A method and system for automatically adjusting and controlling vehicle-mounted light sources based on environmental perception

Abstract

The present invention discloses an automatic adjustment control method and system for vehicle-mounted light sources based on environmental perception, which relates to the field of vehicle control technology, including collecting vehicle data and surrounding environment data, extracting perception features; constructing embedded features according to the perception feature content, respectively calculating the classification loss and correlation loss of the embedded features, generating initial light source adjustment items based on the perception features, and optimizing the initial light source adjustment items according to the embedded features to generate a light source adjustment plan; obtaining the driver's driving data, judging the driving state and adjusting the light source in the car according to the driving state; collecting the driver's feedback data on the light source adjustment plan to optimize the light source adjustment plan, and generating an optimization record for storage. The present invention effectively improves the adaptability and responsiveness of the vehicle-mounted light source to the environment, enhances the flexibility of the vehicle-mounted light source adjustment, and simultaneously adjusts the vehicle-mounted light source by identifying the driver's state, effectively improving driving safety.

Description

Technical Field

[0001] The present invention relates to the technical field of vehicle control, and in particular to an automatic adjustment control method and system for a vehicle-mounted light source based on environment perception. Background Art

[0002] Driven by the development of vehicle automatic control and intelligence, vehicle-mounted light source adjustment technology has undergone a multi-level evolution from fixed light source control to intelligent automatic adjustment. Early vehicle-mounted light source systems mostly used preset modes or manual control to adjust the brightness of the light source, which made it difficult to achieve real-time dynamic response. In recent years, with the rapid development of environmental perception technology and artificial intelligence, vehicle-mounted light source systems have gradually integrated sensing and intelligent control functions, and can perform a certain degree of automatic adjustment based on environmental factors such as vehicle speed, external light intensity, and weather conditions. These systems usually rely on a variety of sensors to collect environmental information inside and outside the vehicle in real time (such as light intensity, vehicle speed, weather conditions, etc.) to generate automatic adjustment strategies for specific scenarios. However, most of the existing light source adjustment systems rely on a single or a small number of environmental features for light source adjustment, and fail to effectively integrate factors such as the dynamic information of the vehicle and environmental change information, resulting in the system's response being not sensitive enough, the adjustment process being not flexible enough, and lacking personalized and adaptive optimization functions. It is difficult to update the light source adjustment strategy in real time based on the driver's driving conditions, which reduces user experience and driving safety. Summary of the invention

[0003] In view of the problems existing in the above-mentioned existing vehicle-mounted light source automatic adjustment control method and system based on environmental perception, the present invention is proposed.

[0004] Therefore, the problem to be solved by the present invention is that the existing light source adjustment system is not sensitive enough in response, the adjustment process is not flexible enough, and lacks personalized and adaptive optimization functions.

[0005] To solve the above technical problems, the present invention provides the following technical solutions: a method for automatic adjustment and control of vehicle-mounted light sources based on environmental perception, which includes collecting vehicle data and surrounding environment data, and extracting perception features; constructing embedded features according to the perception feature content, respectively calculating the classification loss and correlation loss of the embedded features, generating initial light source adjustment items based on the perception features, and optimizing the initial light source adjustment items according to the embedded features to generate a light source adjustment plan; obtaining the driver's driving data, judging the driving state and adjusting the in-vehicle light source according to the driving state; collecting the driver's feedback data on the light source adjustment plan to optimize the light source adjustment plan, and generating an optimization record for storage.

[0006] As a preferred solution of the method for automatic adjustment and control of vehicle-mounted light sources based on environmental perception described in the present invention, wherein: the collection of vehicle data and surrounding environment data and extraction of perception features refers to installing a vehicle speed sensor on the vehicle to obtain vehicle speed data, setting a vehicle speed threshold to divide the vehicle speed level and generate a vehicle speed level label, collecting the vehicle's external environment light data in real time through a light sensor, processing the light data using a Kalman filter, synchronously collecting weather data through a meteorological sensor, identifying the weather through a weather label, and combining the speed level label, light data and weather label to generate a vehicle perception feature vector.

[0007] As a preferred solution of the vehicle-mounted light source automatic adjustment control method based on environmental perception described in the present invention, wherein: the embedding feature is constructed according to the perceived feature content, and the classification loss and correlation loss of the embedded feature are calculated respectively, which refers to randomly generating an embedded feature vector according to each content in the perceived feature vector, initializing the clustering center, using a clustering algorithm to perform clustering analysis on the generated embedded feature vector, and recalculating the category center vector according to the clustering result. :

[0008]

[0009] in is the number of embedded feature vectors in the kth cluster category, is the i-th embedded feature vector in the k-th cluster category;

[0010] According to the category center vector and embedding feature vector Calculate the classification loss of the embedded feature vector :

[0011]

[0012] Where m is the total number of embedded feature vectors, L is the total number of cluster categories, is the category center vector The transpose of is the weight of the i-th embedded feature vector in the k-th cluster category;

[0013] Calculate the cosine similarity between embedded feature vectors , set the similarity threshold, if the cosine similarity between the embedded feature vectors If the similarity is greater than the threshold, it is defined as being in the same semantic neighborhood. The number of empty matrix rows and columns is defined according to the total number of embedded feature vectors. If the embedded feature vectors are in the same semantic neighborhood, their corresponding position values ​​in the similarity matrix are is 1, otherwise 0, and each pair of embedded feature vectors Fill in the corresponding positions in the empty matrix to form a similarity matrix S;

[0014] Through the cosine similarity of the similarity matrix S and the embedded feature vector Calculate the relevance loss of the embedded feature vector :

[0015]

[0016] Where m is the total number of embedded feature vectors.

[0017] As a preferred solution of the vehicle-mounted light source automatic adjustment control method based on environmental perception described in the present invention, wherein: the initial light source adjustment item is generated based on the perception feature, and the initial light source adjustment item is optimized to generate the light source adjustment scheme according to the embedded feature, which means constructing a multi-layer perception network, setting the multi-layer perception network input as the perception feature, and the output as the initial adjustment amount of the vehicle-mounted light source, including brightness, color temperature and angle adjustment amount, using the embedded feature vector as training data, using the training data to train the multi-layer perception network, inputting the perception feature into the trained multi-layer perception network to generate the initial adjustment amount of the vehicle-mounted light source, and optimizing the initial adjustment amount of the vehicle-mounted light source based on the embedded feature vector:

[0018]

[0019] Where Q is the initial adjustment of the vehicle light source, including brightness, color temperature and angle adjustment. Optimize the adjustment amount for vehicle-mounted light sources, is the adjustment coefficient, is the adjustment weight of the i-th embedded feature vector;

[0020] The optimized adjustment quantities of the vehicle-mounted light sources are combined to form an adjustment vector U, and the adjustment loss of the adjustment vector U is constructed based on the Allen-Cahn energy functional. :

[0021]

[0022] in is the gradient of the initial adjustment vector, is the smoothing parameter, x is the integral variable;

[0023] Comprehensive classification loss, correlation loss and adjustment loss to construct an optimized loss function :

[0024]

[0025] The adjustment vector U is iteratively adjusted by the gradient descent method, and the optimization loss function is calculated after each adjustment. Until the loss function is optimized The iteration is stopped when the difference between the values ​​of is less than the set threshold during two consecutive iterations, and the final optimized adjustment vector is output. ;

[0026] Extract the final optimized adjustment vector The brightness, color temperature and angle adjustment are combined to form a vehicle light source adjustment solution.

[0027] As a preferred solution of the vehicle-mounted light source automatic adjustment control method based on environment perception of the present invention, wherein: the acquisition of the driver's driving data, judging the driving state and adjusting the in-vehicle light source according to the driving state refers to using a visual sensor to collect the driver's facial data and motion data when driving, and extracting the driver's facial features and motion features based on an independent component analysis algorithm, and combining the facial features and motion features to form a driving feature;

[0028] Construct a convolutional neural network, set the input as driving characteristics, and the output as the driver's driving status, including normal, fatigue, and abnormal. Use the training data to train the convolutional neural network;

[0029] Input the driver's driving characteristics into the trained convolutional neural network to obtain the driver's driving state;

[0030] If the driver is in normal state, the vehicle light source remains normal;

[0031] If the driver is fatigued, the light source in the car will be adjusted to a cool color temperature and flash synchronously to remind the driver;

[0032] If the driver is in an abnormal state, the light source inside the car will remain on and the turn signals on both sides of the vehicle will flash synchronously to remind other vehicles to give way.

[0033] As a preferred solution of the method for automatic adjustment and control of vehicle-mounted light sources based on environmental perception described in the present invention, wherein: the collection of the driver's feedback data on the light source adjustment plan to optimize the light source adjustment plan refers to collecting the driver's feedback data on the vehicle-mounted light source adjustment plan after the vehicle-mounted light source plan is generated and implemented, and using Q-learning to optimize the vehicle-mounted light source adjustment plan generation process based on the driver's feedback data.

[0034] As a preferred solution of the method for automatic adjustment and control of vehicle-mounted light sources based on environmental perception described in the present invention, wherein: the generation of optimization records for storage means that after each collection of driver feedback data to optimize the vehicle-mounted light source solution generation process, the optimization records are generated and stored in the database with a collection timestamp, and the database regularly uploads the stored data to the cloud for backup.

[0035] Another object of the present invention is to provide an automatic adjustment control system for vehicle-mounted light sources based on environmental perception, which includes a data collection module for collecting vehicle data and environmental data collected by deployed sensors and extracting perception features;

[0036] An adjustment optimization module is used to generate embedded features based on the perception features, generate an initial vehicle light source adjustment amount, and simultaneously optimize the initial vehicle light source adjustment amount based on the embedded features to obtain a vehicle light source adjustment solution;

[0037] A state recognition module is used to collect the driver's facial data and motion data to extract driving characteristics, identify the driver's driving state based on the driving characteristics, and adjust the vehicle-mounted light source based on the driving state;

[0038] The optimization storage module is used to optimize the vehicle light source adjustment solution generation process according to the driver's feedback data and generate optimization records for storage.

[0039] A computer device comprises: a memory and a processor; the memory stores a computer program, and the processor implements the steps of the above-mentioned method for automatically adjusting and controlling a vehicle-mounted light source based on environmental perception when executing the computer program.

[0040] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of the above-mentioned method for automatically adjusting and controlling a vehicle-mounted light source based on environmental perception.

[0041] The beneficial effects of the present invention are as follows: the present invention extracts perception features by collecting vehicle data and environmental data, and constructs embedded features based on the perception features, and optimizes the initial light source adjustment items generated by the perception feature vector based on the embedded features to generate a light source adjustment scheme, thereby effectively improving the adaptability and responsiveness of the vehicle-mounted light source to the environment, improving the flexibility of adjusting the vehicle-mounted light source, and simultaneously adjusting the vehicle-mounted light source by identifying the driver's status, thereby effectively improving driving safety. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0043] Figure 1 The figure is a flow chart of an automatic adjustment control method of an on-board light source based on environmental perception.

[0044] Figure 2 This is a schematic diagram of the structure of the automatic adjustment control system of the vehicle light source based on environmental perception. DETAILED DESCRIPTION

[0045] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific embodiments of the present invention are described in detail below in conjunction with the accompanying drawings.

[0046] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0047] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor is it a separate or selective embodiment that is mutually exclusive with other embodiments.

[0048] Example 1, reference Figure 1 , is the first embodiment of the present invention, which provides an automatic adjustment control method for a vehicle-mounted light source based on environment perception. The automatic adjustment control method for a vehicle-mounted light source based on environment perception includes:

[0049] S1, collect vehicle data and surrounding environment data, and extract perception features;

[0050] Specifically, collecting vehicle data and surrounding environment data and extracting perception features means installing a speed sensor on the vehicle to obtain speed data, setting a speed threshold to divide the speed level and generate a speed level label, for example, 1 represents low speed, 2 represents medium speed, and 3 represents high speed, collecting the vehicle's external environment light data in real time through a light sensor, using a Kalman filter to process the light data, and simultaneously collecting weather data through a meteorological sensor, identifying the weather through a weather label, for example, 1 represents a sunny day, 2 represents a rainy day, 3 represents a cloudy day, 4 represents foggy weather, etc., and combining the speed level label, light data and weather label to generate a vehicle perception feature vector.

[0051] Through multi-source data collection and fusion, the system can acquire and integrate key data of the vehicle and environment in real time in a complex and changeable driving environment. This data fusion method improves the light source adjustment system's sensitive response to the external environment, allowing the on-board light source to provide highly adaptable adjustment solutions under low light, high-speed driving, bad weather and other conditions. By labeling the vehicle speed level, the system can generate personalized light source adjustment solutions under different speed conditions. For example, when driving at high speed, the system will generate a higher brightness and wider range of light source adjustment solutions based on the high-speed label, and reduce the brightness at low speed or parking to avoid interference with surrounding vehicles. This speed-based system The level-based light source adjustment scheme not only improves driving comfort, but also enhances safety. The stability of lighting data is crucial for dynamic light source adjustment. The introduction of Kalman filtering ensures the smoothness of lighting data, allowing the system to avoid frequent light source fluctuation adjustments in scenes such as tunnel entrances and exits, and changes in light in the morning and evening, thereby improving the user's driving experience and adaptability to light source adjustment. Based on weather labeling, the system can automatically identify complex weather conditions and make corresponding light source adjustments to cope with changes in visibility, especially in sudden severe weather such as haze and rain. The system can respond quickly and adapt to changes, greatly improving the driving safety of the vehicle.

[0052] S2. Construct embedded features according to the perceived feature content, calculate the classification loss and correlation loss of the embedded features respectively, generate initial light source adjustment items based on the perceived features, and optimize the initial light source adjustment items according to the embedded features to generate a light source adjustment solution;

[0053] Specifically, embedding features are constructed according to the perceptual feature content, and the classification loss and correlation loss of the embedded features are calculated respectively. This means that an embedded feature vector is randomly generated according to each content in the perceptual feature vector. There are multiple embedded features, each of which represents a perceptual feature category. The clustering center is initialized, and a clustering algorithm is used to perform cluster analysis on the generated embedded feature vector. The category center vector is recalculated based on the clustering results. :

[0054]

[0055] in is the number of embedded feature vectors in the kth cluster category, is the i-th embedded feature vector in the k-th cluster category;

[0056] According to the category center vector and embedding feature vector Calculate the classification loss of the embedded feature vector :

[0057]

[0058] Where m is the total number of embedded feature vectors, L is the total number of cluster categories, is the category center vector The transpose of is the weight of the i-th embedded feature vector in the k-th cluster category;

[0059] Calculate the cosine similarity between embedded feature vectors , set the similarity threshold, if the cosine similarity between the embedded feature vectors If the similarity is greater than the threshold, it is defined as being in the same semantic neighborhood. The number of empty matrix rows and columns is defined according to the total number of embedded feature vectors. If the embedded feature vectors are in the same semantic neighborhood, their corresponding position values ​​in the similarity matrix are is 1, otherwise 0, and each pair of embedded feature vectors Fill in the corresponding positions in the empty matrix to form a similarity matrix S;

[0060] Through the cosine similarity of the similarity matrix S and the embedded feature vector Calculate the relevance loss of the embedded feature vector :

[0061]

[0062] Where m is the total number of embedded feature vectors.

[0063] The generation and clustering of embedded features not only improves the data processing efficiency, but also enhances the system's ability to recognize different environmental conditions. Through the diversification and clustering analysis of embedded features, the system can achieve dynamic adaptation of light source adjustment in different environments. For example, when driving at night and in foggy weather, the system can automatically adjust the brightness and color temperature of the light source to adapt to the needs of different environments, thereby improving driving safety and visual experience. By optimizing the classification loss, the embedded feature vectors can be more closely clustered to their corresponding category centers, improving the accuracy of feature classification. This classification loss optimization ensures that the representative features of each category are not interfered with by other categories, making the light source adjustment more targeted. The minimization of classification loss enables the system to automatically adjust the light source in different environments, such as adjusting the light source brightness and illumination range when driving at high speeds to provide a clearer view of the front, and similar The introduction of the degree matrix can effectively capture the internal correlation of the embedded feature vectors, which helps the system understand the similarity of features of different categories, and then optimize the light source adjustment plan. For example, at low speed, if the system detects weather conditions with similar semantics to the current ambient light intensity, the system will generate similar light source adjustment plans, reduce unnecessary adjustment operations, and improve system efficiency and energy saving effects. By minimizing the correlation loss, the system can ensure the internal consistency of similar features, so that similar embedded feature vectors can be more closely linked together. This feature optimization method can make the light source adjustment plan more stable, especially under environmental conditions with strong continuity, reduce adjustment fluctuations, thereby enhancing driving safety and comfort. The optimization of correlation loss provides a guarantee for the stability of the light source adjustment plan, which is especially suitable for light source control needs in complex weather or environments.

[0064] Furthermore, generating initial light source adjustment items based on perception features, and optimizing the initial light source adjustment items according to embedded features to generate a light source adjustment scheme refers to constructing a multi-layer perception network, setting the multi-layer perception network input as the perception features, and outputting the initial adjustment amount of the vehicle light source, including brightness, color temperature and angle adjustment amount, using the embedded feature vector as training data, using the training data to train the multi-layer perception network, inputting the perception features into the trained multi-layer perception network to generate the initial adjustment amount of the vehicle light source, and optimizing the initial adjustment amount of the vehicle light source based on the embedded feature vector:

[0065]

[0066] Where Q is the initial adjustment of the vehicle light source, including brightness, color temperature and angle adjustment. Optimize the adjustment amount for vehicle-mounted light sources, is the adjustment coefficient, which is set through experience. is the adjustment weight of the i-th embedded feature vector, which is set empirically;

[0067] The optimized adjustment quantities of the vehicle-mounted light sources are combined to form an adjustment vector U, and the adjustment loss of the adjustment vector U is constructed based on the Allen-Cahn energy functional. :

[0068]

[0069] in is the gradient of the initial adjustment vector, is the smoothing parameter, x is the integral variable;

[0070] Comprehensive classification loss, correlation loss and adjustment loss to construct an optimized loss function :

[0071]

[0072] The adjustment vector U is iteratively adjusted by the gradient descent method, and the optimization loss function is calculated after each adjustment. Until the loss function is optimized The iteration is stopped when the difference between the values ​​of is less than the set threshold during two consecutive iterations, and the final optimized adjustment vector is output. ;

[0073] Extract the final optimized adjustment vector The brightness, color temperature and angle adjustment are combined to form a vehicle light source adjustment solution.

[0074] Through the multi-layer perception network, the system can make full use of multi-dimensional environmental characteristics to predict the initial adjustment amount. Compared with the traditional rule-based light source adjustment, the nonlinear mapping ability of the neural network makes the light source adjustment more in line with actual needs. For example, when entering a tunnel or the weather changes suddenly, the system can adjust the brightness and color temperature in real time to ensure a clear driving field of vision and improve driving safety. The introduction of embedded feature vectors realizes the intelligent trade-off of different environmental factors, so that the adjustment amount can not only adapt to external light and weather changes, but also consider the driver's personalized needs. For example, in rainy or foggy conditions, the system can automatically enhance the brightness and expand the illumination range, thereby improving visibility and driving safety. Through the Allen-Cahn energy functional smooth adjustment process, the vehicle-mounted light source can be gradually adjusted during continuous environmental changes (such as day to night, tunnel entry and exit, etc.), avoiding sudden stimulation to the driver's vision. , which significantly improves the comfort and safety of the driving experience. The combination of multiple loss functions can balance the impact of different factors on light source adjustment, ensuring the accuracy and stability of the adjustment. For example, when driving in mountainous areas where lighting conditions change dramatically, the system can balance different environmental requirements, maintain the smoothness of light source adjustment, and prevent unstable driving vision caused by improper adjustment. The application of the gradient descent method realizes the dynamic optimization of the adjustment vector, enabling the system to adaptively approach the optimal light source adjustment scheme. The system can gradually adjust the brightness, color temperature and angle according to the current environmental conditions to ensure the real-time and stability of the adjustment scheme. The final light source adjustment scheme extracted and generated can adapt to environmental requirements in real time, ensure dynamic adjustment under different weather, vehicle speed or light conditions, and provide the driver with a clear and safe field of view. At the same time, this scheme also has good adaptability and stability and can be used in a variety of complex driving environments.

[0075] S3, obtaining the driver's driving data, determining the driving state and adjusting the interior light source according to the driving state;

[0076] Specifically, obtaining the driver's driving data, judging the driving state and adjusting the in-vehicle light source according to the driving state means using a visual sensor to collect the driver's facial data and motion data while driving, and extracting the driver's facial features and motion features based on an independent component analysis algorithm, and combining the facial features and motion features to form a driving feature;

[0077] Construct a convolutional neural network, set the input as driving characteristics, and the output as the driver's driving status, including normal, fatigue, and abnormal. Use the training data to train the convolutional neural network;

[0078] Input the driver's driving characteristics into the trained convolutional neural network to obtain the driver's driving state;

[0079] If the driver is in normal state, the vehicle light source remains normal;

[0080] If the driver is fatigued, the light source in the car will be adjusted to a cool color temperature and flash synchronously to remind the driver;

[0081] If the driver is in an abnormal state, the light source inside the car will remain on and the turn signals on both sides of the vehicle will flash synchronously to remind other vehicles to give way.

[0082] By collecting the driver's facial and motion data in multiple dimensions, the system can monitor the driver's status in all directions during driving and capture subtle changes in the driver. This data collection method provides comprehensive information input for driving status judgment, making the system more accurate in detecting driver fatigue, distraction or abnormal behavior, effectively improving driving safety. The driving features extracted by the ICA algorithm are independent and information-rich, and can clearly reflect the driver's real-time status. This process effectively reduces data redundancy and noise interference, ensuring the validity and accuracy of feature data, which enables the system to capture subtle changes in the driver's status in complex driving scenarios and provide high-quality data support for driving status judgment. By training CNN, the system can accurately identify the different states of the driver, especially in It can quickly detect and respond to driver fatigue or abnormal behavior. Compared with traditional state recognition methods, CNN has higher classification accuracy, can handle complex state characteristics, and is more adaptable. This process provides a high-precision recognition method for real-time monitoring of driving status, making the system intelligent and adaptive. The intelligent light source adjustment method effectively improves driving safety. When the driver is fatigued, the cold color temperature light source and the flashing effect can stimulate the visual nerves and enhance the driver's alertness, thereby reducing the driving risk caused by fatigue. In abnormal conditions, the external warning lights are turned on so that surrounding vehicles can promptly identify the status of the vehicle ahead to avoid collisions. This multi-level light source adjustment mechanism realizes the combination of internal reminders and external warnings for the driver, and improves the safety protection function of the entire vehicle.

[0083] S4. Collecting the driver's feedback data on the light source adjustment scheme to optimize the light source adjustment scheme, and generating an optimization record for storage;

[0084] Specifically, collecting the driver's feedback data on the light source adjustment plan to optimize the light source adjustment plan means collecting the driver's feedback data on the vehicle light source adjustment plan after generating and implementing the vehicle light source adjustment plan, and using Q-learning to optimize the vehicle light source adjustment plan generation process based on the driver's feedback data.

[0085] The collection of feedback data enables the system to understand whether the current adjustment scheme is suitable for the driver's needs, especially in long-term driving, bad weather or other complex environments. By recording the driver's actual response, the system can capture the adaptability of the light source in different driving environments, providing a reliable basis for subsequent optimization. In addition, the real-time collection of feedback data can help the system make quick adjustments when the driver's preferences change or the environmental conditions change, thereby improving the personalization and flexibility of the driving experience. Through the Q-learning algorithm, the system can learn the optimal adjustment strategy from the driver's feedback. With the accumulation of data and the update of the Q value, the system continuously optimizes the selection of the light source adjustment scheme to ensure that the most suitable light source settings are provided to the driver in different environments. This adaptive learning mechanism based on feedback data greatly improves the personalization of the light source adjustment scheme, making the system more flexible in dealing with different driver needs and complex environmental changes. For example, when driving long distances at night, if the driver prefers lower brightness, the system can adjust the scheme based on Q-learning to provide the driver with a more comfortable lighting environment.

[0086] Furthermore, generating optimization records for storage means that after each collection of driver feedback data to optimize the vehicle light source solution generation process, the optimization records are generated by collecting timestamps and stored in the database, and the database regularly uploads the stored data to the cloud for backup.

[0087] Example 2, reference Figure 2 , which is the second embodiment of the present invention, and which is different from the previous embodiment, provides a vehicle-mounted light source automatic adjustment control system based on environment perception, which includes:

[0088] A data collection module, used to collect vehicle data and environmental data from deployed sensors and extract perception features;

[0089] An adjustment optimization module is used to generate embedded features based on the perception features, generate an initial vehicle light source adjustment amount, and simultaneously optimize the initial vehicle light source adjustment amount based on the embedded features to obtain a vehicle light source adjustment solution;

[0090] A state recognition module is used to collect the driver's facial data and motion data to extract driving characteristics, identify the driver's driving state based on the driving characteristics, and adjust the vehicle-mounted light source based on the driving state;

[0091] The optimization storage module is used to optimize the vehicle light source adjustment solution generation process according to the driver's feedback data and generate optimization records for storage.

[0092] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc., which can store program codes.

[0093] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in conjunction with such instruction execution systems, devices or apparatuses. For the purposes of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in conjunction with such instruction execution systems, devices or apparatuses.

[0094] More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection with one or more wires (electronic device), a portable computer disk case (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be a paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, deciphering or, if necessary, processing in another suitable manner, and then stored in a computer memory.

[0095] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or a combination thereof: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0096] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A method for automatically adjusting and controlling a vehicle-mounted light source based on environmental perception, characterized in that: include, Collect vehicle data and surrounding environment data and extract perception features; Construct embedded features according to the perceived feature content, calculate the classification loss and correlation loss of the embedded features respectively, generate initial light source adjustment items based on the perceived features, and optimize the initial light source adjustment items according to the embedded features to generate a light source adjustment solution; Obtain the driver's driving data, determine the driving status and adjust the interior lighting according to the driving status; Collect the driver's feedback data on the light source adjustment plan to optimize the light source adjustment plan, and generate optimization records for storage; The method of constructing embedded features according to the perceived feature content and calculating the classification loss and correlation loss of the embedded features respectively refers to randomly generating an embedded feature vector according to each content in the perceived feature vector, initializing the cluster center, performing cluster analysis on the generated embedded feature vector using a clustering algorithm, and recalculating the category center vector according to the clustering result. : in is the number of embedded feature vectors in the kth cluster category, is the i-th embedded feature vector in the k-th cluster category; According to the category center vector and embedding feature vector Calculate the classification loss of the embedded feature vector : Where m is the total number of embedded feature vectors, L is the total number of cluster categories, is the category center vector The transpose of is the weight of the i-th embedded feature vector in the k-th cluster category; Calculate the cosine similarity between embedded feature vectors , set the similarity threshold, if the cosine similarity between the embedded feature vectors If the similarity is greater than the threshold, it is defined as being in the same semantic neighborhood. The number of empty matrix rows and columns is defined according to the total number of embedded feature vectors. If the embedded feature vectors are in the same semantic neighborhood, their corresponding position values ​​in the similarity matrix are is 1, otherwise 0, and each pair of embedded feature vectors Fill in the corresponding positions in the empty matrix to form a similarity matrix S; Through the cosine similarity of the similarity matrix S and the embedded feature vector Calculate the relevance loss of the embedded feature vector : Where m is the total number of embedded feature vectors; The generating of initial light source adjustment items based on perception features, and optimizing the initial light source adjustment items according to embedded features to generate a light source adjustment scheme refers to constructing a multi-layer perception network, setting the input of the multi-layer perception network as perception features, and outputting the initial adjustment amount of the vehicle light source, including brightness, color temperature and angle adjustment amount, using the embedded feature vector as training data, using the training data to train the multi-layer perception network, inputting the perception features into the trained multi-layer perception network to generate the initial adjustment amount of the vehicle light source, and optimizing the initial adjustment amount of the vehicle light source based on the embedded feature vector: Where Q is the initial adjustment of the vehicle light source, including brightness, color temperature and angle adjustment. Optimize the adjustment amount for vehicle-mounted light sources, is the adjustment coefficient, is the adjustment weight of the i-th embedded feature vector; The optimized adjustment quantities of the vehicle-mounted light sources are combined to form an adjustment vector U, and the adjustment loss of the adjustment vector U is constructed based on the Allen-Cahn energy functional. : in is the gradient of the initial adjustment vector, is the smoothing parameter, x is the integral variable; Comprehensive classification loss, correlation loss and adjustment loss to construct an optimized loss function : The adjustment vector U is iteratively adjusted by the gradient descent method, and the optimization loss function is calculated after each adjustment. Until the loss function is optimized The iteration is stopped when the difference between the values ​​of is less than the set threshold during two consecutive iterations, and the final optimized adjustment vector is output. ; Extract the final optimized adjustment vector The brightness, color temperature and angle adjustment are combined to form a vehicle light source adjustment solution.

2. The method for automatically adjusting and controlling a vehicle-mounted light source based on environmental perception as claimed in claim 1, characterized in that: The collecting of vehicle data and surrounding environment data and extracting perception features refers to installing a speed sensor on the vehicle to obtain speed data, setting a speed threshold to divide speed levels and generate a speed level label, collecting the vehicle's external environment light data in real time through a light sensor, processing the light data using a Kalman filter, synchronously collecting weather data through a meteorological sensor, identifying the weather through a weather label, and combining the speed level label, light data and weather label to generate a vehicle perception feature vector.

3. The method for automatically adjusting and controlling a vehicle-mounted light source based on environmental perception as claimed in claim 2, characterized in that: The obtaining of the driver's driving data, judging the driving state and adjusting the in-vehicle light source according to the driving state refers to using a visual sensor to collect the driver's facial data and motion data while driving, extracting the driver's facial features and motion features based on an independent component analysis algorithm, and combining the facial features and motion features to form a driving feature; Construct a convolutional neural network, set the input as driving characteristics, and the output as the driver's driving status, including normal, fatigue, and abnormal. Use the training data to train the convolutional neural network; Input the driver's driving characteristics into the trained convolutional neural network to obtain the driver's driving state; If the driver is in normal state, the vehicle light source remains normal; If the driver is fatigued, the light source in the car will be adjusted to a cool color temperature and flash synchronously to remind the driver; If the driver is in an abnormal state, the light source inside the car will remain on and the turn signals on both sides of the vehicle will flash synchronously to remind other vehicles to give way.

4. The method for automatically adjusting and controlling a vehicle-mounted light source based on environmental perception as claimed in claim 3, characterized in that: The collecting of the driver's feedback data on the light source adjustment scheme to optimize the light source adjustment scheme refers to collecting the driver's feedback data on the vehicle-mounted light source adjustment scheme after generating and implementing the vehicle-mounted light source adjustment scheme, and optimizing the vehicle-mounted light source adjustment scheme generation process using Q-learning based on the driver's feedback data.

5. The method for automatically adjusting and controlling a vehicle-mounted light source based on environment perception as claimed in claim 4, characterized in that: Generating optimization records for storage means that after each collection of driver feedback data to optimize the vehicle light source solution generation process, the optimization records are generated and stored in the database with a timestamp, and the database regularly uploads the stored data to the cloud for backup.

6. An automatic adjustment control system for vehicle light sources based on environment perception, used to implement the automatic adjustment control method for vehicle light sources based on environment perception as claimed in any one of claims 1 to 5, characterized in that: include, A data collection module, which is used to deploy sensors to collect vehicle data and environmental data and extract perception features; An adjustment optimization module is used to generate embedded features based on the perception features, generate an initial vehicle light source adjustment amount, and simultaneously optimize the initial vehicle light source adjustment amount based on the embedded features to obtain a vehicle light source adjustment solution; A state recognition module is used to collect the driver's facial data and motion data to extract driving characteristics, identify the driver's driving state based on the driving characteristics, and adjust the vehicle-mounted light source based on the driving state; The optimization storage module is used to optimize the vehicle light source adjustment solution generation process according to the driver's feedback data and generate optimization records for storage.

7. A computer device comprising: Memory and processor; The memory stores a computer program, characterized in that: when the processor executes the computer program, the steps of the vehicle-mounted light source automatic adjustment control method based on environmental perception described in any one of claims 1 to 5 are implemented.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method for automatic adjustment and control of a vehicle-mounted light source based on environment perception as claimed in any one of claims 1 to 5 are implemented.

Images