AI-BASED VEHICLE HEADLIGHT ADJUSTMENT SYSTEM THAT RESPONDS TO DYNAMIC ROAD CONDITIONS.

Abstract

The invention relates to an artificial intelligence-based system and method that detects sudden and short-duration impacts originating from the road surface using data obtained from vehicle dynamics, and dynamically and adaptively adjusts the vehicle's headlight level based on these detections. Within the scope of the invention, data relating to vehicle dynamics such as acceleration, speed, braking, steering movements, etc., obtained via the in-vehicle communication network are collected in real time and processed as time-series data structures. This data is subjected to labeling processes and used to train artificial intelligence-based models, and real-time inferences are made through the trained models to detect road surface-related events. The system is specifically designed to distinguish characteristic vehicle dynamic changes occurring during bump crossings from other vehicle behaviors such as braking, acceleration, or load changes. In this way, road events that cause short-term and sudden pitch changes can be determined with high accuracy.Based on the classification results obtained, the vehicle's headlight level is adjusted dynamically and adaptively in real-time, thus reducing temporary glare effects for oncoming drivers and increasing driving safety. Furthermore, the invention can operate solely on existing data from the vehicle itself, without the need for cameras, lidar, or similar external sensors, thus offering a cost-effective and feasible solution. Thanks to its data-driven structure, the system can be adapted to different vehicle types and improved over time.

Description

1 TARIFF ARTIFICIAL INTELLIGENCE THAT RESPONDES TO DYNAMIC ROAD CONDITIONS BASE-BASED VEHICLE HEADLIGHT ADJUSTMENT SYSTEM Technical Area This invention is used in the automotive industry for vehicle lighting systems and driving safety. It is focused on the area of ​​vehicle headlights, specifically their level, direction, and light distribution. It is concerned with systems and methods for dynamic control. 10 The invention primarily targets data obtained through in-vehicle communication networks, data relating to vehicle dynamics, driving conditions and environmental parameters processing of this data and time series information obtained from it using artificial intelligence. By analyzing the headlight settings with algorithms, adaptive adjustments can be made. It includes the implementation. State of the Art Headlight leveling systems used in vehicles today generally... with manual control mechanisms or limited automatic control systems 20 Automatic headlight leveling systems are mostly implemented in vehicles. load condition, suspension position, or tilt of the vehicle body along the longitudinal axis It operates based on parameters such as pitch variation, and previously It uses defined threshold values ​​and rule-based algorithms. These types of systems adapt to specific and relatively slowly changing driving conditions. while being able to provide instantaneous and high-frequency changes in road conditions (e.g., speed bumps, road irregularities, or sudden surface changes) sufficient unable to accurately perceive and effectively assess the effect of these changes on headlight adjustment. They are unable to manage it properly. This situation is particularly problematic for short but sudden moments when the vehicle is about to crash. 2 temporary glare for oncoming drivers during their movements This can lead to problems and negatively affect driving safety. Upon examining various patent documents in the literature, it is observed that the existing solutions... It appears to be based primarily on the following approaches: In some solutions, pitch changes derived from vehicle dynamics are used for braking, for example. Headlight angle is adjusted in relation to specific driving situations. This type of The document US2004125608A1 can be given as an example of such approaches. However, this Such approaches do not take into account sudden and irregular effects caused by the road surface, and It does not treat phenomena like bumps as a separate issue. Other solutions are longer-lasting and relatively quick, depending on the vehicle load. It detects consistent pitch changes and adjusts the headlight setting accordingly. This The document US2025332987A1 can be given as an example of this approach. These systems It generally operates based on fixed threshold values, and is dynamic and short-term. It lacks the ability to distinguish changes. In some solutions, the effect of road defects is eliminated, and only the load is 15. The aim is to detect slope changes caused by various factors. This approach One of the solutions representing this is document US2025332986A1. However, this Because the approach does not involve direct modeling of road-related impacts, accurate characterization of bumps and similar sudden events It cannot provide. 20 Additionally, some systems include environmental sensing sensors such as cameras and / or lidar. Solutions based on this approach are proposed. Examples of approaches within this scope include: The document US2025001928A1 can be cited. However, such systems... They require high equipment costs and are susceptible to environmental factors such as low light, rain, and fog. Because they are affected by conditions, they are suitable for every use case and 25 It does not offer a reliable solution. 3 The common disadvantage of the existing technical solutions summarized above is the road surface. sudden and short-term vehicle dynamics caused by (especially speed bump crossings) (temporary pitch changes occurring during this process) reliable detection and this Insufficient ability to isolate effects from other vehicle dynamics. Their remaining position. Therefore, these systems require the headlight angle to be in line with actual driving conditions. 5 It is unable to ensure that the necessary adjustments are made in a timely manner and in accordance with the conditions, which This causes glare problems, especially for oncoming drivers. In addition, in the literature and in current practices, in-vehicle communication networks The data obtained from this process is artificially engineered by considering the time series characteristics. Analysis using intelligence methods and dynamic effects originating from the road surface in these 10 By identifying it in this way and integrating it directly into the headlight control system. There is no comprehensive solution. Brief Description of the Invention The present invention meets the aforementioned requirements and overcomes all the disadvantages of 15 a response to dynamic road conditions that eliminates and introduces some additional advantages This relates to an AI-based vehicle headlight adjustment system. The aim of this invention is to obtain vehicle dynamics via an in-vehicle communication network. By continuously monitoring and processing data, sudden 20% of road surface-related accidents can be identified. and the detection of short-term dynamic effects using artificial intelligence-based methods to provide. In this context, the system processes CAN signals obtained from the vehicle in real time. It collects data and organizes it into suitable data sets. The collected data is then labeled. 25 training time series-based artificial intelligence models by putting them through these processes by providing and training the model and making real-time inferences. The aim is to distinguish road phenomena such as bumps from other vehicle dynamics. 4 Another purpose of the invention is to address temporary pitch that occurs, especially when passing over bumps. changes in other vehicles such as braking, acceleration and load changes isolating it from its dynamics and, depending on the result of this isolation, the vehicle's headlights It is the real-time, dynamic, and adaptive control of its level. Thus, similar dynamics such as braking, road imperfections, and acceleration are considered. By distinguishing the bump effect from behaviors, oncoming drivers Potential temporary headlight glare is prevented. The invention also allows for the use of devices without the need for cameras, lidar, or similar external sensors. low-cost, high-performance systems that operate solely on available data from the vehicle. to offer an accurate and continuously improveable headlight adjustment system This is the aim. In this respect, the invention involves data collection, model training, and real-world application. by offering a timely decision-making mechanism together with existing headlight level control. a smarter, more adaptable and higher accuracy solution compared to existing systems It provides. 15 All the advantages mentioned above and explained in detail below. The present invention aims to achieve this by obtaining data via an in-vehicle communication network. a system that allows the adjustment of vehicle headlights using vehicle dynamics data The system collects multivariate time series data obtained from the instrument. 20 a data collection unit structured to process that data and guide it an artificial intelligence system designed to detect surface-generated dynamic events intelligence-based analysis unit and vehicle preview based on detected road events. a control that dynamically adjusts the level, direction and / or light distribution of the headlights It includes a unit and an AI-based analysis unit that detects 25 bumps during the bump crossing. the resulting changes in vehicle dynamics, such as braking, acceleration and / or load changes characterized by being trained in a way that distinguishes it from other vehicle dynamics. It is a system that is implemented. In a possible configuration of the invention, vehicle dynamics data could be incorporated into vehicle speed, engine speed, wheel speeds, steering column torque, brake and accelerator pedal signals and includes at least one of the longitudinal, lateral, and vertical acceleration components. In a possible configuration of the invention, the artificial intelligence-based analysis unit would consist of 5 It is structured to analyze multivariate time series data. In a possible configuration of the invention, the artificial intelligence model would be supervised. learning, deep learning, recurrent neural networks (RNN), long-term and short-term memory (LSTM) or convolutional neural networks (CNN) at least one of these methods 10 It includes. In a feasible configuration of the invention, the data collection unit would be in-vehicle. Controller Area Network (CAN-BUS) and / or similar data communication network It uses transmission infrastructures. 15 In a possible configuration of the invention, the artificial intelligence-based analysis unit, short-term and characteristic signal changes that occur when passing over a speed bump, distinguishing signal changes caused by poor road conditions and / or sudden braking. It is structured in such a way as to... 20 In a possible configuration of the invention, the control unit is the detected path. Depending on the situation, you can adjust the headlight angle, light intensity, and / or light distribution accordingly. It adjusts according to the time. Again, all of the above-mentioned points and the detailed explanation below will be understood. The present invention aims to realize the advantages of adjusting vehicle headlights. It is a method aimed at vehicle dynamics through the in-vehicle communication network. data collection, the organization of this data into time series data structures processing, analyzing the processed data with an artificial intelligence-based model, way 30 Detection of surface-related incidents and vehicle analysis based on the detected incidents. 6 It includes steps for dynamically adjusting the headlights and bump crossing. Characterized by the ability to distinguish its signal patterns from other vehicle dynamics. It is a method that is used. The structural and characteristic features and all the advantages of the invention are given below in 5. a detailed explanation written with figures and references to these figures This will make it clearer, and therefore the evaluation will also be based on this. This should be done taking into account the figures and detailed explanations. Explaining the Figures 10 The best way to structure the current invention and its advantages with additional elements. In order for it to be understood, the figures explained below are included. It needs to be evaluated. Figure 1. State Diagram Figure 2. Creating the Data Set Figure 3. Creating an Artificial Intelligence Model Figure 4. Road Bump Scenario Data. Figure 4.1. Passing a Speed ​​Bump with Vehicle Braking 20 Figure 4.2. Longitudinal Acceleration During Flat Road and Bump Crossing. Comparison Figure 4.3. Longitudinal Acceleration Occurring on Straight and Uneven Roads. Comparison The parts in the figures are individually numbered, and these numbers correspond to: It is given below. The drawings do not necessarily need to be scaled, and the existing invention... Details that are not necessary for understanding may have been omitted. From this point onwards... 7 other, at least substantially identical or at least substantially identical Elements with functions are represented by the same number. Reference numbers 1.1. Vehicle Approaching a Speed ​​Bump 1.2. Headlight Beam 1.3. Bump 1.4. Headlight Angle of Vehicle on Bump 1.5. Light Beam That Causes Glare 10 1.6. Oncoming Vehicle 2.1 Vehicle 2.2 Vehicle Headlight 2.3 ECU 15 2.4 CAN Bus Communication 2.5 CAN-BUS Data 2.6 Vehicle Communication Recording Device 3.1 Collection of CAN-BUS data 20 3.2 Data Collected from the Vehicle 3.3 Separating data into test and training data 3.4 Artificial Intelligence Algorithm Selection 3.5 Selection of Hyperparameters and Metrics 3.6 Comparison of Artificial Intelligence Models 25 3.7 Integration of the Suitable Model into the Vehicle ECU 4.1.1. Longitudinal Acceleration Signal During Bump Passage 4.1.2. Vehicle Brake Pedal Signal 4.1.3. Steering Column Torque Signal 30 8 4.2.1. Road Without Bumps 4.2.2. Speed ​​Bump Crossing 4.3.1. Longitudinal Acceleration on a Damaged Road Detailed Description of the Invention 5 This detailed explanation describes the invention as an artificial intelligence system that responds to dynamic road conditions. The preferred configurations for the intelligence-based vehicle headlight adjustment system are only... to better understand the subject and without any limiting influence It is explained in a way that will not create a problem. 10 This invention provides multivariate vehicle information obtained through in-vehicle communication systems. dynamics data processing and time series characteristics of this data By analyzing road surface-related issues using artificial intelligence-based algorithms Reliable 15 for sudden and short-term dynamic impacts, especially road events such as bumps. the determination of the level and direction of the vehicle's headlights based on these determinations and / or a system that allows for dynamic adjustment of light distribution. It relates to the method. The vehicle data used within the scope of the invention is Controller Area Network (CAN-BUS) 20 This is primarily obtained through in-vehicle communication infrastructures, The use of different data transmission protocols in alternative applications also It is possible. The system obtains data from sensors already present on the vehicle. It is configured to use the received signals, eliminating the need for additional hardware. It is designed to minimize this. 25 According to the situation diagram shown in Figure 1, in the vehicle approaching the bump (1.1), Under normal driving conditions, the headlight beam (1.2) illuminates the road surface at a certain angle and It illuminates with distribution. When the vehicle goes over a bump (1.3), the suspension Depending on the system and vehicle mass distribution, there is a sudden dynamic response in the vehicle body. 30 This occurs and results in the vehicle's nose pitching. 9 As a result of the movement, the headlight angle (1.4) temporarily changes upwards, and this the situation is that the beam of light causes glare for oncoming vehicles (1.5) leads to the formation of. The driver of the oncoming vehicle (1.6) causes this temporary However, it is negatively affected by sudden increases in light. Within the scope of the invention, road phenomena such as speed bumps can be prevented from occurring beforehand. detection with high accuracy just before or at the moment of occurrence, and this Proactive adjustment of headlight settings based on available information. is the goal. As shown in Figure 2, the system consists of the vehicle (2.1), vehicle headlight system (2.2), and electronics. control unit (ECU) (2.3), in-vehicle communication network (2.4) and data recording / processing The unit consists of (2.6) components. Data obtained via CAN-BUS Within the scope of (2.5); vehicle speed, engine speed, gas and brake pedal positions, steering Column torque, wheel speeds, and suspension movements, along with longitudinal and lateral 15 and numerous vehicle dynamics parameters such as vertical acceleration components are real. It is collected periodically. This collected data can be in raw signal form or subjected to specific pre-processing. noise reduction, normalization, and sampling frequency are performed through these steps. It can be subjected to processes such as compatibility testing and feature extraction. Especially over time. signals are transmitted at specific time intervals using windowing techniques. Patterns are extracted and these patterns are used as input for the artificial intelligence model. It is presented. Characteristic signals that occur when passing over a speed bump include; front and rear. Temporary differences between wheel speeds, longitudinal and vertical acceleration. Sudden changes in its components depend on the oscillation frequency of the vehicle body. These include fluctuations and micro-changes that are reflected in the steering system. This multidimensional signal structure is a complex 30 that cannot be expressed by a single parameter. This behavior is exhibited and therefore requires multivariate analysis. Within the process shown in Figure 3, the data obtained during the data collection phase (3.1) Vehicle data is processed through labeling procedures to detect bumps, rough roads, braking, They are classified to represent different driving situations, such as acceleration. This dataset is divided into training and test datasets (3.3) and appropriate artificial intelligence 5 Model training is carried out by selecting intelligence algorithms (3.4). Among the artificial intelligence methods that can be used are; supervised learning. algorithms, deep learning-based time series models, recurrent neural networks networks (RNN), long-short term memory networks (LSTM) or convolutional neural networks 10 Methods such as central networking (CNN) may be involved. However, the invention is specific to a particular area. Not limited to algorithms, but also patterns from multivariate time series data. The scope has been broadly defined to include all types of artificial intelligence models capable of learning. Hyperparameter optimization during model training process (3.5), performance 15 By determining the metrics and performing model comparisons (3.6), the most The appropriate model is selected and integrated into the vehicle ECU (3.7). The integrated model makes inferences from real-time data streams, providing instantaneous insights. classifying driving conditions and road phenomena such as bumps, for other vehicles. It distinguishes itself from its dynamics. 20 Figure 4 examines data from different driving scenarios. Figure In case of passing over a speed bump while braking as in 4.1, the brake signal (4.1.2) is activated. A significant relationship is observed between longitudinal acceleration (4.1.1) and also Characteristic changes occur in the steering column torque (4.1.3) signal. This is reflected in Figure 4, which analyzes data from different driving scenarios. This is shown in Figure 4.1, which illustrates the bump that occurs simultaneously with braking. The transition situation has been examined. The vertical line in the graphs indicates the vehicle's passage over a bump. It represents the moment it passes over. At this moment, the longitudinal acceleration signal (4.1.1) A clear relationship is observed between the brake signal (4.1.2) and 30 11 The characteristic of the steering column torque signal (4.1.3) is specific to bump crossing. Changes are occurring. In Figure 4.2, when comparing the flat road (4.2.1) and the bump crossing (4.2.2), the bump The oscillatory behaviors specific to this transition appear to be distinctly different. 5 Figure 4.3 shows the signal changes that occur under poor road conditions (4.3.1). It has been investigated and it has been found that these signals may show similarities to bump crossings, however It appears to contain certain pattern differences. The artificial intelligence model developed as part of the invention analyzes the signals belonging to these different scenarios. By learning their patterns, we can minimize false positives and reduce bumps. It makes its detection with a high degree of accuracy. Based on the classification result obtained, the vehicle's front headlight system is actual. It is checked periodically and the headlight angle, light intensity or light distribution is adjusted 15 It adjusts dynamically. This allows for proper handling when passing over speed bumps. Potential temporary glare effects are prevented or minimized in advance. is being done. In conclusion, this invention eliminates the need for additional sensors by using existing vehicle data. eliminating, AI-powered, learnable, adaptive, and highly accurate It offers a headlight control system that enhances driving safety and driving comfort. It provides an innovative solution.

Claims

12 REQUESTS 1. Vehicle dynamics data obtained via the in-vehicle communication network. using which dynamic control of vehicle headlights is enabled It is a system, and its feature is; 5 - collecting multivariate time series data obtained from the instrument data collection unit structured accordingly, - processing the data to identify dynamic events originating from the road surface. AI-based analysis unit that detects and - Depending on detected road incidents, the vehicle's headlight level will be adjusted to 10. Control that dynamically adjusts direction and / or light distribution. It includes the unit and the artificial intelligence-based analysis unit, as shown in Figure 1. Interaction of the vehicle approaching the shown bump (1.1) and the bump (1.3) characteristic vehicle dynamics changes that occur during by learning other vehicle techniques such as braking, acceleration and / or load changes. by being trained to distinguish it from its own dynamics It is a characterized system.

2. The system is based on claim 1 and its feature is; vehicle dynamics data of the vehicle (2.1), 20 obtained via ECU (2.3) and CAN-BUS communication infrastructure (2.4) vehicle speed, engine RPM, wheel speeds, steering column torque, brake and accelerator. pedal signals and longitudinal, lateral and vertical acceleration components, at least It involves one of them.

3. The system is defined as either Claim 1 or 2, and its characteristic is: artificial intelligence-based analysis. 25 multivariate in accordance with the data processing process of the unit (3.1–3.7) It is structured to analyze time series data.

4. A system that meets any of the previous criteria and whose characteristic is artificial intelligence. In accordance with the model selection process of the model, supervised learning, deep 30 13 learning, at least one of the following methods: RNN, LSTM or CNN based. It includes.

5. A system that meets any of the previous requirements, and whose characteristic is data collection. the unit's in-vehicle communication network is the Controller Area Network (CAN-5 This involves using data transmission infrastructures (BUS) and / or similar systems.

6. A system that meets any of the previous criteria and whose characteristic is artificial intelligence. Based on the road scenarios of the base analysis unit, bump crossing (4.2.2) short-term and characteristic signal changes that occur during this time, faulty path 10 (4.3.1) and signal changes occurring during braking scenarios It is structured in a way that allows for differentiation.

7. The system, according to any of the previous requirements, has the following characteristic: control. the unit's headlight angle, light intensity and / or light distribution are actually 15 It is a time-based adjustment.

8. This is a method for dynamic control of vehicle headlights, and its characteristic feature is; - Vehicle dynamics data via the in-vehicle communication network. collection, 20 - Processing the collected data in the form of time series data structures, - Analysis using an artificial intelligence model, - Separating speed bump transitions from other vehicle dynamics and - It includes steps for dynamically adjusting the headlight system. It is a characterized method. 25