Vehicle lamp dynamic control method and system based on graph neural network and state machine

Abstract

The invention provides a vehicle light dynamic control method and system based on a graph neural network and a state machine, and relates to the technical field of vehicle light control, and the method comprises the steps: obtaining a vehicle driving state, a vehicle inclination angle change and external environment data, converting all the data into graph structure data, and storing the graph structure data; modeling a state machine in a self-adaptive vehicle lamp control algorithm into a directed graph, and forming a vehicle-environment interaction graph structure by taking the state of the vehicle as a node and taking the external environment data as an edge; inputting a vehicle-environment interaction graph structure into a graph neural network model, learning a dependency relationship among nodes in a state machine graph and state representation of each node, obtaining the influence of the dependency relationship among the nodes on the rotation angle of a vehicle motor and vehicle lamp adjustment, dynamically adjusting three modes C, V and E of a self-adaptive vehicle lamp control algorithm, and obtaining the self-adaptive vehicle lamp control algorithm. According to the classification result, the current vehicle lamp irradiation angle and the motor rotation angle are controlled, and the optimal matching state with the current environment is ensured.

Description

Technical Field

[0001] The present disclosure relates to the technical field of vehicle lighting control, and in particular to a vehicle lighting dynamic control method and system based on a graph neural network and a state machine. Background Art

[0002] The statements in this section merely provide background information related to the present disclosure and do not necessarily constitute prior art.

[0003] Traditional headlight ALS algorithms play an important role in headlight control, but their decisions are mainly based on preset rules and patterns, lacking keen perception and adaptability to real-time environmental changes. With the development of intelligent vehicles, higher requirements are placed on headlight systems, including fine perception of complex road environments, dynamic adjustment of lighting modes, and improved energy efficiency and safety.

[0004] There are relatively few artificial intelligence algorithms in the field of headlight algorithms, among which graph neural networks are used even less. The traditional ALS algorithm is generally implemented by accepting CAN signals, establishing a corresponding state machine system, and establishing rules to output the rotation angle of the headlight motor; and the traditional static pitch adjustment function of vehicle headlights means that when the vehicle is stationary (vehicle speed = 0km / h), the body load and load position change, resulting in changes in the body inclination angle and the angle and position of the headlights. The headlight algorithm system should adjust the headlights up and down to the correct angle.

[0005] Therefore, the traditional ALS algorithm system for headlights is a rule-based system, which still has the following deficiencies and problems:

[0006] Traditional ALS algorithm systems mainly rely on preset rules, which are usually based on assumptions about specific scenarios, such as adjusting the angle of headlights when the vehicle turns. However, such preset rules cannot respond to complex and changing driving environments in real time, such as the sudden appearance of pedestrians, animals, or non-standard road signs. The lack of keen perception and adaptability to real-time environmental changes results in the system's inability to respond promptly and appropriately when faced with emergencies. When designing a rule-based ALS system, the rule set is fixed and difficult to modify or expand once set. However, the actual driving environment is dynamically changing, including weather, lighting, traffic conditions, etc. Fixed rule sets are difficult to cope with the diversity of the environment, especially in extreme weather or complex traffic conditions. The system may not be able to provide the best lighting strategy, thus affecting driving safety and energy efficiency. Because the rule-based ALS system relies on preset general rules, it cannot be personalized according to the driver's personal preferences or specific driving style. For example, some drivers may prefer stronger lighting when driving at night, while others may pay more attention to energy saving. Rule-based systems are difficult to meet such personalized needs, resulting in limited user experience. Summary of the invention

[0007] In order to solve the above problems, the present invention proposes a dynamic control method and system for headlights based on graph neural networks and state machines, which constructs the state machine, vehicles, roads and environmental conditions into a graph structure, and uses the graph neural network to learn the state representation of each node in the graph, providing a decision-making basis for the state transfer of the state machine, realizing the transition from traditional rule-driven to data-driven, ensuring the best match between the motor rotation angle of the headlights and the illumination angle of the headlights and the current environmental conditions, and further optimizing the representation of the node states to achieve refined and intelligent control of the headlight system.

[0008] According to some embodiments, the present disclosure adopts the following technical solutions:

[0009] The vehicle light dynamic control method based on graph neural network and state machine includes:

[0010] Obtain vehicle driving status, vehicle tilt angle changes, and external environment data, and convert all data into graph structure data;

[0011] Based on graph structure data, an adaptive headlight control algorithm is introduced, and the state machine in the adaptive headlight control algorithm is modeled as a directed graph, in which the vehicle state is used as a node and the external environment data is used as an edge, forming a graph structure of vehicle-environment interaction;

[0012] The graph structure of vehicle-environment interaction is input into the graph neural network model, the dependencies between nodes in the state machine graph and the state representation of each node are learned, the influence of the dependencies between nodes on the rotation angle of the vehicle motor and the adjustment of the headlights are obtained, the three modes C, V, and E of the adaptive headlight control algorithm are dynamically adjusted, and the classification results of the three modes are output. According to the classification results, the current headlight illumination angle and motor rotation angle are controlled to ensure the best matching state with the current environment.

[0013] According to some embodiments, the present disclosure adopts the following technical solutions:

[0014] The dynamic control system of car lights based on graph neural network and state machine includes:

[0015] The data acquisition module is used to obtain the vehicle driving status, vehicle tilt angle change and external environment data, and convert all the data into graph structure data;

[0016] A graph structure modeling module is used to introduce an adaptive headlight control algorithm based on graph structure data, and model the state machine in the adaptive headlight control algorithm as a directed graph, in which the vehicle state is used as a node and the external environment data is used as an edge, forming a graph structure of vehicle-environment interaction;

[0017] The adaptive control module is used to input the graph structure of vehicle-environment interaction into the graph neural network model, learn the dependencies between nodes in the state machine graph and the state representation of each node, obtain the influence of the dependencies between nodes on the rotation angle of the vehicle motor and the adjustment of the headlights, dynamically adjust the three modes C, V, and E of the adaptive headlight control algorithm, and output the classification results of the three modes. According to the classification results, the current headlight illumination angle and motor rotation angle are controlled to ensure the best matching state with the current environment.

[0018] According to some embodiments, the present disclosure adopts the following technical solutions:

[0019] A computer program product includes a computer program, and when the computer program is executed by a processor, the method for dynamic control of vehicle lights based on a graph neural network and a state machine is implemented.

[0020] According to some embodiments, the present disclosure adopts the following technical solutions:

[0021] A non-transitory computer-readable storage medium, wherein the non-transitory computer-readable storage medium is used to store computer instructions. When the computer instructions are executed by a processor, the vehicle light dynamic control method based on a graph neural network and a state machine is implemented.

[0022] According to some embodiments, the present disclosure adopts the following technical solutions:

[0023] An electronic device includes: a processor, a memory and a computer program; wherein the processor is connected to the memory, the computer program is stored in the memory, and when the electronic device is running, the processor executes the computer program stored in the memory so that the electronic device executes the method for dynamic control of vehicle lights based on a graph neural network and a state machine.

[0024] Compared with the prior art, the present invention has the following beneficial effects:

[0025] The disclosed method for dynamic control of vehicle lights based on a graph neural network and a state machine constructs graph structure data, constructs a graph data structure for training a dynamic state perception graph neural network (DSGNN), uses a graph neural network to process complex graph connection relationships, learns node representations through information propagation, and can dynamically adjust the motor rotation angle. The graph neural network uses a message passing mechanism to learn the dependencies between nodes and intelligently adjust the motor rotation angle and vehicle light illumination mode. The results of the graph neural network training are used to dynamically adjust the three modes of ALS, C, V, and E, to ensure the best match between the illumination angle and the current environmental conditions. The classification results output by the graph neural network are used to determine which of the three modes in the CVE of the ALS belongs to, and output the angle that the motor needs to be adjusted.

[0026] The present invention discloses a method for dynamic control of vehicle lights based on graph neural network and state machine. After the introduction of dynamic state perception graph neural network (DSGNN), the vehicle light system can analyze environmental changes in real time, such as road type, weather conditions, and light level, and intelligently adjust the lighting mode according to the data collected by the sensors during the driving process of the vehicle, which is significantly better than the traditional rule-driven ALS algorithm; the dynamic graph data processing capability of the graph neural network ensures that the system can quickly respond to vehicle state changes, such as sudden acceleration, deceleration, and ramp driving, and automatically adjust the angle of the headlights to improve driving safety and lighting efficiency. At the same time, in response to new scenarios, research and development can update the graph database in real time according to new scenarios to optimize the model obtained by graph neural network training in real time. Through continuous learning and optimization, the vehicle light system can continuously adjust the motor rotation angle and the headlight irradiation angle according to the accumulated real-time data to achieve the best match with environmental conditions.

[0027] The present invention discloses a method for dynamic control of vehicle lights based on a graph neural network and a state machine. Unlike traditional ALS systems that require customized rules based on different vehicle models and shapes, the present invention converts the state machine into a graph structure, and the graph neural network can learn the complex relationship between nodes and edges, providing data support for intelligent mode switching. Through comparison, it is found that the present invention can reduce power consumption, and by intelligently adjusting the motor rotation angle, the system can effectively reduce unnecessary lighting adjustments, reduce energy consumption, and improve overall energy efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] The accompanying drawings constituting a part of the present disclosure are used to provide a further understanding of the present disclosure. The illustrative embodiments of the present disclosure and their descriptions are used to explain the present disclosure and do not constitute an improper limitation on the present disclosure.

[0029] Figure 1 Schematic diagram of the traditional ALS algorithm implementation process. DETAILED DESCRIPTION

[0030] The present disclosure is further described below in conjunction with the accompanying drawings and embodiments.

[0031] It should be noted that the following detailed descriptions are all illustrative and are intended to provide further explanation of the present disclosure. Unless otherwise specified, all technical and scientific terms used herein have the same meanings as those commonly understood by those skilled in the art to which the present disclosure belongs.

[0032] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit the exemplary embodiments according to the present disclosure. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that when the terms "comprising" and / or "including" are used in this specification, it indicates the presence of features, steps, operations, devices, components and / or combinations thereof.

[0033] Example 1

[0034] In one embodiment of the present disclosure, a method for dynamic control of vehicle lights based on a graph neural network and a state machine is provided, the steps comprising:

[0035] Step 1: Obtain the vehicle driving status, vehicle tilt angle change and external environment data, and convert all data into graph structure data;

[0036] Step 2: Based on graph structure data, the adaptive headlight control algorithm is introduced, and the state machine in the adaptive headlight control algorithm is modeled as a directed graph, in which the vehicle state is used as a node and the external environment data is used as an edge, forming a graph structure of vehicle-environment interaction;

[0037] Step 3: Input the graph structure of vehicle-environment interaction into the graph neural network model, learn the dependencies between nodes in the state machine graph and the state representation of each node, obtain the impact of the dependencies between nodes on the rotation angle of the vehicle motor and the adjustment of the headlights, dynamically adjust the C, V, and E modes of the adaptive headlight control algorithm, and output the classification results of the three modes. According to the classification results, the current headlight illumination angle and motor rotation angle are controlled to ensure the best matching state with the current environment.

[0038] As an embodiment, in a method for dynamic control of vehicle lights based on a graph neural network and a state machine, the present invention utilizes a graph neural network structure, and the graph neural network is able to process data with complex graph connection relationships. The basic idea of ​​the graph neural network is to propagate information through the relationship between nodes, thereby learning the node representations in the graph. The representations of each node are learned through a neural network, and these representations contain the structural information and feature information of the node. Similarly, the representation of each edge is learned through a neural network, and these representations reflect the relationship between the two nodes connected by the edge. The representation of the entire graph is obtained through pooling operations, which can be used for graph classification tasks. Different car companies have different state machine representations for adaptive car lights. The present invention uses a large number of state machine graphs accumulated by car companies to construct a directed graph as input, which is input into the graph neural network for learning.

[0039] The traditional ALS algorithm (vehicle control algorithm) is generally implemented by receiving CAN signals, establishing a corresponding state machine system, and establishing rules to output the rotation angle of the headlight motor.

[0040] The static pitch adjustment function of traditional vehicle headlights means that when the vehicle is stationary (vehicle speed = 0km / h), the body load and load position change, resulting in changes in the body inclination angle, causing the angle and position of the headlights to change. The headlight algorithm system should adjust the headlights up and down to the correct angle. The static dimming function should be targeted at the following vehicle conditions:

[0041] 1. Drivers and passengers getting on and off the vehicle

[0042] 2. Movement of occupants in the vehicle

[0043] 3. Changes in luggage load

[0044] 4. Oil loss

[0045] When the vehicle is stationary, the headlight algorithm monitors the changes in the body inclination angle at any time, maintains the number of samples in the constant and actively filters the noise. Within the time T after the body inclination angle changes, the ECU will drive the dimming motor to complete the light adjustment action. In order to avoid too frequent adjustment of the headlights, which will affect the driver's subjective experience. The slight change in the body inclination angle is too small and will not drive the headlight algorithm to adjust the headlight angle.

[0046] During static pitch adjustment, if the light signal is turned off, the static pitch adjustment will be discontinued and the headlights will be adjusted back to the initial position. If the light signal is turned on again and the static pitch adjustment start conditions are met, the headlights will be adjusted to the correct static pitch adjustment position.

[0047] Considering various working conditions of the vehicle during driving (sudden acceleration and deceleration, slopes, bumpy roads), the dynamic pitch adjustment function of the vehicle headlights needs to be divided into the following mode functions:

[0048] 1. Rapid acceleration and deceleration dynamic pitch adjustment function

[0049] 2. Dynamic pitch adjustment function on ramp

[0050] 3. Dynamic pitch adjustment function for bumpy road recognition

[0051] During rapid acceleration, the vehicle's upward pitch will cause the light height to increase, so the headlight algorithm system should adjust the light to lower the angle of illumination to prevent dazzling oncoming drivers. During rapid deceleration, the vehicle's downward pressure will cause the light height to decrease, which will cause the lighting distance to decrease rapidly. The headlight algorithm system should adjust the light to raise the angle of illumination to maintain a reasonable lighting distance on the road and ensure driving safety. During acceleration and deceleration, the vehicle's pitch angle changes very quickly, and the response time of the headlight algorithm system's dimming system should be less than a certain time.

[0052] When the speed of the vehicle increases by more than a certain speed, it is determined that the vehicle has entered the rapid acceleration mode. Once the rapid acceleration mode is entered, the headlights are adjusted to lower the illumination angle in a very short time: when the speed of the vehicle increases by less than a certain speed, it is determined that the vehicle has left the acceleration mode. Once the rapid acceleration mode is exited, the headlights are adjusted to return to the normal illumination angle in a very short time. When the speed of the vehicle decreases by more than a certain speed, it is determined that the vehicle has entered the rapid deceleration mode. Once the rapid deceleration mode is entered, the headlights are adjusted to raise the illumination angle in a very short time; when the speed of the vehicle decreases by less than a certain speed, it is determined that the vehicle has left the rapid deceleration mode. Once the vehicle has left the deceleration mode, the headlights are adjusted to return to the normal illumination angle in a very short time.

[0053] During the rapid acceleration / deceleration pitch adjustment function, if the light signal is turned off, the rapid acceleration / deceleration pitch adjustment function will be immediately disengaged and the headlights will be adjusted back to the initial position: if the light signal is turned on again and the rapid acceleration / deceleration pitch adjustment function start conditions are met, the headlights will be immediately adjusted to the correct position for rapid acceleration / deceleration pitch adjustment. Optimization and adjustment can be made based on actual vehicle verification conditions.

[0054] When the vehicle travels on a slope for more than T, the center of gravity of the vehicle load shifts due to the slope, causing the body inclination to change. At this time, the headlight algorithm system should sense the vehicle inclination through the height sensor and adjust the headlight angle to compensate for the vehicle pitch caused by the slope, providing a suitable lighting distance on the slope.

[0055] During the ramp dimming pitch adjustment function, if the light signal is turned off, the ramp dimming pitch adjustment function will be immediately disengaged and the headlights will be adjusted back to the initial position: if the light signal is turned on again and the start conditions of the ramp dimming pitch adjustment function are met, the headlights will be immediately adjusted to the correct ramp dimming pitch adjustment position.

[0056] Therefore, in view of the problems existing in the above prior art, the present disclosure provides a vehicle light dynamic control method based on a graph neural network and a state machine. The specific implementation process is as follows:

[0057] Step 1: Obtain the vehicle driving status, vehicle tilt angle change and external environment data, and convert all data into graph structure data;

[0058] Specifically, sensors are used to collect vehicle driving status, vehicle inclination changes and external environment data in real time, and all data are converted into graph structure data; the vehicle driving status includes vehicle driving speed, vehicle position and vehicle driving direction, and the external environment data includes road type, weather data and light level. The attributes of the edges are the switching conditions and migration conditions between various vehicle states.

[0059] Step 2: Based on the graph structure data, the adaptive headlight control algorithm is introduced, and the state machine in the adaptive headlight control algorithm is modeled as a directed graph, in which the vehicle state is used as a node and the external environment data is used as an edge, forming a graph structure of vehicle-environment interaction;

[0060] An adaptive headlight control algorithm is introduced, and the state machine of the adaptive headlight control is represented as a directed graph, in which the vehicle state and environmental conditions are modeled as attributes of nodes and edges, forming a graph expression of vehicle-environment interaction.

[0061] Specifically, we first construct graph structure data, based on the large amount of data accumulated by Chery on the ALS algorithm of headlights in actual test scenarios and various state machine structure data accumulated previously, to construct the graph data structure for training graph neural networks. For example, node features include but are not limited to vehicle status, location, speed, direction, etc.; edge attributes include road type, weather conditions, light level, etc.

[0062] Step 3: Input the graph structure of vehicle-environment interaction into the graph neural network model, learn the dependencies between nodes in the state machine graph and the state representation of each node, obtain the impact of the dependencies between nodes on the rotation angle of the vehicle motor and the adjustment of the headlights, dynamically adjust the C, V, and E modes of the adaptive headlight control algorithm, and output the classification results of the three modes. According to the classification results, the current headlight illumination angle and motor rotation angle are controlled to ensure the best matching state with the current environment.

[0063] The attributes of the edges are the switching conditions and migration conditions between various states. Graph neural networks are used to process complex graph connection relationships, and node representations are learned through information propagation.

[0064] This paper proposes a dynamic state perception graph neural network (DSGNN) for real-time adaptive headlight control. The network realizes data-driven C / V / E mode decision-making through a dynamic vehicle environment interaction graph (node ​​= vehicle state, edge = environmental conditions and transfer rules), which includes four major modules:

[0065] 1. Dynamic graph builder: node feature h iv =[speed,Δθ,GPS,a x ,a y ], edge feature h ije = [road type, light intensity, weather], Data update graph structure G_t;

[0066] 2. Multi-head edge-aware convolutional layer: message generation m ij =MLP(h iv ,h ije ), attention weight Node Update

[0067] 3. Hierarchical graph pooling: Generate graph-level embedding h_G = connection (mean pooling, maximum pooling) through node L2 norm TOP-K clustering;

[0068] 4. Adaptive Classifier: Output Pattern Probability The control instruction θ_adjustment is triggered according to the probability threshold (e.g. -5°±α when p_C>0.7). Experiments show that training based on the state diagram achieves 93.7% classification accuracy and 18ms latency, reducing 41% of invalid motor adjustments compared to traditional ALS. The core innovations include: edge feature-driven state transition modeling, real-time node clustering and pooling, and CAN feedback online optimization parameters (W_a, W_g).

[0069] Among them, Δθ is the change in inclination angle, a_x / y = acceleration, α / β are environmental compensation coefficients, SUM is the summation, and σ is the activation function.

[0070] The dynamic state-aware graph neural network (DSGNN) can dynamically adjust the motor rotation angle according to graph systems such as (sudden acceleration and deceleration, ramps, bumpy roads, road types, weather conditions, light levels, C, V, E under different states). The graph neural network uses a message passing mechanism to learn the dependencies between nodes and intelligently adjust the motor rotation angle and the lighting mode of the headlights. The results of the graph neural network training are used to dynamically adjust the C, V, and E modes of the ALS to ensure the best match between the lighting angle and the current environmental conditions. The classification results output by the graph neural network are used to determine which of the three modes in the CVE of the ALS belongs to, and output the angle that the motor needs to be adjusted.

[0071] Specifically, the dynamic state-aware graph neural network (DSGNN) processes complex graph structure data, learns the state representation of each node through message passing between nodes, and then optimizes the motor control and headlight adjustment in the ALS algorithm. For different vehicle states and environmental conditions, the graph neural network can dynamically adjust the control strategy through the input state machine graph to ensure the best lighting mode. The final output of the graph neural network is the classified category information. By reducing the amount of calculation through the output category information, the motor angle can be adjusted more accurately.

[0072] The present disclosure can automatically learn the dependencies between nodes and how these relationships affect the adjustment of the headlights through a message passing mechanism during the model training process. This dynamic graph data processing capability enables the graph neural network to update the node status in real time, and uses the state machine in the headlight ALS algorithm as the input of the graph neural network. The graph neural network can determine the state transition and realize intelligent switching from one state to another. This closed-loop mechanism enables the headlight system to continuously learn and optimize during operation. The graph neural network can process complex graph structure data and learn the state representation of each node in the graph. These states can be high-dimensional embeddings of node features, providing decision-making basis for the state transition of the state machine, and realizing the transition from traditional rule-driven to data-driven. Ensure that the motor rotation angle of the headlight and the illumination angle of the headlight are optimally matched to the current environmental conditions. Further optimize the representation of the node state to achieve refined and intelligent control of the headlight system.

[0073] Through real vehicle testing, this paper verifies the response speed and adjustment accuracy of the system under conditions such as rapid acceleration, rapid deceleration, and ramp driving. In addition, by collecting and analyzing a large amount of driving data, the parameters of the graph neural network model are continuously optimized to improve the adaptability and stability of the system under various environmental conditions. The experimental results show that in scenes such as rapid acceleration and deceleration, and ramp driving, the ALS system fused with the graph neural network can significantly reduce glare, maintain a reasonable lighting distance, and improve driving safety. At the same time, it is found that it can significantly improve energy efficiency and reduce energy consumption.

[0074] Example 2

[0075] In one embodiment of the present disclosure, a vehicle light dynamic control system based on a graph neural network and a state machine is provided, including:

[0076] The data acquisition module is used to obtain the vehicle driving status, vehicle tilt angle change and external environment data, and convert all the data into graph structure data;

[0077] A graph structure modeling module is used to introduce an adaptive headlight control algorithm based on graph structure data, and model the state machine in the adaptive headlight control algorithm as a directed graph, in which the vehicle state is used as a node and the external environment data is used as an edge, forming a graph structure of vehicle-environment interaction;

[0078] The adaptive control module is used to input the graph structure of vehicle-environment interaction into the graph neural network model, learn the dependencies between nodes in the state machine graph and the state representation of each node, obtain the influence of the dependencies between nodes on the rotation angle of the vehicle motor and the adjustment of the headlights, dynamically adjust the three modes C, V, and E of the adaptive headlight control algorithm, and output the classification results of the three modes. According to the classification results, the current headlight illumination angle and motor rotation angle are controlled to ensure the best matching state with the current environment.

[0079] Example 3

[0080] In one embodiment of the present disclosure, a computer program product is provided, including a computer program, and when the computer program is executed by a processor, the method for dynamic control of vehicle lights based on a graph neural network and a state machine is implemented.

[0081] Example 4

[0082] In one embodiment of the present disclosure, a non-transitory computer-readable storage medium is provided, wherein the non-transitory computer-readable storage medium is used to store computer instructions. When the computer instructions are executed by a processor, the vehicle light dynamic control method based on a graph neural network and a state machine is implemented.

[0083] Example 5

[0084] In one embodiment of the present disclosure, an electronic device is provided, including: a processor, a memory, and a computer program; wherein the processor is connected to the memory, and the computer program is stored in the memory. When the electronic device is running, the processor executes the computer program stored in the memory, so that the electronic device executes the method for dynamic control of vehicle lights based on graph neural network and state machine.

[0085] The present disclosure is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present disclosure. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0086] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0087] Although the above describes the specific implementation methods of the present disclosure in conjunction with the accompanying drawings, it is not intended to limit the scope of protection of the present disclosure. Technical personnel in the relevant field should understand that on the basis of the technical solution of the present disclosure, various modifications or variations that can be made by those skilled in the art without creative work are still within the scope of protection of the present disclosure.

Claims

1. A vehicle light dynamic control method based on graph neural network and state machine, characterized in that: include: Obtain vehicle driving status, vehicle tilt angle changes, and external environment data, and convert all data into graph structure data; Based on graph structure data, an adaptive headlight control algorithm is introduced, and the state machine in the adaptive headlight control algorithm is modeled as a directed graph, in which the vehicle state is used as a node and the external environment data is used as an edge, forming a graph structure of vehicle-environment interaction; The graph structure of vehicle-environment interaction is input into the graph neural network model, the dependencies between nodes in the state machine graph and the state representation of each node are learned, the influence of the dependencies between nodes on the rotation angle of the vehicle motor and the adjustment of the headlights are obtained, the three modes C, V, and E of the adaptive headlight control algorithm are dynamically adjusted, and the classification results of the three modes are output. According to the classification results, the current headlight illumination angle and motor rotation angle are controlled to ensure the best matching state with the current environment.

2. The vehicle light dynamic control method based on graph neural network and state machine as claimed in claim 1, characterized in that: The vehicle driving state includes the vehicle driving speed, vehicle position and vehicle driving direction, the external environment data includes road type, weather data and light level, and the attributes of the edge are the switching conditions and migration conditions between various vehicle states.

3. The vehicle light dynamic control method based on graph neural network and state machine as claimed in claim 1, characterized in that: In the graph neural network model, through message passing and node feature updating, the dependencies between nodes in the state machine graph structure and the impact of the dependencies on the adjustment of the headlights are automatically learned, the structural features of each node and each edge are extracted, the node status is updated through the message passing mechanism, the graph data is dynamically learned, and the optimized node status representation is output.

4. The vehicle light dynamic control method based on graph neural network and state machine as claimed in claim 3, characterized in that: The output optimized node state represents the decision of state machine state transfer, and the node state represents the state transfer, realizing intelligent switching from one state to another, dynamically adjusting the C, V, and E modes of the adaptive headlight control algorithm to ensure the best match between the motor rotation angle of the headlight and the illumination angle of the headlight and the current environmental conditions.

5. The vehicle light dynamic control method based on graph neural network and state machine as claimed in claim 1, characterized in that: When the vehicle is stationary, the changes in the vehicle status are obtained at any time. Within the time T after the vehicle's inclination angle changes, the ECU drives the dimming motor to complete the lighting adjustment action. During the static pitch adjustment, when the light signal is turned off, the static pitch adjustment is immediately disengaged and the headlights are adjusted back to the initial position: when the light signal is turned on again and the static pitch adjustment starting conditions are met, the headlights can be adjusted to the correct static pitch adjustment position.

6. The vehicle light dynamic control method based on graph neural network and state machine as claimed in claim 1, characterized in that: When the vehicle speed increases faster than a certain speed, it is determined that the vehicle enters the rapid acceleration mode, and the headlights are adjusted to lower the illumination angle within a very short time; when the vehicle speed increases slower than a certain speed, it is determined that the vehicle leaves the acceleration mode, and the headlights are adjusted to return to the normal illumination angle within a very short time; when the vehicle speed decreases faster than a certain speed, it is determined that the vehicle enters the rapid deceleration mode, and the headlights are adjusted to raise the illumination angle within a very short time; when the vehicle speed decreases slower than a certain speed, it is determined that the vehicle leaves the rapid deceleration mode, and the headlights are adjusted to return to the normal illumination angle within a very short time.

7. A dynamic control system for vehicle lights based on graph neural network and state machine, characterized in that: include: The data acquisition module is used to obtain the vehicle driving status, vehicle tilt angle change and external environment data, and convert all the data into graph structure data; A graph structure modeling module is used to introduce an adaptive headlight control algorithm based on graph structure data, and model the state machine in the adaptive headlight control algorithm as a directed graph, in which the vehicle state is used as a node and the external environment data is used as an edge, forming a graph structure of vehicle-environment interaction; The adaptive control module is used to input the graph structure of vehicle-environment interaction into the graph neural network model, learn the dependencies between nodes in the state machine graph and the state representation of each node, obtain the influence of the dependencies between nodes on the rotation angle of the vehicle motor and the adjustment of the headlights, dynamically adjust the three modes C, V, and E of the adaptive headlight control algorithm, and output the classification results of the three modes. According to the classification results, the current headlight illumination angle and motor rotation angle are controlled to ensure the best matching state with the current environment.

8. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the vehicle light dynamic control method based on a graph neural network and a state machine as described in any one of claims 1 to 6 is implemented.

9. A non-transitory computer-readable storage medium, characterized in that: The non-transitory computer-readable storage medium is used to store computer instructions. When the computer instructions are executed by the processor, the vehicle light dynamic control method based on the graph neural network and the state machine as described in any one of claims 1 to 6 is implemented.

10. An electronic device, characterized in that: include: A processor, a memory and a computer program; wherein the processor is connected to the memory, the computer program is stored in the memory, and when the electronic device is running, the processor executes the computer program stored in the memory so that the electronic device executes the method for dynamic control of vehicle lights based on a graph neural network and a state machine as described in any one of claims 1-6.

Images