begun to introduce dynamic headlight control systems. Existing systems usually detect the driving status of the vehicle through sensors and adjust the beam direction or brightness of the headlights according to fixed preset rules. However, these systems lack flexibility in complex road environments and it is difficult to perform precise lighting control according to real-time changes in the driving environment.

[0003] Most current headlight control systems rely on simple sensor inputs and adjust the light beam based on a preset set of rules. However, these systems have several shortcomings:

[0004] 1. Due to the limitations of fixed rules, the dynamic response speed and adaptability of the headlights are insufficient, making it difficult to cope with complex driving environments;

[0005] 2. The existing technology fails to optimize the adjustment of the light beam by combining the vehicle's driving path prediction data, resulting in a gap between the lighting effect of the headlights and the actual driving needs;

[0006] 3. Due to the lack of a real-time closed-loop feedback mechanism, the adjustment of the headlights cannot be dynamically corrected according to the actual lighting effect.

[0007] The above problems need to be solved urgently. Summary of the invention

[0008] The purpose of the present invention is to overcome at least one technical problem existing in the prior art and to provide an intelligent vehicle light dynamic control system and method based on support vector regression.

[0009] On the one hand, an embodiment of the present invention provides an intelligent headlight dynamic control system based on support vector regression, the intelligent headlight dynamic control system includes: a data acquisition module, a data processing module, an SVR module, a headlight dynamic control module, a motor module and a headlight projection subsystem, the output end of the data acquisition module is electrically connected to the input end of the data processing module, the first output end of the data processing module is electrically connected to the input end of the SVR module, the second output end of the data processing module is electrically connected to the first input end of the headlight dynamic control module, the first output end of the SVR module is electrically connected to the second input end of the headlight dynamic control module, the output end of the headlight dynamic control module is electrically connected to the motor module and the input end of the headlight projection subsystem respectively, and the motor module and the headlight projection subsystem are externally connected to the headlight system; the data acquisition module is used to obtain the dynamic data of the vehicle in real time based on the vehicle-mounted sensor; the data processing module is used to perform data processing on the dynamic data of the vehicle based on the Kalman filtering algorithm to generate feature data; the SVR module integrates a model creation module, a model optimization module, a model training module and a headlight projection subsystem. and a prediction result generation module; the model creation module is used to create a basic SVR model by introducing a Gaussian kernel function; the model optimization module is used to optimize the kernel parameters and regularization coefficients in the basic SVR model based on the Osprey optimization algorithm; the model training module is used to train the optimized basic SVR model based on a pre-stored training set; the prediction result generation module is used to input the feature data into the trained SVR model to generate the vehicle's driving path prediction data and vehicle state prediction data in the next few seconds; the headlight dynamic control module is used to generate corresponding headlight control instructions based on the driving path prediction data and the feature data, and the headlight control instructions include one or a combination of the headlight beam angle control instructions, the headlight brightness control instructions and the headlight projection spot shape control instructions; the motor module is used to drive the headlight system to perform corresponding operations based on the headlight beam angle control instructions; the headlight projection subsystem is used to control the light brightness of the headlight system and the image data to be projected based on the headlight brightness control instructions and the headlight projection spot shape control instructions.

[0010] Furthermore, the intelligent headlight dynamic control system also includes a real-time monitoring feedback module, the output end of the real-time monitoring feedback module is electrically connected to the third input end of the headlight dynamic control module, the second output end of the SVR module is electrically connected to the first input end of the real-time monitoring feedback module, and the third output end of the data processing module is electrically connected to the second input end of the real-time monitoring feedback module; the real-time monitoring feedback module is used to monitor the illumination effect of the headlight in real time through the vehicle-mounted camera, and compare it with the vehicle's driving path prediction data and vehicle status data generated by the SVR module in the next few seconds and the vehicle's front road information sent by the data processing module, and send a control instruction to the headlight dynamic control module based on the comparison result, so that the headlight dynamic control module adaptively adjusts the control rules for generating headlight control instructions based on the control instructions, and the illumination effect includes one or a combination of the light intensity, illumination angle and light spot shape of the headlight illumination area.

[0011] Furthermore, the data acquisition module is integrated with vehicle-mounted sensors, vehicle-mounted cameras and / or lidars, which are used to collect in real time one or a combination of vehicle speed, acceleration, steering wheel angle, gyroscope data, road curvature, GPS information, ambient light intensity, speed limit warning, and the distance between the obstacle in front of the vehicle and the vehicle.

[0012] Furthermore, the data processing module is used to generate characteristic data after cleaning, filtering, denoising and data normalization of the dynamic data of the vehicle based on a Kalman filter;

[0013] The Kalman filter algorithm is integrated in the Kalman filter, and its state equation is:

[0014] x k =Ax k-1 +Bu k-1 +ω k-1 ;

[0015] Among them, x k is the state of the system at time k, A represents the state transfer matrix, B represents the control matrix, and u k-1 is the control input, ω k-1 is the process noise;

[0016] The observation process describing the system is:

[0017] z k =Hx k +v k ;

[0018] Among them, z k is the observed value, H is the observation matrix, describing the relationship between the state and the measured value, v k To measure noise;

[0019] The Kalman filter performs real-time denoising on the vehicle's dynamic data by combining the prediction model and observations.

[0020] Furthermore, the objective function of the basic SVR model is:

[0021] f(x)=ω T Φ(x)+b;

[0022] Among them, ω is the weight vector, Φ(x) is the feature map of the input data x, the low-dimensional data is mapped to the high-dimensional space through the kernel function, and b is the bias term;

[0023] The Gaussian kernel function is:

[0024]

[0025] Among them, x i For any point in space, x j is the center of the kernel function, σ is the width parameter of the kernel function, ||x i -x j || is x i to x j The Euclidean distance between .

[0026] Furthermore, the model optimization module is used to: set the root mean square error of the historical data input into the SVR model as the fitness function of the Osprey optimization algorithm; the Osprey optimization algorithm optimizes the regularization coefficient and the kernel parameter in the basic SVR model by minimizing the root mean square error corresponding to the fitness function, thereby completing the process of optimizing the kernel parameters and the regularization coefficient in the basic SVR model, including: setting the initial parameters in the Osprey optimization algorithm, iterating in sequence, and reaching the maximum number of iterations or the Osprey optimization algorithm tending to be stable, indicating that the optimal value has been found.

[0027] Furthermore, the root mean square error function corresponding to the fitness function is:

[0028]

[0029] Where m is the number of samples, F i_predict is the predicted value, F i_actual is the actual value.

[0030] Furthermore, the driving path prediction data and vehicle state prediction data include: one or a combination of the vehicle's predicted speed, predicted acceleration, predicted steering wheel angle, predicted gyroscope data, predicted road curvature, predicted GPS information, predicted ambient light intensity, and predicted distance between obstacles in front of the vehicle and the vehicle in the next few seconds.

[0031] Furthermore, the dynamic control module of the headlights integrates a beam direction adjustment submodule, a beam brightness adjustment submodule and a spot shape adjustment submodule; the beam direction adjustment submodule is used to generate a control instruction for dynamically adjusting the direction of the headlight beam based on the current driving direction of the vehicle in the feature data and the predicted steering wheel angle in the driving path prediction data; the beam brightness adjustment submodule is used to generate a control instruction for dynamically adjusting the beam brightness based on the predicted vehicle speed and the predicted distance between the obstacle in front of the vehicle and the vehicle in the vehicle state prediction data; the spot shape adjustment submodule is used to generate a control instruction for dynamically adjusting the image data to be projected based on the road information in front of the vehicle sent by the data processing module.

[0032] In a second aspect, an embodiment of the present invention provides a method for dynamic control of intelligent headlights based on support vector regression, and the method is applied to the above-mentioned intelligent headlight dynamic control system based on support vector regression, and the method includes: system initialization; obtaining the dynamic data of the vehicle in real time based on the vehicle-mounted sensor through the data acquisition module; the data processing module processes the dynamic data of the vehicle based on the Kalman filter algorithm to generate feature data; the SVR module creates a basic SVR model by introducing a Gaussian kernel function; optimizes the kernel parameters and regularization coefficients in the basic SVR model based on the Osprey optimization algorithm; trains the optimized basic SVR model based on a pre-stored training set; inputs the feature data into the trained SVR model to generate the vehicle's driving path prediction data and vehicle status prediction data in the next few seconds; The headlight dynamic control module generates a corresponding headlight control instruction based on the driving path prediction data and the characteristic data; the motor module drives the headlight system to perform corresponding operations based on the headlight beam angle control instruction; the headlight projection subsystem controls the light brightness of the headlight system and the image data to be projected based on the headlight brightness control instruction and the headlight projection spot shape control instruction; the real-time monitoring feedback module monitors the illumination effect of the headlight in real time through the on-board camera, and compares it with the driving path prediction data and vehicle status data of the vehicle in the next few seconds generated by the SVR module and the road information ahead of the vehicle sent by the data processing module, and sends a control instruction to the headlight dynamic control module based on the comparison result, so that the headlight dynamic control module adaptively adjusts the control rules for generating the headlight control instruction based on the control instruction.

[0033] In a third aspect, an embodiment of the present invention further provides an electronic device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program, when executed by the processor, implements the above-mentioned intelligent vehicle light dynamic control method based on support vector regression.

[0034] In a fourth aspect, an embodiment of the present invention further provides a readable storage medium, which, when the instructions in the storage medium are executed by a processor of an electronic device, enables the electronic device to execute the above-mentioned intelligent vehicle light dynamic control method based on support vector regression.

[0035] The beneficial effects of the present invention are as follows: the present invention provides an intelligent vehicle light dynamic control system and method based on support vector regression, which can effectively extract data from the data acquisition module for real-time denoising by embedding the Kalman filter algorithm in the data processing module, and provide a more accurate state estimation; at the same time, the hyperparameters of the SVR are optimized using the swarm intelligence optimization algorithm (Osprey optimization algorithm), which can effectively improve the robustness and generalization of the SVR model, thereby further improving the prediction performance of the SVR. After the prediction algorithm is embedded in the intelligent vehicle light dynamic control system, the intelligence of the system is improved, thereby improving the safety and comfort of night driving. The system and method of the present invention can be widely used in the fields of intelligent assisted driving and unmanned driving, and also provide a new solution for the vehicle light controller algorithm. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0037] Figure 1 It is a schematic diagram of the structure of an intelligent vehicle light dynamic control system based on support vector regression provided in Example 1 of the present invention.

[0038] Figure 2 This is a schematic diagram of the structure of another intelligent vehicle light dynamic control system based on support vector regression provided in Example 1 of the present invention.

[0039] Figure 3 This is a flow chart of a method for dynamic control of intelligent vehicle lights based on support vector regression provided in Example 2 of the present invention.

[0040] Figure 4 This is a partial block diagram of an electronic device provided in Embodiment 3 of the present invention. DETAILED DESCRIPTION

[0041] It should be mentioned before discussing the exemplary embodiments in more detail that some exemplary embodiments are described as processes or methods depicted as flow charts. Although the flow charts describe the operations as sequential processes, many of the operations therein can be implemented in parallel, concurrently or simultaneously. In addition, the order of the operations can be rearranged. The process can be terminated when its operation is completed, but can also have additional steps not included in the accompanying drawings. The process can correspond to a method, function, procedure, subroutine, subprogram, etc.

[0042] It should be understood that, although the terms "first", "second", etc. may be used herein to describe various units, these units should not be limited by these terms. These terms are used only to distinguish one unit from another unit. For example, without departing from the scope of the exemplary embodiments, the first unit may be referred to as the second unit, and similarly the second unit may be referred to as the first unit. The term "and / or" used herein includes any and all combinations of one or more of the listed associated items.

[0043] The present invention will now be described in detail with reference to the accompanying drawings. This figure is a simplified schematic diagram, which only illustrates the basic structure of the present invention in a schematic manner, and therefore only shows the components related to the present invention.

[0044] For ease of understanding, the following professional terms are explained here:

[0045] Kalman filter: is a recursive estimation algorithm widely used in prediction, smoothing and denoising of time series data. It is mainly used for linear systems and performs real-time correction based on the noise characteristics of sensor data under a known system model. Kalman filtering can infer a more accurate state based on historical information and physical models in observation data containing noise. Kalman filtering predicts the state of the next moment based on the dynamic model of the system, and then makes corrections based on the actual measurement data. Its core idea is to give the optimal estimate based on Bayesian estimation, combining previous predictions with current observations.

[0046] SVR: (Support Vactor Regression) Support Vector Regression is a machine learning algorithm used to solve regression problems, that is, predicting the value of a continuous variable. Its principle is similar to that of Support Vector Machine (SVM), but its goal is different. SVR aims to establish a prediction model so that the error between the predicted value and the true value is as small as possible.

[0047] Osprey Optimization Algorithm (OOA) is an optimization algorithm based on osprey behavior proposed in 2023. The algorithm simulates the hunting behavior of osprey to find the optimal solution. It has the characteristics of strong optimization ability and fast convergence speed.

[0048] Example 1

[0049] For ease of understanding, the inventive concept is generally described below before describing the embodiments of the present invention in detail:

[0050] The present application provides an intelligent headlight dynamic control system and method based on support vector regression (SVR) to improve driving safety and driving experience, and realize intelligent headlight control and adaptive adjustment. It is mainly composed of 7 parts, including data acquisition module, data processing module, SVR module, intelligent headlight dynamic control module, motor module, headlight projection subsystem, and real-time feedback module. Among them, the data acquisition module is responsible for collecting the driving state data of the vehicle, the data processing module performs noise smoothing and data normalization on the collected raw data, and the SVR module builds and trains the SVR model based on the processed feature data, establishes a nonlinear relationship between the input data and the future vehicle driving path and headlight requirements, and outputs the vehicle's driving trajectory in the next few seconds. The intelligent headlight dynamic control module automatically adjusts the illumination direction, illumination angle, brightness and spot shape of the headlight according to the output of the SVR model. The motor module receives control instructions from the intelligent headlight dynamic control module and drives the headlight to perform corresponding adjustment operations. The headlight projection system is mainly responsible for projecting image data onto the road ahead. The real-time feedback module monitors the lighting effect of the headlights and environmental changes in real time, and feeds back to the system for further optimization and adjustment. The intelligent headlight dynamic control system of the present application dynamically adjusts the lighting area and brightness of the headlights by intelligently processing the vehicle and the environment, combined with the prediction ability of the support vector regression model, thereby improving the safety and driving comfort of nighttime driving.

[0051] The specific implementation is as follows:

[0052] like Figure 1 , which is a schematic diagram of the structure of an intelligent vehicle light dynamic control system based on support vector regression provided by the present invention.

[0053] As an example, the intelligent headlight dynamic control system includes: a data acquisition module 1, a data processing module 2, an SVR module 3, a headlight dynamic control module 4, a motor module 5 and a headlight projection subsystem 6, the output end of the data acquisition module 1 is electrically connected to the input end of the data processing module 2, the first output end of the data processing module 2 is electrically connected to the input end of the SVR module 3, the second output end of the data processing module 2 is electrically connected to the first input end of the headlight dynamic control module 4, the first output end of the SVR module 3 is electrically connected to the second input end of the headlight dynamic control module 4, the output end of the headlight dynamic control module 4 is electrically connected to the input ends of the motor module 5 and the headlight projection subsystem 6 respectively, and the motor module 5 and the headlight projection subsystem 6 are externally connected to the headlight system.

[0054] In some feasible implementations, the data acquisition module 1 is used to obtain the dynamic data of the vehicle in real time based on the vehicle-mounted sensors. The data acquisition module 1 is integrated with vehicle-mounted sensors, vehicle-mounted cameras and / or laser radars, and is used to collect the vehicle speed, acceleration, steering wheel angle, gyroscope data, road curvature, GPS information, ambient light intensity, speed limit warning, and one or a combination of the distance between the obstacle in front of the vehicle and the vehicle in real time. Specifically, the data acquisition module 1 uses vehicle-mounted sensors such as vehicle speed sensors, steering angle sensors, acceleration sensors, etc. to obtain the dynamic data of the vehicle in real time, including but not limited to vehicle speed, steering angle, acceleration, braking status, etc., which reflect the current motion state and driving behavior of the vehicle. At the same time, photosensitive sensors, vehicle-mounted cameras, laser radars, etc. are integrated to detect the surrounding environmental conditions in real time, including but not limited to light intensity, speed limit warning, distance of obstacles in front of the vehicle, etc., which affect the vision and road condition perception of the vehicle when driving. The data acquisition module 1 packages this information and sends it to the data processing module 2.

[0055] In some feasible implementations, the data processing module 2 is used to process the dynamic data of the vehicle based on the Kalman filter algorithm to generate characteristic data. Preferably, the data processing module 2 is used to generate characteristic data after cleaning, filtering, denoising and data normalization of the dynamic data of the vehicle based on the Kalman filter algorithm; the Kalman filter algorithm is integrated in the Kalman filter, and its state equation is:

[0056] x k =Ax k-1 +Bu k-1 +ω k-1 ;

[0057] Among them, x k is the state of the system at time k, A represents the state transfer matrix, B represents the control matrix, and u k-1 is the control input, ω k-1 is the process noise;

[0058] The observation process describing the system is:

[0059] z k =Hx k +v k ;

[0060] Among them, z k is the observed value, H is the observation matrix, describing the relationship between the state and the measured value, v k To measure noise; the Kalman filter combines the prediction model and observations to perform real-time denoising on the vehicle's dynamic data and provide a more accurate state estimate.

[0061] The data processing module 2 performs preliminary cleaning and denoising on the raw data obtained from the sensor, removes outliers and unreliable data to ensure the quality of data for subsequent processing, thereby reducing the complexity of the SVR model and improving its prediction accuracy, ensuring that the data input into the SVR model is more accurate, which is the basis for the SVR model to perform efficient predictions.

[0062] In some feasible embodiments, in combination with Figure 2 As shown, the SVR module 3 integrates a model creation module 310, a model optimization module 320, a model training module 330 and a prediction result generation module 340; the model creation module 310 is used to create a basic SVR model by introducing a Gaussian kernel function; the model optimization module 320 is used to optimize the kernel parameters and regularization coefficients in the basic SVR model based on the Osprey optimization algorithm; the model training module 330 is used to train the optimized basic SVR model based on a pre-stored training set; the prediction result generation module 340 is used to input the feature data into the trained SVR model to generate the vehicle's driving path prediction data and vehicle state prediction data in the next few seconds.

[0063] Preferably, the objective function of the basic SVR model is:

[0064] f(x)=ω T Φ(x)+b;

[0065] Among them, ω is the weight vector, Φ(x) is the feature map of the input data x, the low-dimensional data is mapped to the high-dimensional space through the kernel function, and b is the bias term;

[0066] The Gaussian kernel function is:

[0067]

[0068] Among them, x i For any point in space, x j is the center of the kernel function, σ is the width parameter of the kernel function, ||x i -x j || is x i to x j The Euclidean distance between .

[0069] Specifically, the core idea of ​​SVR is to perform regression through a hyperplane that fits the data as closely as possible and allows errors within a certain tolerance range. This tolerance range is controlled by a parameter ε, called the epsilon-insensitive tube.

[0070] The goal of SVR is to make the error within the allowable range while the weight ω is as small as possible in order to smooth the function and avoid overfitting. To achieve this goal, SVR optimizes the following formula:

[0071]

[0072] Where C is the regularization coefficient, ξ i and is a slack variable, representing the error of the predicted value outside the range of ε.

[0073] After the kernel function is introduced into SVR, it can handle nonlinear regression problems. In this embodiment, the Gaussian kernel function is preferably used.

[0074] In some feasible implementations, the model optimization module 320 is used to: set the root mean square error of the historical data input into the SVR model as the fitness function of the Osprey optimization algorithm; the Osprey optimization algorithm optimizes the regularization coefficient and the kernel parameter in the basic SVR model by minimizing the root mean square error corresponding to the fitness function, thereby completing the process of optimizing the kernel parameters and the regularization coefficient in the basic SVR model, including: setting the initial parameters in the Osprey optimization algorithm, iterating in sequence, reaching the maximum number of iterations or the Osprey optimization algorithm tending to be stable, indicating that the optimal value has been found. Among them, the root mean square error function corresponding to the fitness function is:

[0075]

[0076] Where m is the number of samples, F i_predict is the predicted value, F i_actual is the actual value.

[0077] Specifically, in recent years, the emerging intelligent optimization algorithms have been proven to have the ability to coordinate global exploration and local development, so they have good performance when dealing with some engineering optimization problems. In this embodiment, OOA is used to optimize the kernel parameters and regularization coefficients of SVR.

[0078] The process of changing the population location in OOA consists of two stages: the first stage is when the osprey identifies the location of the fish and catches them (global exploration), and the second stage is when the fish are brought to the appropriate location (local exploitation).

[0079] In the first stage, the osprey detects the location of underwater fish. After determining the location of the fish, it attacks it and preys underwater. The first stage is modeled based on the simulation of this natural behavior of the osprey. For each osprey, the locations of other ospreys with better objective function values ​​in the search space are considered as underwater fish. The fish group for each osprey is expressed by the following formula:

[0080] FP i ={Xk |k∈{1,2,...,N}∧F k <F i}∪{X best};

[0081] Among them, FP i is the fish set of the i-th osprey, F k and F i represents the fitness value corresponding to the position of the kth and ith osprey, N is the population size, X best Is the best Osprey location.

[0082] The update formula for the first stage is as follows:

[0083]

[0084] Among them, x i,j Represents the location information of the Osprey. is the new position of the i-th osprey in the first stage, and r i,j is a random number between [0,1], SF i is the target selected by the ith osprey, SF i,j is its corresponding dimension, I i,j is a random number from the set {1,2}.

[0085]

[0086] Among them, lb j is the lower boundary corresponding to the jth dimension, ub j is the upper boundary corresponding to the j-th dimension.

[0087] In the second phase, after catching a fish, the osprey will take it to a relatively safe location to eat. The second phase of population update in OOA is modeled based on the simulation of this natural behavior of the osprey. The modeling of bringing the fish to the appropriate location causes a small change in the position of the osprey in the search space, which leads to an increase in the OOA's ability to exploit local search.

[0088] The second phase update formula is as follows:

[0089]

[0090] in, is the new position of the i-th osprey in the j-th dimension in the second stage, is its corresponding fitness value, r is a random number between [0,1], t and T are the current number of iterations and the maximum number of iterations respectively.

[0091] The root mean square error of the historical data input to SVR is set as the fitness function of OOA. OOA can optimize the regularization coefficient and kernel parameter of SVR by minimizing the root mean square error. The initial parameters in OOA are set and iterated in sequence. When the maximum number of iterations is reached or the algorithm tends to be stable, it means that the optimal value has been found;

[0092] The formula of the root mean square error function corresponding to the fitness function is as follows:

[0093]

[0094] Where m is the number of samples, F i_predict is the predicted value, F i_actual is the actual value.

[0095] OOA can effectively improve the prediction accuracy of the SVR model and optimize its performance. After obtaining the real-time sensor data provided by the data processing module, the SVR will predict the vehicle's driving path and vehicle status data in the next few seconds based on these data. That is, the SVR module will generate corresponding prediction data based on the input vehicle dynamic data. The driving path prediction data and vehicle status prediction data include: the predicted vehicle speed, predicted acceleration, predicted steering wheel angle, predicted gyroscope data, predicted road curvature, predicted GPS information, predicted ambient light intensity, and predicted distance between the obstacle in front of the vehicle and the vehicle in the next few seconds or a combination thereof.

[0096] In some feasible implementations, the headlight dynamic control module 4 is used to generate a headlight control instruction corresponding to the driving path prediction data and the characteristic data, and the headlight control instruction includes one or a combination of a headlight beam angle control instruction, a headlight brightness control instruction, and a headlight projection spot shape control instruction. Preferably, in combination with Figure 2As shown, the headlight dynamic control module 4 integrates a beam direction adjustment submodule 410, a beam brightness adjustment submodule 420 and a spot shape adjustment submodule 430; the beam direction adjustment submodule 410 is used to generate a control instruction for dynamically adjusting the direction of the headlight beam based on the current driving direction of the vehicle in the feature data and the predicted steering wheel angle in the driving path prediction data; the beam brightness adjustment submodule 420 is used to generate a control instruction for dynamically adjusting the beam brightness based on the predicted vehicle speed and the predicted distance between the obstacle in front of the vehicle and the vehicle in the vehicle state prediction data; the spot shape adjustment submodule 430 is used to generate a control instruction for dynamically adjusting the image data to be projected based on the road information in front of the vehicle sent by the data processing module. Specifically, the beam direction adjustment submodule 410 will generate a control instruction based on the current driving direction of the vehicle and the estimated steering angle, and transmit it to the motor module 5, so as to realize the dynamic adjustment of the beam direction of the headlight. The light beam will automatically shift with the driving direction of the vehicle to ensure that the light beam can illuminate the road area to be traveled when turning or changing lanes. The beam brightness adjustment submodule 420 will dynamically adjust the beam brightness according to the predicted vehicle speed and the distance of the obstacle ahead. When the vehicle is traveling faster, the irradiation distance of the beam will increase. The added real-time prediction mechanism can ensure that the driver has enough reaction time. The spot shape adjustment submodule 430 will dynamically adjust the image data to be projected according to the road ahead information collected by the camera, and transmit it to the headlight projection subsystem 6 for projection.

[0097] In some feasible implementations, the intelligent headlight dynamic control system also includes a real-time monitoring feedback module 7, the output end of the real-time monitoring feedback module 7 is electrically connected to the third input end of the headlight dynamic control module 4, the second output end of the SVR module 3 is electrically connected to the first input end of the real-time monitoring feedback module 7, and the third output end of the data processing module 2 is electrically connected to the second input end of the real-time monitoring feedback module 7; the real-time monitoring feedback module 7 is used to monitor the illumination effect of the headlights in real time through the vehicle-mounted camera, and compare it with the vehicle's driving path prediction data and vehicle status data generated by the SVR module 3 in the next few seconds and the vehicle's front road information sent by the data processing module 2, and send a control instruction to the headlight dynamic control module 4 based on the comparison result, so that the headlight dynamic control module 4 adaptively adjusts the control rules for generating headlight control instructions based on the control instructions, and the illumination effect includes one or a combination of the light intensity, illumination angle and spot shape of the headlight illumination area. Specifically, the real-time monitoring feedback module 7 monitors the lighting conditions of the headlights in real time through the on-board camera. The specific operation steps are as follows: Lighting range monitoring: The camera detects the actual illumination area and brightness value of the headlights, and compares these data with the target value output by the SVR module 3. Error calculation and adjustment: When it is detected that the actual lighting effect deviates greatly from the target value, the real-time monitoring feedback module 7 will send a corresponding control signal to the headlight dynamic control module 4 according to the error value, so that it can dynamically adjust the control rules based on the control signal, and then adjust the beam direction and brightness in real time. For example, when the error value is within the first error range, a first control signal is sent to the headlight dynamic control module 4, and the headlight dynamic control module 4 adjusts the control rules of the beam direction and brightness slightly based on the first control signal; when the error value is within the second error range, a second control signal is sent to the headlight dynamic control module 4, and the headlight dynamic control module 4 adjusts the control rules of the beam direction and brightness significantly based on the second control signal. It should be noted that the adjustment method of the control rule can be independently changed and set based on the actual needs of relevant technicians. That is, the real-time monitoring feedback module 7 can establish a real-time data monitoring mechanism, which will monitor the actual lighting effect of the headlights in real time according to the on-board camera, and feed back the collected information such as the light intensity, angle, shape, etc. of the headlight irradiation area to the headlight dynamic control module. At the same time, the real-time data is compared with the prediction results of the SVR model to detect model anomalies or model failures. And according to the comparison results, the dynamic adjustment and optimization of the headlights are realized to maintain the best adaptability and optimization performance, and finally achieve closed-loop control.

[0098] In some feasible implementations, the motor module 5 is used to drive the headlight system to perform corresponding operations based on the headlight beam angle control instruction. The headlight projection subsystem 6 is used to control the light brightness of the headlight system and the image data to be projected based on the headlight brightness control instruction and the headlight projection spot shape control instruction. Among them, the motor module 5 is mainly responsible for adjusting the angle of the headlight, and the headlight projection subsystem 6 is responsible for projecting the image screen onto the road ahead.

[0099] In the above embodiment, by embedding the Kalman filter algorithm in the data processing module, the data in the data acquisition module can be effectively extracted for real-time denoising, giving a more accurate state estimation. At the same time, the hyperparameters of the SVR are optimized using the swarm intelligence optimization algorithm (Osprey optimization algorithm), which can effectively improve the robustness and generalization of the SVR model, thereby further improving the prediction performance of the SVR. After embedding the prediction algorithm in the intelligent headlight dynamic control system, the intelligence of the system is improved, thereby improving the safety and comfort of night driving. The system and method of the present invention can be widely used in the fields of intelligent assisted driving and unmanned driving, and also provide a new solution for the headlight controller algorithm.

[0100] It is worth mentioning that all modules involved in this embodiment are logical units. In practical applications, a logical unit can be a physical unit, a part of a physical unit, or a combination of multiple physical units. In addition, in order to highlight the innovative part of the present invention, this embodiment does not introduce units that are not closely related to solving the technical problem proposed by the present invention, but this does not mean that there are no other units in this embodiment.

[0101] Example 2

[0102] See also Figure 3 , this embodiment provides a flow chart of an intelligent vehicle light dynamic control method based on support vector regression.

[0103] As an example, the method is applied to the intelligent vehicle light dynamic control system based on support vector regression described in Example 1, and the method includes:

[0104] S310: System initialization.

[0105] S320: Acquire the dynamic data of the vehicle in real time based on the vehicle-mounted sensors through the data acquisition module.

[0106] S330: The data processing module processes the dynamic data of the vehicle based on a Kalman filter algorithm to generate feature data.

[0107] S340: The SVR module creates a basic SVR model by introducing a Gaussian kernel function.

[0108] S350: Optimizing kernel parameters and regularization coefficients in the basic SVR model based on the Osprey optimization algorithm.

[0109] S360: Training the optimized basic SVR model based on the pre-stored training set.

[0110] S370: Input the feature data into the trained SVR model to generate the vehicle's driving path prediction data and vehicle status prediction data in the next few seconds.

[0111] S380: The vehicle light dynamic control module generates a corresponding vehicle light control instruction based on the driving path prediction data and the characteristic data.

[0112] S390: The motor module drives the headlight system to perform corresponding operations based on the beam angle control instruction of the headlight.

[0113] S3100: The vehicle light projection subsystem controls the light brightness of the vehicle light system and the image data to be projected based on the vehicle light brightness control instruction and the vehicle light projection spot shape control instruction.

[0114] S3110: The real-time monitoring feedback module monitors the illumination effect of the headlights in real time through the on-board camera, and compares it with the predicted driving path data and vehicle status data of the vehicle in the next few seconds generated by the SVR module and the road information ahead of the vehicle sent by the data processing module, and sends a control instruction to the headlight dynamic control module based on the comparison result, so that the headlight dynamic control module adaptively adjusts the control rules for generating the headlight control instruction based on the control instruction.

[0115] It is not difficult to find that this embodiment is a method embodiment corresponding to the first embodiment, and this embodiment can be implemented in conjunction with the first embodiment. The relevant technical details mentioned in the first embodiment are still valid in this embodiment, and in order to reduce repetition, they are not repeated here. Accordingly, the relevant technical details mentioned in this embodiment can also be applied in the first embodiment.

[0116] Example 3

[0117] See also Figure 4 An embodiment of the present invention further provides an electronic device, comprising: a memory and a processor; the memory stores at least one program instruction; the processor loads and executes the at least one program instruction to implement the intelligent vehicle light dynamic control method based on support vector regression provided in Example 2.

[0118] The memory 702 and the processor 701 are connected in a bus manner, and the bus may include any number of interconnected buses and bridges, and the bus connects various circuits of one or more processors 701 and the memory 702 together. The bus can also connect various other circuits such as peripheral devices, voltage regulators, and power management circuits, which are well known in the art and are therefore not further described herein. The bus interface provides an interface between the bus and the transceiver. The transceiver can be one element or multiple elements, such as multiple receivers and transmitters, providing a unit for communicating with various other devices on a transmission medium. The data processed by the processor 701 is transmitted on a wireless medium via an antenna, and further, the antenna also receives data and transmits the data to the processor 701.

[0119] The processor 701 is responsible for managing the bus and general processing, and can also provide various functions, including timing, peripheral interfaces, voltage regulation, power management and other control functions. The memory 702 can be used to store data used by the processor 701 when performing operations.

[0120] Example 4

[0121] The embodiment of the present invention further provides a storage medium, on which is stored a method for dynamic control of intelligent headlights based on support vector regression, and when the program for dynamic control of intelligent headlights based on support vector regression is executed, the steps of the method for dynamic control of intelligent headlights based on support vector regression as described above are implemented. Since the storage medium adopts all the technical solutions of all the above embodiments, it has at least all the beneficial effects brought by the technical solutions of the above embodiments, which will not be described one by one here.

[0122] The above is only an embodiment of the present invention. The common sense such as the known specific structure and characteristics in the scheme is not described in detail here. The ordinary technicians in the relevant field know all the common technical knowledge in the technical field of the invention before the application date or priority date, can know all the existing technologies in the field, and have the ability to apply the conventional experimental means before that date. The ordinary technicians in the relevant field can improve and implement this scheme in combination with their own abilities under the enlightenment given by this application. Some typical known structures or known methods should not become obstacles for ordinary technicians in the relevant field to implement this application. It should be pointed out that for those skilled in the art, without departing from the structure of the present invention, several deformations and improvements can be made, which should also be regarded as the scope of protection of the present invention, which will not affect the effect of the implementation of the present invention and the practicality of the patent. The scope of protection required by this application shall be based on the content of its claims, and the specific implementation methods and other records in the specification can be used to interpret the content of the claims.

Claims

1. An intelligent vehicle light dynamic control system based on support vector regression, characterized in that: The intelligent headlight dynamic control system comprises: a data acquisition module, a data processing module, an SVR module, a headlight dynamic control module, a motor module and a headlight projection subsystem, wherein the output end of the data acquisition module is electrically connected to the input end of the data processing module, the first output end of the data processing module is electrically connected to the input end of the SVR module, the second output end of the data processing module is electrically connected to the first input end of the headlight dynamic control module, the first output end of the SVR module is electrically connected to the second input end of the headlight dynamic control module, the output end of the headlight dynamic control module is electrically connected to the motor module and the input end of the headlight projection subsystem respectively, and the motor module and the headlight projection subsystem are externally connected to the headlight system; The data acquisition module is used to obtain the dynamic data of the vehicle in real time based on the vehicle-mounted sensors; The data processing module is used to process the dynamic data of the vehicle based on the Kalman filter algorithm to generate characteristic data; The SVR module integrates a model creation module, a model optimization module, a model training module and a prediction result generation module; The model creation module is used to create a basic SVR model by introducing a Gaussian kernel function; The model optimization module is used to optimize the kernel parameters and regularization coefficients in the basic SVR model based on the Osprey optimization algorithm; The model training module is used to train the optimized basic SVR model based on a pre-stored training set; The prediction result generation module is used to input the feature data into the trained SVR model to generate the vehicle's driving path prediction data and vehicle state prediction data within the next few seconds; The headlight dynamic control module is used to generate a headlight control instruction corresponding to the driving path prediction data and the characteristic data, wherein the headlight control instruction includes one or a combination of a headlight beam angle control instruction, a headlight brightness control instruction and a headlight projection spot shape control instruction; The motor module is used to drive the headlight system to perform corresponding operations based on the beam angle control instruction of the headlight; The vehicle lamp projection subsystem is used to control the light brightness of the vehicle lamp system and the image data to be projected based on the vehicle lamp brightness control instruction and the vehicle lamp projection spot shape control instruction.

2. The intelligent vehicle light dynamic control system based on support vector regression according to claim 1 is characterized in that: The intelligent headlight dynamic control system further comprises a real-time monitoring feedback module, the output end of the real-time monitoring feedback module is electrically connected to the third input end of the headlight dynamic control module, the second output end of the SVR module is electrically connected to the first input end of the real-time monitoring feedback module, and the third output end of the data processing module is electrically connected to the second input end of the real-time monitoring feedback module; The real-time monitoring feedback module is used to monitor the illumination effect of the headlights in real time through the on-board camera, and compare it with the vehicle's driving path prediction data and vehicle status data in the next few seconds generated by the SVR module and the vehicle's front road information sent by the data processing module, and send a control instruction to the headlight dynamic control module based on the comparison result, so that the headlight dynamic control module adaptively adjusts the control rules for generating headlight control instructions based on the control instructions. The illumination effect includes one or a combination of the light intensity, illumination angle and light spot shape of the headlight illumination area.

3. The intelligent vehicle light dynamic control system based on support vector regression according to claim 1, characterized in that: The data acquisition module integrates vehicle-mounted sensors, vehicle-mounted cameras and / or laser radars, and is used to collect vehicle speed, acceleration, steering wheel angle, gyroscope data, road curvature, GPS information, ambient light intensity, speed limit warnings, and one or a combination of the distance between the obstacle in front of the vehicle and the vehicle in real time.

4. The intelligent vehicle light dynamic control system based on support vector regression according to claim 1, characterized in that: The data processing module is used to generate feature data after cleaning, filtering, denoising and data normalization of the dynamic data of the vehicle based on the Kalman filter algorithm; The Kalman filter algorithm is integrated in the Kalman filter, and its state equation is: x k =Ax k-1 +Bu k-1 +ω k-1 4 Among them, x k is the state of the system at time k, A represents the state transfer matrix, B represents the control matrix, and u k-1 is the control input, ω k-1 is the process noise; The observation process describing the system is: z k =Hx k +v k ; Among them, z k is the observed value, H is the observation matrix, describing the relationship between the state and the measured value, v k To measure noise; The Kalman filter performs real-time denoising on the vehicle's dynamic data by combining the prediction model and observations.

5. The intelligent vehicle light dynamic control system based on support vector regression according to claim 1, characterized in that: The objective function of the basic SVR model is: f(x)=ω T Φ(x)+b; Among them, ω is the weight vector, Φ(x) is the feature map of the input data x, the low-dimensional data is mapped to the high-dimensional space through the kernel function, and b is the bias term; The Gaussian kernel function is: Among them, x i For any point in space, x j is the center of the kernel function, σ is the width parameter of the kernel function, ‖x i -x j For x i to x j The Euclidean distance between .

6. The intelligent vehicle light dynamic control system based on support vector regression according to claim 1, characterized in that: The model optimization module is used to: The root mean square error of the historical data input to the SVR model is set as the fitness function of the Osprey optimization algorithm; The Osprey optimization algorithm optimizes the regularization coefficient and the kernel parameter in the basic SVR model by minimizing the root mean square error corresponding to the fitness function, thereby completing the process of optimizing the kernel parameters and the regularization coefficient in the basic SVR model, including: setting the initial parameters in the Osprey optimization algorithm, iterating in sequence, and reaching the maximum number of iterations or the Osprey optimization algorithm tending to be stable, indicating that the optimal value has been found.

7. The intelligent vehicle light dynamic control system based on support vector regression according to claim 6 is characterized in that: The root mean square error function corresponding to the fitness function is: Where m is the number of samples, F i_predict is the predicted value, F i_actual is the actual value.

8. The intelligent vehicle light dynamic control system based on support vector regression according to claim 1, characterized in that: The driving path prediction data and vehicle state prediction data include: one or a combination of the vehicle's predicted speed, predicted acceleration, predicted steering wheel angle, predicted gyroscope data, predicted road curvature, predicted GPS information, predicted ambient light intensity, and predicted distance between obstacles in front of the vehicle and the vehicle in the next few seconds.

9. The intelligent vehicle light dynamic control system based on support vector regression according to claim 8, characterized in that: The vehicle light dynamic control module integrates a beam direction adjustment submodule, a beam brightness adjustment submodule and a light spot shape adjustment submodule; The light beam direction adjustment submodule is used to generate a control instruction for dynamically adjusting the light beam direction of the vehicle based on the current driving direction of the vehicle in the characteristic data and the predicted steering wheel angle in the driving path prediction data; The light beam brightness adjustment submodule is used to generate a control instruction for dynamically adjusting the light beam brightness based on the predicted vehicle speed and the predicted distance between the obstacle in front of the vehicle and the vehicle in the vehicle state prediction data; The light spot shape adjustment submodule is used to generate a control instruction for dynamically adjusting the image data to be projected based on the road information ahead of the vehicle sent by the data processing module.

10. An intelligent vehicle light dynamic control method based on support vector regression, characterized in that: The method is applied to the intelligent vehicle light dynamic control system based on support vector regression as described in any one of claims 1 to 9, and the method comprises: System initialization; The data acquisition module acquires the dynamic data of the vehicle in real time based on the vehicle-mounted sensors; The data processing module processes the dynamic data of the vehicle based on a Kalman filter algorithm to generate characteristic data; The SVR module creates a basic SVR model by introducing a Gaussian kernel function; Optimizing the kernel parameters and regularization coefficients in the basic SVR model based on the Osprey optimization algorithm; Training the optimized basic SVR model based on the pre-stored training set; Inputting the feature data into the trained SVR model to generate the vehicle's driving path prediction data and vehicle status prediction data within the next few seconds; The vehicle light dynamic control module generates a corresponding vehicle light control instruction based on the driving path prediction data and the characteristic data; The motor module drives the vehicle lamp system to perform corresponding operations based on the beam angle control instruction of the vehicle lamp; The vehicle lamp projection subsystem controls the light brightness of the vehicle lamp system and the image data to be projected based on the vehicle lamp brightness control instruction and the vehicle lamp projection spot shape control instruction; The real-time monitoring feedback module monitors the illumination effect of the headlights in real time through the on-board camera, and compares it with the predicted driving path data and vehicle status data of the vehicle in the next few seconds generated by the SVR module and the road information ahead of the vehicle sent by the data processing module, and sends a control instruction to the headlight dynamic control module based on the comparison result, so that the headlight dynamic control module adaptively adjusts the control rules for generating the headlight control instruction based on the control instruction.

Images