Driving behavior specification method and system based on complex scene, medium and equipment

Abstract

The invention discloses a driving behavior specification method and system based on a complex scene, a medium and equipment, and relates to the technical field of vehicle safety management and control. The method comprises the following steps: acquiring driving data and environment data of various drivers in real time; identifying and analyzing the driving data by using the behavior model to obtain a driving behavior evaluation result; generating a corresponding intervention decision according to a driving behavior evaluation result, classifying non-standard driving behaviors into self-factor influence behaviors and environmental factor influence behaviors, performing correction analysis on the self-factor influence behaviors to generate an internal cause intervention decision, extracting environmental data when the environmental factor influence behaviors occur, and determining the internal cause intervention decision according to the extracted environmental data. Generating an external cause intervention decision by combining the environment data on the environment factor influence behavior; and performing behavioral standard correction and continuous test on the driver by using the intervention decision. According to the invention, various driving behaviors of the driver can be identified in real time in a complex driving scene, and non-standard driving behaviors are effectively corrected through targeted continuous intervention.

Description

Technical Field

[0001] The present invention relates to the field of vehicle safety management and control technology, and in particular to a method, system, medium and equipment for regulating driving behavior based on complex scenarios. Background Art

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

[0003] As road traffic becomes increasingly complex, the behavioral norms of drivers are crucial to ensuring driving safety. However, in actual driving, there are often bad driving habits such as fatigue driving, speeding, not wearing seat belts, and distracted driving, which significantly increase the risk of traffic accidents.

[0004] Traditional driving monitoring methods often rely on manual supervision or simple vehicle-mounted equipment alarms. These methods can remind drivers of the risks ahead to a certain extent. However, they can usually only provide passive warnings for specific single risk factors, lack in-depth understanding of the driver's overall behavior pattern and active intervention capabilities, and have limited recognition of complex driving scenarios (such as subjective states such as fatigue and distraction), making it difficult to achieve accurate, efficient, and personalized driving behavior management. Therefore, the existing technology lacks a method and system that can provide continuous feedback training and precise intervention for different irregular behaviors of drivers in complex driving scenarios. Summary of the invention

[0005] In view of the deficiencies in the prior art, the purpose of the present invention is to provide a method, system, medium and device for regulating driving behavior based on complex scenarios, which can identify the driver's various driving behaviors in real time in complex driving scenarios, and effectively correct non-standard driving behaviors through targeted and continuous intervention.

[0006] In order to achieve the above object, the present invention is implemented through the following technical solutions:

[0007] A first aspect of the present invention provides a method for regulating driving behavior based on complex scenarios, comprising the following steps:

[0008] Acquire driving data and environmental data of multiple drivers in real time, and pre-process the driving data and environmental data;

[0009] Use the behavior model to identify and analyze the preprocessed driving data to obtain the driving behavior evaluation results;

[0010] Generate corresponding intervention decisions based on the driving behavior assessment results. Among them, classify non-standard driving behaviors into behaviors affected by self-factors and behaviors affected by environmental factors. Perform correction analysis on behaviors affected by self-factors to generate internal intervention decisions. Extract environmental data when behaviors affected by environmental factors occur, and combine environmental data to generate external intervention decisions on behaviors affected by environmental factors.

[0011] Use decision intervention to modify driver behavior and conduct ongoing testing.

[0012] Furthermore, the driving data includes the driver's physiological state, line of sight direction and operating actions.

[0013] Furthermore, the specific steps of using the behavior model to identify and analyze the preprocessed driving data are as follows:

[0014] Use deep learning algorithms to build behavioral models;

[0015] The existing database data is combined with the driver's historical driving data to form a training set, and the behavior model is trained;

[0016] The trained behavior model is used to evaluate the real-time driving data to obtain the driving behavior evaluation results.

[0017] Furthermore, the driving behavior assessment result includes standard driving behavior and non-standard driving behavior, and the risk level is judged according to the severity of the non-standard driving behavior.

[0018] Furthermore, the environmental data includes weather condition data, road condition complexity data, traffic flow data and potential hazard source data.

[0019] Furthermore, the intervention decision is a staged intervention decision, the number of stages is determined according to the severity of the non-standard driving behavior, and the subsequent number of stages and decision content are adjusted in real time according to the driver's correction feedback.

[0020] Furthermore, continuous testing is used to test the driver's corrective behavior for the implementation effect of the intervention decision at each stage and generate driver correction feedback.

[0021] A second aspect of the present invention provides a driving behavior regulation system based on complex scenarios, comprising:

[0022] A data acquisition module is configured to acquire driving data and environmental data of various drivers in real time and pre-process the driving data and environmental data;

[0023] A behavior recognition and evaluation module is configured to use the behavior model to recognize and analyze the preprocessed driving data to obtain a driving behavior evaluation result;

[0024] A decision-making module is configured to generate corresponding intervention decisions according to the driving behavior evaluation results, wherein the non-standard driving behaviors are classified into behaviors affected by self-factors and behaviors affected by environmental factors, correction analysis is performed on behaviors affected by self-factors to generate internal intervention decisions, environmental data when behaviors affected by environmental factors occur are extracted, and external intervention decisions are generated for behaviors affected by environmental factors in combination with the environmental data;

[0025] The correction feedback module is configured to utilize intervention decisions to perform behavioral correction and continuity testing on the driver.

[0026] The third aspect of the present invention provides a medium having a program stored thereon, which, when executed by a processor, implements the steps in the method for regulating driving behavior based on complex scenarios as described in the first aspect of the present invention.

[0027] The fourth aspect of the present invention provides a device, including a memory, a processor, and a program stored in the memory and executable on the processor, wherein when the processor executes the program, the steps in the method for regulating driving behavior based on complex scenarios as described in the first aspect of the present invention are implemented.

[0028] One or more of the above technical solutions have the following beneficial effects:

[0029] The present invention designs a method, system, medium and equipment for regulating driving behavior based on complex scenarios, which can realize real-time identification of all-round non-standard driving behaviors through training with a large amount of data. Compared with the existing driving behavior monitoring methods, the present invention can also analyze the factors affecting driving behavior in complex driving scenarios, generate more targeted intervention decisions, and greatly improve the effectiveness of correcting driving behavior norms. In order to consolidate its correction effect, the present invention also innovates a continuous staged intervention method, which ensures that the driver's behavior is not maintained in a short period of time through adaptive stage testing and correction, allowing the driver to form muscle habits, improve the driving safety factor, and provide a stronger guarantee for safe driving.

[0030] Advantages of additional aspects of the present invention will be given in part in the following description, and in part will become obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] The accompanying drawings in the specification, which constitute a part of the present invention, are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute improper limitations on the present invention.

[0032] Figure 1 This is a flow chart of a driving behavior regulation method based on complex scenarios in Embodiment 1 of the present invention;

[0033] Figure 2 This is a structural diagram of the data collection process in the first embodiment of the present invention;

[0034] Figure 3 This is a structural diagram of the behavior recognition and evaluation process in the first embodiment of the present invention;

[0035] Figure 4 A structural diagram of a decision-making designation process in Embodiment 1 of the present invention;

[0036] Figure 5 This is a structural diagram of the interactive feedback and guidance process in the first embodiment of the present invention;

[0037] Figure 6 This is a structural diagram of a driving behavior regulation system based on complex scenarios in Embodiment 2 of the present invention. DETAILED DESCRIPTION

[0038] It should be noted that the following detailed descriptions are exemplary and are intended to provide further explanation of the present invention. 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 invention belongs.

[0039] 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 invention. 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 "include" and / or "include" are used in this specification, it indicates the presence of features, steps, operations, devices, components and / or their combinations;

[0040] Embodiment 1:

[0041] Embodiment 1 of the present invention provides a driving behavior regulation method based on complex scenarios, such as Figure 1 As shown, the following steps are included:

[0042] Step 1: Acquire driving data and environmental data of multiple drivers in real time, and pre-process the driving data and environmental data.

[0043] Driving data includes the driver's physiological state, sight direction and operating actions. Environmental data includes weather conditions, road complexity, traffic flow and potential danger sources.

[0044] In this embodiment, weather conditions and potential hazard source data can be obtained through vehicle body sensors, etc., and road condition complexity data and traffic flow data can be obtained through navigation software.

[0045] During data collection, Figure 2As shown in the figure, the driver is taken as the collection object, and multiple sensor devices installed in the car, including but not limited to high-definition cameras, biometric sensors, vehicle motion sensors and CAN bus interfaces, are used to collect multivariate data such as the driver's physiological state, line of sight direction and operating actions. After that, the data is preprocessed and then transmitted to the data fusion and synchronization unit for synchronization and storage. Among them, vehicle motion sensors such as accelerometers, gyroscopes, wheel speed sensors, etc. Driver's physiological state such as fatigue level and concentration. Operational actions such as steering, acceleration, braking force and other vehicle dynamic information, vehicle dynamic information such as speed, position, heading, driving path, etc. The data collection process ensures the real-time and accuracy of the information, and can provide a basic source of teaching data for subsequent behavior recognition and evaluation.

[0046] In a specific implementation, the data collection process includes the following steps:

[0047] 1. Hardware deployment: Deploy various sensor devices in the vehicle, including but not limited to high-definition cameras, biometric sensors, vehicle motion sensors, and connections to the vehicle CAN bus interface to obtain vehicle driving data. Among them, the high-definition camera is installed in front of the driver to capture facial expressions, line of sight, etc., biometric sensors such as heart rate sensors and eye tracking sensors are used to monitor the driver's physiological state, and vehicle motion sensors such as accelerometers, gyroscopes, wheel speed sensors, etc. are used to capture vehicle dynamic information.

[0048] 2. Data interface configuration: Ensure that the data interface of each sensor device and the vehicle-mounted AI assistant system is correctly connected and configured to ensure the stability and real-time performance of data transmission.

[0049] 3. Data preprocessing: Perform necessary preprocessing on the received raw sensor data, such as filtering and denoising, data format conversion, unit unification, etc., to facilitate subsequent module processing.

[0050] 4. Data synchronization and storage: The pre-processed data is transmitted to the processing unit of the AI ​​assistant in real time, and stored in the local cache or on-board storage device as needed. At the same time, it is synchronized to the cloud server through the on-board communication module, such as 4G / 5G, Wi-Fi, etc., to support remote management and data analysis.

[0051] Step 2: Use the behavior model to identify and analyze the preprocessed driving data to obtain the driving behavior evaluation results.

[0052] like Figure 3As shown, the data collected in step 1 is input, and the driver behavior, vehicle operation, and environmental factors are extracted. Then, the data is input into the constructed behavior model, the behavior is identified and classified, and finally, risk assessment and warning are performed based on the results. Specifically, it includes:

[0053] Step 2.1: Use deep learning algorithms combined with computer vision, pattern recognition and other technologies to build a behavior model, and use machine learning models such as support vector machines, random forests, etc. for classification. In this embodiment, a convolutional neural network is used to build the behavior model.

[0054] Step 2.2: Use the existing database data and the driver's historical driving data to form a training set and train the behavior model.

[0055] Step 2.3: Use the trained behavior model to evaluate the real-time driving data to obtain a driving behavior evaluation result, wherein the driving behavior evaluation result includes standard driving behavior and non-standard driving behavior, and the risk level is determined according to the severity of the non-standard driving behavior.

[0056] The evaluation of real-time driving data using the trained behavior model is divided into the following steps:

[0057] (1) Driver behavior recognition: Analyze the driver's facial expressions, eye movement characteristics, head posture and other information to identify whether the driver is in a state that affects driving safety, such as fatigue, distraction, or emotional fluctuations.

[0058] More specifically, facial expression recognition: Use deep learning models to analyze the driver's facial images captured by the camera in real time to identify facial expression features such as fatigue, distraction, anger, etc. that may affect driving safety.

[0059] In a specific implementation, the real-time analysis of the driver's facial image captured by the camera using a deep learning model is based on the following aspects:

[0060] 1. Recognition based on eye status.

[0061] 1 Blink frequency:

[0062] Under normal circumstances, the driver's blinking frequency is within a certain range. When the driver is tired, the blinking frequency will change. Generally speaking, the normal blinking frequency is about 10-20 times per minute. If the blinking frequency is too high (such as more than 30 times per minute, it may be caused by dry eyes, etc., but it may also be a sign of early fatigue) or too low (such as less than 5 times per minute), it may be a sign of fatigue. For example, an extremely tired driver may unconsciously close his eyes for a long time. This is a response of the brain to fatigue, trying to rest by reducing visual information input.

[0063] 2. Eye closure time:

[0064] This is a key indicator. When blinking normally, the eye closure time is short, usually between 0.1-0.4 seconds. When the driver is tired, the eye closure time may be extended. If the eye closure time exceeds 1.5 seconds, there is a high suspicion of fatigue driving. Some studies have found through the analysis of a large amount of experimental data that when the driver drives continuously for 3-4 hours, the number of times the eye closure time exceeds 1 second will increase significantly.

[0065] 3. Eyelid drooping degree:

[0066] The eyelids of a tired driver will droop significantly. By analyzing the image of the eye area, the distance or angle between the eyelid and the eyeball can be measured to determine the degree of eyelid droop. For example, using a computer vision algorithm, without locating the eye feature points, the vertical distance between the edge of the upper eyelid and the center of the eyeball is calculated. Under normal circumstances, this distance is relatively stable. When the driver is tired, this distance will decrease, and the eyelid will be closer to the eyeball, covering part of the eyeball.

[0067] 4. Yawning:

[0068] Yawning is a clear sign of physical fatigue. When yawning, the driver's mouth will open wide and the facial muscles will have a specific stretching movement. Yawning can be detected by detecting the degree of opening and closing of the mouth and the movement pattern of the facial muscles. Generally speaking, when yawning, the mouth opens at an angle of about 60-90 degrees, and the muscles in the cheeks, chin and other parts will have obvious stretching movements. And yawning is usually accompanied by a slight backward tilt or shaking of the head. These comprehensive movement characteristics can be detected by the camera installed in the car and the corresponding image recognition algorithm.

[0069] 2. Recognition based on head posture.

[0070] 1. Nodding action:

[0071] When the driver is tired, the head may nod unconsciously. During normal driving, the head position is relatively stable. However, when the driver is tired, the head may nod more frequently and with a larger amplitude as the vehicle shakes. The movement trajectory of the head can be tracked by sensors or cameras installed in the car. For example, an inertial measurement unit (IMU) or a vision-based head posture estimation method can be used. If the head nods up and down multiple times in a short period of time (such as 10-15 seconds), and the nodding amplitude exceeds a certain angle (such as 10-15 degrees), it may indicate that the driver is in a state of fatigue.

[0072] 2. Head tilt degree:

[0073] A fatigued driver may have his head tilted to one side. This is because muscle fatigue makes it difficult for the head to maintain a normal vertical position. By estimating the head posture in three dimensions, the angle between the central axis of the head and the vertical direction can be calculated. Under normal circumstances, this angle is small, but when the driver is fatigued, the angle may increase. For example, in some experiments, when the driver's fatigue index reaches a certain level, the angle between the central axis of the head and the vertical direction may exceed 15-20 degrees, tilting to one side.

[0074] 3. Recognition based on facial muscle status.

[0075] 1. Degree of facial relaxation:

[0076] The facial expression of a tired driver is usually relaxed. Under normal circumstances, the facial muscles have a certain tension, the eyes are bright, and the eyebrows and other facial features are in normal position. When tired, the facial muscle tension decreases, the eyebrows may be pulled down, and the corners of the mouth may droop slightly. The degree of relaxation of the expression can be determined by analyzing the movement and morphology of the key facial muscle groups. For example, the facial action coding system (FACS) is used to analyze the action units of facial muscles. If multiple action units related to relaxed expression (such as drooping eyebrows, relaxed cheek muscles, etc.) are detected at the same time, it can be inferred that the driver may be in a state of fatigue.

[0077] 2. Frowning frequency and degree:

[0078] When the driver is fatigued, he may frown frequently due to strong light, lack of concentration, etc. At the same time, the degree of frowning may be more severe than normal. The degree of frowning can be judged by analyzing the muscle contraction in the glabella area. For example, the degree of frowning can be quantified by measuring the depth and length of the glabella wrinkles. Under normal circumstances, the glabella wrinkles are shallow and short. When the driver is fatigued and frowns, the glabella wrinkles will deepen and lengthen. The frequency of frowning will also increase. When driving normally, you may frown rarely, but when you are fatigued, you may frown more than 3-5 times per minute.

[0079] In order to prevent misjudgment, this embodiment can determine whether the driver is in a fatigue state based on the comprehensive situation of the above-mentioned identification indicators.

[0080] Eye tracking: Through the eye tracking technology in computer vision technology or the camera-based eye estimation algorithm, it is determined whether the driver's eyes are focused on the road ahead or the dashboard, and distraction behavior is identified.

[0081] Physiological status monitoring: Analyze biometric sensor data, such as heart rate, blinking frequency, etc., to assess the driver's fatigue level and concentration.

[0082] (2) Driving action analysis: Analyze signals such as steering wheel angle, accelerator / brake pedal pressure changes, and vehicle posture changes to identify whether the driver's operation complies with safe driving principles, such as oversteering, sudden braking and starting, and inappropriate following distance.

[0083] More specifically, vehicle motion state analysis: analyzing sensor data such as accelerometers, gyroscopes, and wheel speed sensors, and calculating key parameters such as the vehicle's real-time speed, acceleration, steering angle, and yaw rate.

[0084] Driving behavior pattern recognition: Use machine learning models such as support vector machines and random forests to classify driving action data and identify irregular driving behaviors such as sudden acceleration, sudden braking, oversteering, and inappropriate following distance.

[0085] (3) Environmental factor analysis: Analyze external environmental factors, such as weather conditions, road complexity, traffic flow, potential hazards, and other external environments that may have an impact on driver behavior.

[0086] Risk assessment and classification: Combining the driver's behavioral characteristics with the results of driving action analysis, the pre-trained risk assessment model is used to quantify the risk level of the current driving behavior and divide it into three levels: low risk, medium risk, and high risk, providing a basis for subsequent decision-making.

[0087] (4) Risk assessment and classification:

[0088] More specifically, risk feature extraction: combining driver behavior characteristics with driving action analysis results, extracting a set of risk assessment indicators, such as fatigue index, distraction duration, frequency of emergency operations, etc.

[0089] Risk model construction: Use historical driving data to train risk assessment models, such as logistic regression, neural networks, etc. The model input is risk characteristics and the output is risk level. In this embodiment, the risk level is set to low risk, medium risk and high risk.

[0090] Real-time risk assessment: The driver behavior and driving action data collected in real time are input into the risk assessment model to obtain the risk level of the current driving behavior.

[0091] In a specific implementation, driver behaviors that are prone to accidents, such as fatigue driving, risky behaviors, and random behaviors, are classified as high risk. Among them, risky behaviors include speeding, running red lights, and sharp turns. Random behaviors include not keeping a safe distance, changing lanes at will, not using turn signals, and not giving way to pedestrians. Bad habits such as not using seat belts, using lights inaccurately, and leaving hands off the steering wheel for a long time are classified as medium risk. If the number of irregular behaviors during this trip does not reach the set threshold, it is considered low risk.

[0092] Step 3: Generate corresponding intervention decisions based on the driving behavior assessment results.

[0093] Specifically, Figure 4 As shown, the behavior assessment results in step 2 are classified into low risk, medium risk and high risk. If it is determined to be medium risk or high risk, a standardized behavior intervention decision is specified and a driving suggestion is generated. For medium risk, only voice and visual warnings are performed, and for high risk, in addition to voice and visual warnings, ADAS system intervention is also included.

[0094] In this embodiment, non-standard driving behaviors are classified into behaviors affected by self-factors and behaviors affected by environmental factors. Corrective analysis is performed on behaviors affected by self-factors to generate internal intervention decisions. Environmental data when environmental factors affect behaviors are extracted, and external intervention decisions are generated for behaviors affected by environmental factors in combination with the environmental data.

[0095] Step: 3.1: Based on the results of behavior recognition and evaluation, the AI ​​assistant makes corresponding intervention decisions in real time:

[0096] Step 3.1: Early warning.

[0097] Internal factors intervene in decision-making: For medium- and high-risk driving behaviors, the AI ​​assistant outputs customized voice prompts through the in-car speakers, such as "Please keep a safe distance", "It is recommended that you take a proper rest to prevent fatigue driving", "Please give way to pedestrians", "Please obey traffic rules and do not run red lights", etc., and displays corresponding warning icons on the dashboard, HUD or other display screens.

[0098] 1) Warning rule setting: Set corresponding warning trigger rules according to the risk level, such as "Please keep a safe distance" when the risk is medium, and "It is recommended that you take a proper rest to prevent fatigue driving" when the risk is high.

[0099] 2) Speech synthesis and output: When the warning rule is triggered, the preset voice prompt is converted into an audio signal using speech synthesis technology and played to the driver through the vehicle's speakers.

[0100] 3) Visual warning display: Display corresponding warning icons or text messages on the instrument panel, HUD or other display screens to enhance the visual warning effect.

[0101] External factors intervene in decision-making: Reset voice prompts and visual warnings based on environmental factors, such as the voice prompts "The road is slippery on rainy days, please be careful not to make sharp turns" and "The light is dim at night, please turn on the low beam".

[0102] Step 3.2: Assisted driving intervention: When necessary, the AI ​​assistant can automatically adjust the vehicle's assisted driving function parameters (such as adaptive cruise control, lane departure warning, emergency brake assist, etc.) through deep integration with the vehicle's ADAS system, fine-tune the vehicle's driving status, and guide the vehicle to return to the standard driving trajectory to reduce the risk of accidents.

[0103] More specifically, Figure 5 As shown, it is deeply integrated with the ADAS system:

[0104] 1) AI assistant is deeply integrated with the vehicle's ADAS system (such as adaptive cruise control, lane keeping, pre-collision warning, etc.) to obtain its control interface authority.

[0105] 2) Intervention strategy setting:

[0106] Internal intervention decision: Set the intervention strategy of the assisted driving function according to the risk level and driving scenario, such as moderately slowing down the vehicle speed, fine-tuning the steering angle, and enhancing lane keeping strength.

[0107] External intervention decision: Based on the internal intervention decision, the intervention strategy is re-formulated in combination with environmental factors such as rainy days, nights, and narrow roads.

[0108] 3) Real-time command sending: When assisted driving intervention is required, the AI ​​assistant sends corresponding control commands to the ADAS system through the control interface to guide the vehicle back to the standard driving trajectory.

[0109] Step 3.3: Personalized suggestion push: Based on the driver's historical driving habits, current road conditions, weather conditions and other factors, the AI ​​assistant provides real-time, personalized driving suggestions, such as recommending the appropriate speed, prompting the reasonable time to change lanes, and reminding of precautions for special sections of the road.

[0110] 1) Driving advice generation: Based on real-time traffic conditions, weather information, the driver's historical driving habits and other data, a recommendation algorithm is used to generate personalized driving advice, such as recommended speed, lane change timing, precautions for special road sections, etc.

[0111] 2) Display and broadcast of suggestions: Driving suggestions are displayed in text or graphic form on the in-vehicle screen and broadcast to the driver through speech synthesis technology.

[0112] Step 3.4: Generate a test question bank based on the scenarios and specific behaviors of the driver's irregular behaviors, and conduct real-time monitoring and early warning of behaviors encountered in similar scenarios during the rest of the driving process. Among them, the early warning includes the methods described in steps 3.1, 3.2, and 3.3.

[0113] Step 4: Use intervention decisions to correct driver behavior and conduct ongoing testing.

[0114] In this embodiment, the intervention decision is a staged intervention decision, the number of stages is determined according to the severity of the non-standard driving behavior, and the number of subsequent stages and the content of the decision are adjusted in real time according to the driver's correction feedback. The continuous test is to test the driver's corrective behavior for the implementation effect of the intervention decision at each stage and generate the driver's correction feedback.

[0115] In a specific implementation, score thresholds and stages are set, and the driver is tested at each stage node. For example, if the driver tends to make sharp turns when encountering a bend, it is set to one stage for every ten turns. During the ten turns, if the driver maintains normal speed and angle to turn more than eight times, and the score reaches a passing score in the test after one stage, then one stage of the test is reduced, and only voice prompts can be given in the next stage without intervention of assisted driving. Completion of all stages of the test and passing is considered a successful correction, and no advance warning or test is required at the next turn. If the driver's test does not meet the standards or, then an additional stage of assessment and warning will be added, and assisted driving intervention will be given priority in the next stage.

[0116] The above assessment settings can be customized according to actual conditions.

[0117] If the non-standard driving behavior involves external factors, such as speeding in rainy days, in order to correct the driver as soon as possible, in addition to monitoring in real scenes, simulation training can also be carried out according to different scenarios, which is also counted as part of the driver's test.

[0118] The subsequent interactive feedback and guidance during the intervention process are as follows: Figure 5 As shown, a driving behavior report is generated based on the driving behavior data and stored in an established driving knowledge base for the driver to query, and remote monitoring and guidance are provided.

[0119] Step 4.1: Visual feedback: The AI ​​assistant presents a concise and clear driving behavior evaluation and improvement suggestion chart through the in-vehicle display screen, allowing the driver to intuitively understand his or her driving performance and clearly identify the areas that need improvement.

[0120] 1) Driving behavior report generation: Generate driving behavior reports regularly (such as at the end of each trip, daily summary), including driver behavior scores, problem behavior statistics, improvement suggestions, etc.

[0121] 2) Driving behavior charting: Use charts (such as bar charts, line charts, radar charts, etc.) to visualize driving behavior data, so that drivers can intuitively understand their own driving performance.

[0122] 3) Report and chart display: Integrate driving behavior reports and charts into the designated interface of the vehicle screen for the driver to view at any time.

[0123] Step 4.2: Interactive training: By integrating a rich driving knowledge base and simulation training resources, the AI ​​assistant can provide customized online tutorials, video explanations, virtual driving exercises and other interactive training content based on the driver's shortcomings or common mistakes, helping the driver improve his driving skills and ability to cope with complex road conditions.

[0124] 1) Construction of driving knowledge base: Collect and organize various driving knowledge, regulations, skills, etc. to form a structured driving knowledge base.

[0125] 2) Online tutorial production: Develop graphic tutorials, short video tutorials for common driving problems, and virtual simulation training software for specific driving skills.

[0126] 3) Training resource push: Based on the driver’s driving shortcomings or common mistakes, relevant online tutorials and video resources are pushed through the in-vehicle screen or supporting mobile applications to encourage the driver to participate in learning and practice.

[0127] Step 4.3: Remote management and data sharing: The AI ​​assistant connects to the cloud server to achieve real-time synchronization, storage and analysis of driving data. Car owners and fleet managers can remotely view driver behavior reports, receive warning notifications, and conduct remote guidance through a dedicated application. At the same time, the cloud platform can conduct in-depth mining of large-scale driver behavior data and provide decision-making support such as driving behavior research and risk area identification for traffic management departments.

[0128] 1) Cloud service connection: The AI ​​assistant system establishes a stable data transmission channel with the cloud server to realize real-time upload and download of data.

[0129] 2) Driving data storage and management: The cloud server receives and stores driving behavior data from the in-vehicle AI assistant, and supports data query, screening, statistics and other functions.

[0130] 3) Remote monitoring and guidance: Car owners and fleet managers can log in to the cloud platform through a dedicated mobile application to view driver behavior reports, receive early warning notifications, and remotely send guidance information or suggestions to drivers.

[0131] Through the above implementation steps, the AI ​​assistant of the present invention can fully realize the four major functions of data collection, behavior recognition and evaluation, decision making, interactive feedback and guidance, and effectively assist the driver to standardize driving behavior and improve driving safety.

[0132] The present invention designs a driving behavior regulation method based on complex scenarios. By integrating multiple data sources such as in-vehicle cameras, biosensors, and CAN bus data, an all-round and multi-level driver behavior monitoring system is constructed to accurately capture subtle changes in the driver's physiological state, visual focus, gestures, and other subtle changes, providing rich materials for in-depth understanding of driving behavior. By adopting AI technologies such as deep learning algorithms, image recognition technology, and natural language processing, the data collected by the perception module are analyzed and feature extracted in real time to identify the driver's behavior pattern, emotional state, concentration, and whether there is illegal driving behavior. The decision-making module determines whether there is potentially dangerous driving behavior and determines the corresponding intervention strategy. Finally, through speech synthesis, tactile feedback, visual prompts, etc., early warning information, corrective instructions, or safety suggestions are conveyed to the driver to guide him to adjust his driving behavior, and personalized driving training tutorials and improvement suggestions are provided.

[0133] Embodiment 2:

[0134] Embodiment 2 of the present invention provides a second aspect of the present invention provides a driving behavior regulation system based on complex scenarios, such as Figure 6 As shown, including:

[0135] The data acquisition module is configured to acquire driving data and environmental data of various drivers in real time, and pre-process the driving data and environmental data.

[0136] The data acquisition module uses a variety of sensor devices installed in the car, including but not limited to high-definition cameras, biometric sensors, vehicle motion sensors, and CAN bus interfaces to collect multivariate data such as the driver's physiological state, line of sight direction, and operating actions. The data is then preprocessed and sent to the data fusion and synchronization unit for synchronization and storage. Among them, vehicle motion sensors such as accelerometers, gyroscopes, wheel speed sensors, etc. Driver physiological state such as fatigue level and concentration. Operating actions such as steering, acceleration, braking force and other vehicle dynamic information, vehicle dynamic information such as speed, position, heading, driving path, etc. The data collection process ensures the real-time and accuracy of the information, and can provide a basic source of teaching data for subsequent behavior recognition and evaluation.

[0137] In a specific implementation, the data acquisition module is configured as follows:

[0138] 1. Hardware deployment: Deploy various sensor devices in the vehicle, including but not limited to high-definition cameras, biometric sensors, vehicle motion sensors, and connections to the vehicle CAN bus interface to obtain vehicle driving data. Among them, the high-definition camera is installed in front of the driver to capture facial expressions, line of sight, etc., biometric sensors such as heart rate sensors and eye tracking sensors are used to monitor the driver's physiological state, and vehicle motion sensors such as accelerometers, gyroscopes, wheel speed sensors, etc. are used to capture vehicle dynamic information.

[0139] 2. Data interface configuration: Ensure that the data interface of each sensor device and the on-board Al assistant system is correctly connected and configured to ensure the stability and real-time performance of data transmission.

[0140] 3. Data preprocessing: Perform necessary preprocessing on the received raw sensor data, such as filtering and denoising, data format conversion, unit unification, etc., to facilitate subsequent module processing.

[0141] 4. Data synchronization and storage: The pre-processed data is transmitted to the processing unit of the AI ​​assistant in real time, and stored in the local cache or on-board storage device as needed. At the same time, it is synchronized to the cloud server through the on-board communication module, such as 4G / 5G, Wi-Fi, etc., to support remote management and data analysis.

[0142] The behavior recognition and evaluation module is configured to use the behavior model to identify and analyze the pre-processed driving data to obtain a driving behavior evaluation result.

[0143] The behavior recognition and evaluation module uses advanced technologies such as deep learning, computer vision, and pattern recognition to process and analyze the collected data in real time.

[0144] The decision-making module is configured to generate corresponding intervention decisions based on the driving behavior evaluation results, wherein non-standard driving behaviors are classified into behaviors affected by self-factors and behaviors affected by environmental factors, correction analysis is performed on behaviors affected by self-factors to generate internal intervention decisions, environmental data when environmental factors affect behaviors are extracted, and external intervention decisions are generated for behaviors affected by environmental factors in combination with the environmental data.

[0145] The decision-making module uses the results of behavior recognition and evaluation to enable the AI ​​assistant to make corresponding intervention decisions in real time.

[0146] The correction feedback module is configured to utilize intervention decisions to perform behavioral correction and continuity testing on the driver.

[0147] According to the correction feedback module, drivers can view detailed driving behavior reports during parking breaks, participate in online training courses, and continuously improve their driving quality. Remote managers can use the cloud platform to effectively supervise and guide driver behavior.

[0148] Embodiment three:

[0149] Embodiment 3 of the present invention provides a medium on which a program is stored. When the program is executed by a processor, the steps in the method for regulating driving behavior based on complex scenarios as described in Embodiment 1 of the present invention are implemented.

[0150] Embodiment 4:

[0151] Embodiment 4 of the present invention provides a device, including a memory, a processor, and a program stored in the memory and executable on the processor. When the processor executes the program, the steps in the method for regulating driving behavior based on complex scenarios as described in Embodiment 1 of the present invention are implemented.

[0152] The steps involved in the above embodiments 2, 3 and 4 correspond to the method embodiment 1. For the specific implementation methods, please refer to the relevant description part of embodiment 1.

[0153] Those skilled in the art should understand that the modules or steps of the present invention described above can be implemented by a general-purpose computer device, or alternatively, they can be implemented by a program code executable by a computing device, so that they can be stored in a storage device and executed by the computing device, or they can be made into individual integrated circuit modules, or multiple modules or steps therein can be made into a single integrated circuit module for implementation. The present invention is not limited to any specific combination of hardware and software.

[0154] Although the above describes the specific implementation mode of the present invention in conjunction with the accompanying drawings, it is not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art on the basis of the technical solution of the present invention without creative work are still within the scope of protection of the present invention.

[0155] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A driving behavior regulation method based on complex scenarios, characterized in that: The following steps are involved: Acquire driving data and environmental data of multiple drivers in real time, and pre-process the driving data and environmental data; Use the behavior model to identify and analyze the preprocessed driving data to obtain the driving behavior evaluation results; Generate corresponding intervention decisions based on the driving behavior assessment results. Among them, classify non-standard driving behaviors into behaviors affected by self-factors and behaviors affected by environmental factors. Perform correction analysis on behaviors affected by self-factors to generate internal intervention decisions. Extract environmental data when behaviors affected by environmental factors occur, and combine environmental data to generate external intervention decisions on behaviors affected by environmental factors. Use decision intervention to modify driver behavior and conduct ongoing testing.

2. The method for regulating driving behavior based on complex scenarios as claimed in claim 1, characterized in that: Driving data includes the driver's physiological state, line of sight and operating actions.

3. The method for regulating driving behavior based on complex scenarios as claimed in claim 1, characterized in that: The specific steps to use the behavior model to identify and analyze the preprocessed driving data are: Use deep learning algorithms to build behavioral models; The existing database data is combined with the driver's historical driving data to form a training set, and the behavior model is trained; The trained behavior model is used to evaluate the real-time driving data to obtain the driving behavior evaluation results.

4. The method for regulating driving behavior based on complex scenarios as claimed in claim 3, characterized in that: The driving behavior assessment result includes standard driving behavior and non-standard driving behavior, and the risk level is judged according to the severity of the non-standard driving behavior.

5. The method for regulating driving behavior based on complex scenarios as claimed in claim 1, characterized in that: Environmental data includes weather condition data, road condition complexity data, traffic flow data and potential hazard source data.

6. The method for regulating driving behavior based on complex scenarios as claimed in claim 1, characterized in that: The intervention decision is a staged intervention decision, the number of stages is determined according to the severity of the non-standard driving behavior, and the subsequent number of stages and decision content are adjusted in real time according to the driver's correction feedback.

7. The method for regulating driving behavior based on complex scenarios as claimed in claim 6, characterized in that: The continuous test is to test the driver's corrective behavior on the implementation effect of the intervention decision at each stage and generate driver correction feedback.

8. A driving behavior regulation system based on complex scenarios, characterized in that: include: A data acquisition module is configured to acquire driving data and environmental data of various drivers in real time and pre-process the driving data and environmental data; A behavior recognition and evaluation module is configured to use the behavior model to recognize and analyze the preprocessed driving data to obtain a driving behavior evaluation result; A decision-making module is configured to generate corresponding intervention decisions according to the driving behavior evaluation results, wherein the non-standard driving behaviors are classified into behaviors affected by self-factors and behaviors affected by environmental factors, correction analysis is performed on behaviors affected by self-factors to generate internal intervention decisions, environmental data when behaviors affected by environmental factors occur are extracted, and external intervention decisions are generated for behaviors affected by environmental factors in combination with the environmental data; The correction feedback module is configured to utilize intervention decisions to perform behavioral correction and continuity testing on the driver.

9. A computer-readable storage medium, characterized in that: A plurality of instructions are stored therein, and the instructions are suitable for being loaded by a processor of a terminal device and executing the complex scenario-based driving behavior regulation method described in any one of claims 1-7.

10. A terminal device, characterized in that: It includes a processor and a computer-readable storage medium, the processor is used to implement each instruction; the computer-readable storage medium is used to store multiple instructions, and the instructions are suitable for being loaded by the processor and executing the complex scenario-based driving behavior standardization method described in any one of claims 1-7.

Images