Emotion road condition early warning method and system of AI intelligent agent and traffic large model

Abstract

The application provides an emotion road condition early warning method and system of an AI intelligent agent and a traffic large model, and belongs to the technical field of intelligent networked vehicles.The method comprises the following steps: acquiring vehicle behavior data from a roadside edge AI intelligent agent in real time, and acquiring driver emotion data from a vehicle-mounted AI intelligent agent in real time; inputting the vehicle behavior data and the driver emotion data into a traffic emotion large model, and outputting emotion road condition indexes by the traffic emotion large model; inputting the emotion road condition indexes, real-time traffic flow data and real-time environmental factor data into a time sequence fusion model, and outputting emotion road condition index trend prediction results by the time sequence fusion model; and generating early warning information according to the emotion road condition index trend prediction results.The application breaks through the physical limitations of traditional traffic monitoring, first introduces driver emotions into a road condition evaluation system, establishes a new dimension of "person-vehicle-road" collaborative perception, and upgrades the traffic system from simple physical state monitoring to comprehensive perception containing psychological characteristics.

Description

Technical Field

[0001] This invention relates to the field of intelligent connected vehicle technology, and in particular to an AI agent and a traffic big data model for emotional road condition early warning method and system. Background Technology

[0002] With the acceleration of urbanization and the surge in motor vehicle ownership, traffic congestion and accident prevention have become global challenges. Traditional intelligent transportation systems focus on physical layer monitoring, such as traffic flow, speed, and road occupancy. However, practice shows that most traffic accidents are directly related to abnormal driver emotions—drivers' reaction speed decreases in an anxious state, and the probability of illegal lane changes due to anger increases.

[0003] The existing technology system has three major flaws:

[0004] 1. Data dimensions are limited to the physical parameters of "vehicle-road", creating a perception blind spot. Mainstream roadside systems, such as millimeter-wave radar and cameras, only collect the vehicle's motion status, completely severing the connection between the driver's emotions and driving behavior, and failing to capture the transmission chain of "emotional alienation - dangerous behavior - traffic conflict".

[0005] 2. Data processing is fragmented and isolated. Driver physiological data, such as heart rate and EEG, are collected independently by onboard equipment, while vehicle behavior data belongs to the traffic management system. The lack of a standardized integration mechanism between the two makes it impossible to quantify and analyze the impact of emotions on traffic flow.

[0006] 3. The response mechanism is in a "passive triggering" stage. Existing early warning systems, such as collision warnings and congestion alerts, rely on events that have already occurred and have not formed a proactive safety logic of "emotional prediction - behavioral intervention," thus missing the golden window for risk mitigation. Current technologies only classify emotions through single physiological signals and have not been linked to traffic scenarios; they are based on historical traffic data and have not incorporated dynamic emotional variables. Summary of the Invention

[0007] This invention aims to solve at least one of the aforementioned problems in the existing technology, and innovatively constructs a dual-dimensional perception system of "vehicle behavior - driver emotion". Through a traffic emotion big data model, it achieves cross-domain data fusion and forms a closed-loop chain of "perception - fusion - prediction - response", filling the technological gap in the field of active safety.

[0008] In a first aspect, embodiments of the present invention provide an emotion-based traffic condition early warning method based on an AI agent and a large traffic model, comprising:

[0009] Real-time vehicle behavior data is obtained from roadside edge AI agents, and real-time driver emotion data is obtained from in-vehicle AI agents.

[0010] The vehicle behavior data and driver emotion data are input into the traffic emotion big data model, and the traffic emotion big data model outputs the emotion road condition index.

[0011] The emotional traffic condition index, real-time traffic flow data, and real-time environmental factor data are input into the time series fusion model, and the time series fusion model outputs the trend prediction result of the emotional traffic condition index.

[0012] Early warning information is generated based on the trend prediction results of the aforementioned emotional traffic conditions index.

[0013] In a preferred embodiment, the method further includes: pushing warning information to vehicles in the corresponding target area according to the level of the warning information.

[0014] In a preferred embodiment, the method further includes: after the vehicle-mounted AI agent performs differentiated response measures based on the level of the warning information, acquiring real-time driving behavior change data and driver emotion change data collected by the vehicle-mounted AI agent, and inputting them into the traffic emotion big data model to update the model parameters.

[0015] In a preferred embodiment, the step of inputting the vehicle behavior data and driver emotion data into a traffic emotion big data model, and the traffic emotion big data model outputting emotion-based road condition indicators, includes:

[0016] The vehicle behavior data and driver emotion data are preprocessed;

[0017] The preprocessed vehicle behavior data and driver emotion data are input into the traffic emotion big data model for feature fusion.

[0018] Two types of emotional road condition indicators are generated based on the fusion feature vector: anxiety index and conflict probability.

[0019] In a preferred embodiment, the preprocessing step of the vehicle behavior data and driver emotion data includes:

[0020] Data alignment: Based on GPS timestamps, the vehicle behavior data and driver emotion data are synchronized to a unified timeline, and the vehicle behavior data of the target vehicle is bound to the driver emotion data of the target vehicle through vehicle ID association;

[0021] Data cleaning: For the vehicle behavior data, incomplete trajectories caused by sensor obstruction are removed; for the driver emotion data, abrupt changes are smoothed using Kalman filtering.

[0022] Standardization: The vehicle behavior data and driver emotion data are mapped to the [0,1] interval using the min-max normalization method;

[0023] The step of inputting the preprocessed vehicle behavior data and driver emotion data into the traffic emotion big data model for feature fusion includes:

[0024] Scene recognition: Based on preprocessed vehicle behavior data, the current scene is determined, which is either smooth traffic, slow traffic, or congestion. The scene confidence score is output through a random forest model.

[0025] Weighting: The weights of vehicle behavior features and driver emotion features are calculated using the following formulas: Driver emotion feature weight = 0.5 + 0.1 × (congestion confidence - smooth traffic confidence), Vehicle behavior feature weight = 1 - Driver emotion feature weight.

[0026] Feature fusion: The weighted feature vectors of vehicle behavior features and driver emotion features are concatenated by a multilayer perceptron, processed by a 3-layer fully connected network, and output as a fused feature vector.

[0027] The steps for generating two types of emotional road condition indicators—anxiety index and conflict probability—based on fused feature vectors include:

[0028] The anxiety index is calculated using a weighted summation formula: Anxiety Index = 0.3 × Vehicle Behavior Abnormality + 0.7 × Group Emotion Abnormality, where the Vehicle Behavior Abnormality is generated by fusing abnormal vehicle behavior data from vehicle behavior data; and the Group Emotion Abnormality is the average of the anxiety probabilities of all drivers within a preset area.

[0029] The probability of conflict is calculated using a logistic regression model: Conflict probability = 1 / [1+exp (-(w1×anxiety index+w2×traffic density+b))], where w1 and w2 are model parameters, and b is a bias term.

[0030] In a preferred embodiment, the step of inputting the sentiment traffic condition index, real-time traffic flow data, and real-time environmental factor data into a time-series fusion model, and the time-series fusion model outputting a trend prediction result for the sentiment traffic condition index, includes:

[0031] The emotional road condition index, real-time traffic flow data, and real-time environmental factor data from the first preset time period are input into the time series fusion model;

[0032] The time-series fusion model uses the sliding window method to learn short-term evolution patterns;

[0033] The time-series fusion model outputs the trend prediction results of anxiety index and conflict probability at a second preset time in the future.

[0034] In a preferred embodiment, generating early warning information based on the trend prediction result of the emotional road condition index includes:

[0035] A red alert is triggered when the probability of conflict is greater than the first probability threshold; an orange alert is triggered when the probability of conflict is greater than or equal to the second probability threshold and less than or equal to the first probability threshold; and a yellow alert is triggered when the probability of conflict is greater than or equal to the third probability threshold and less than the second probability threshold.

[0036] The early warning information includes: the boundary of the risk area, the expected duration, and the scope of impact;

[0037] The step of pushing early warning information to vehicles in the corresponding target area according to the level of the early warning information includes:

[0038] When the warning information level is red, the warning information is pushed to all vehicles within a preset distance threshold around the boundary of the risk area via V2X broadcast; when the warning information level is orange or yellow, the warning information is pushed to vehicles about to enter the boundary of the risk area, wherein the push protocol adopts the ITS-G5 standard.

[0039] In a second aspect, embodiments of the present invention provide an emotion-based traffic condition early warning system based on an AI agent and a large traffic model, configured to implement any of the methods described in the first aspect, the system comprising:

[0040] The acquisition module is used to acquire vehicle behavior data in real time from the roadside edge AI agent and driver emotion data in real time from the in-vehicle AI agent.

[0041] The emotional road condition index module is used to input the vehicle behavior data and driver emotional data into the traffic emotional big model, and the traffic emotional big model outputs the emotional road condition index.

[0042] The prediction module is used to input the emotional traffic condition index, real-time traffic flow data, and real-time environmental factor data into the time series fusion model, and the time series fusion model outputs the trend prediction result of the emotional traffic condition index.

[0043] The early warning module is used to generate early warning information based on the trend prediction results of the emotional road condition index.

[0044] Thirdly, embodiments of the present invention provide an electronic device, including:

[0045] One or more processors;

[0046] Memory, used to store one or more programs;

[0047] When the one or more programs are executed by the one or more processors, the one or more processors implement any of the methods described in the first aspect.

[0048] Fourthly, embodiments of the present invention provide a computer-readable medium storing a computer program that, when executed by a processor, implements the steps of any of the methods described in the first aspect.

[0049] Beneficial effects of this invention:

[0050] 1. Pioneering a new paradigm of emotion-traffic integration: This invention breaks through the physical limitations of traditional traffic monitoring and for the first time incorporates driver emotions into the road condition assessment system, establishing a new dimension of collaborative perception of "people-vehicle-road", upgrading the traffic system from simple physical state monitoring to comprehensive perception that includes psychological characteristics.

[0051] 2. Constructing a proactive safety defense system: This invention achieves a shift from "post-event handling" to "pre-event intervention" by predicting emotional risk trends in advance, giving vehicles sufficient time to adjust their strategies, reducing the possibility of traffic conflicts at the source, and improving the inherent safety level of the traffic system.

[0052] 3. Achieving intelligent evolution capability of the system: Through the closed-loop design of "perception-fusion-prediction-response-feedback", this invention enables the system to continuously learn new patterns in actual traffic scenarios, constantly optimize models and strategies, and adapt to complex and ever-changing traffic environments.

[0053] 4. Enhance the human-machine collaborative experience: This invention incorporates driver emotions into the system decision-making dimension, and alleviates driving stress and enhances driving comfort and safety through personalized cockpit adjustment and intelligent route planning, thereby upgrading the transportation system from a "tool attribute" to a "service attribute". Attached Figure Description

[0054] Figure 1 This is a schematic diagram of a method for providing early warning of emotional traffic conditions using an AI agent and a large traffic model, as provided in an embodiment of the present invention.

[0055] Figure 2 This is a schematic flowchart of an optional implementation of step S2 provided in an embodiment of the present invention.

[0056] Figure 3 This is a structural block diagram of an electronic device provided in an embodiment of the present invention.

[0057] Figure 4 The overall process framework diagram of another AI intelligent agent and traffic big model emotion traffic condition early warning system provided in the embodiment of the present invention is shown.

[0058] Figure 5 The flowchart of the traffic emotion big data model fusion process corresponding to another AI intelligent agent and traffic big data model emotion traffic condition early warning system provided in the embodiment of the present invention.

[0059] Figure 6 The following is a flowchart of the early warning response for an AI agent and traffic big data model-based emotion-based traffic condition early warning system provided in an embodiment of the present invention. Detailed Implementation

[0060] To enable those skilled in the art to better understand the technical solutions of the present invention, exemplary embodiments of the present invention are described below in conjunction with the accompanying drawings, including various details of the embodiments of the present invention to aid understanding. These should be considered merely exemplary. Therefore, those skilled in the art should recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of the present invention. Similarly, for clarity and brevity, descriptions of well-known functions and structures are omitted in the following description.

[0061] Where there is no conflict, the various embodiments of the present invention and the features thereof may be combined with each other.

[0062] As used herein, the term “and / or” includes any and all combinations of one or more related enumerated entries.

[0063] The terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the invention. As used herein, the singular forms “a” and “the” are intended to include the plural forms as well, unless the context clearly indicates otherwise. It will also be understood that when the terms “comprising” and / or “made of” are used in this specification, the presence of the stated feature, integral, step, operation, element, and / or component is specified, but the presence or addition of one or more other features, integrals, steps, operations, elements, components, and / or groups thereof is not excluded. Terms such as “connected” or “linked” are not limited to physical or mechanical connections but can include electrical connections, whether direct or indirect.

[0064] Unless otherwise specified, all terms used herein (including technical and scientific terms) have the same meaning as commonly understood by one of ordinary skill in the art. It will also be understood that terms such as those defined in commonly used dictionaries should be interpreted as having the meaning consistent with their meaning in the context of the relevant art and the invention, and will not be interpreted as having an idealized or overly formal meaning unless expressly so defined herein.

[0065] In the technical solution of this invention, the collection, storage, use, processing, transmission, provision, and disclosure of user personal information all comply with relevant laws and regulations and do not violate public order and good morals. The use of user data in this technical solution follows relevant national laws and regulations (e.g., the "Information Security Technology - Personal Information Security Specification"). For example: appropriate measures are taken for personal information access control; restrictions are imposed on the display of personal information; the purpose of using personal information does not exceed the scope of direct or reasonable association; and explicit identity targeting is eliminated when using personal information to avoid precisely locating a specific individual.

[0066] Some abbreviations and key terms in this invention are defined as follows:

[0067] 1. AI Agent (AIA): An artificial intelligence entity with the ability to perceive the environment, make autonomous decisions and perform collaborative execution, and achieve closed-loop control of specific traffic tasks through multimodal data interaction.

[0068] 2. Traffic Emotion Large Model (TEGM): A deep learning model trained on a large scale of parameters, which integrates vehicle behavior and driver physiological and psychological data to generate emotional road condition features and predict traffic conflict risks.

[0069] 3. Emotional Traffic Conditions (ER): A dynamic traffic condition representation that integrates traffic flow status and the emotional characteristics of group drivers. Core indicators include anxiety index and conflict probability.

[0070] 4. Roadside Edge Unit (RSEU): Intelligent sensing nodes deployed at the edge of the road, integrating multiple types of sensors and edge computing modules to realize real-time acquisition and preprocessing of vehicle behavior data.

[0071] 5. Multimodal Biosensing Cockpit (MBC): An intelligent in-vehicle system integrating physiological sensors, acoustic modules, and computer vision units to collect driver emotional characteristic data.

[0072] 6. Vehicle-Road-Cloud Integrated Platform (VRC): A distributed architecture that enables data interconnection between the vehicle, roadside, and cloud, supporting real-time data aggregation, model inference, and control command issuance.

[0073] In this embodiment, for ease of description, the cloud is used as the execution subject in the following description. The cloud can be a software module, or other electronic devices capable of performing the following functions.

[0074] Figure 1 This is a flowchart illustrating an emotion-based traffic condition early warning method based on an AI agent and a large traffic model, as provided in an embodiment of the present invention. Figure 1 As shown, the method includes:

[0075] Step S1: Obtain vehicle behavior data in real time from the roadside edge AI agent and driver emotion data in real time from the in-vehicle AI agent;

[0076] Step S2: Input the vehicle behavior data and driver emotion data into the Traffic Emotion Big Data Model (TEGM), and the Traffic Emotion Big Data Model outputs an emotion-based road condition index.

[0077] Step S3: Input the emotional traffic condition index, real-time traffic flow data (such as flow rate, density, etc.), and real-time environmental factor data (weather, time period, etc.) into the time series fusion model (LSTM+Transformer), and the time series fusion model outputs the trend prediction result of the emotional traffic condition index.

[0078] Step S4: Generate early warning information based on the trend prediction results of the emotional road condition index.

[0079] Among them, the roadside edge AI intelligent agent integrates three types of devices: LiDAR, high-definition camera and acoustic array to acquire vehicle behavior data in real time. For example, LiDAR can acquire abnormal behaviors such as sudden braking and continuous lane changes, high-definition camera can acquire vehicle light status data, and acoustic array can acquire vehicle horn sound source data. The in-vehicle AI intelligent agent is embedded in the intelligent cockpit to form a multimodal biosensing cockpit (MBC) to capture the driver's physiological and behavioral characteristics in real time as driver emotion data. The multimodal biosensing cockpit can deploy three types of core devices: photoelectric heart rate sensor, infrared camera and microphone array. The photoelectric heart rate sensor is attached to the steering wheel to collect heart rate variability data, the infrared camera is installed on the dashboard to capture facial micro-expressions, and the microphone array is integrated into the roof to collect voice signals.

[0080] This invention breaks through the limitation of "physical parameter dependence" in traditional traffic monitoring. It innovatively constructs a dual-dimensional data collection system of "vehicle behavior - driver emotion" through roadside edge AI agents and vehicle-mounted AI agents, solving the problem of the separation between physical parameters and psychological characteristics, and realizing multimodal dynamic representation of emotional road conditions.

[0081] This invention breaks through the data silo dilemma by establishing a deep correlation mechanism for heterogeneous data through a traffic sentiment big data model. This solves the problem of distortion in the mapping of emotion to risk caused by isolated data processing in existing technologies, and generates accurate sentiment-based road condition characteristic indicators.

[0082] This invention revolutionizes the traffic forecasting paradigm by establishing a prediction mechanism based on emotional dynamics. It breaks through the bottleneck of traditional road condition prediction relying solely on physical parameters, enabling early prediction of emotional risk trends and solving the problem of insufficient early warning timeliness.

[0083] In some embodiments, the method further includes: step S5, pushing warning information to vehicles in the corresponding target area according to the level of the warning information.

[0084] This invention can achieve precise delivery of warning messages to vehicles in different areas based on different warning information levels.

[0085] In some embodiments, the method further includes: step S6, after the vehicle-mounted AI agent performs differentiated response measures according to the level of the warning information, the method acquires real-time driving behavior change data (such as braking frequency and steering angle) and driver emotion change data (such as the magnitude of heart rate decrease) collected by the vehicle-mounted AI agent, and inputs them into the Traffic Emotion Model (TEGM) to update the model parameters and realize the dynamic optimization of the TEGM model.

[0086] This invention constructs a vehicle-road-cloud collaborative response architecture by matching differentiated response measures to the vehicle based on the level of early warning information through an in-vehicle AI agent. It innovates the "prediction-intervention" proactive control mode, solves the problem of "early warning-execution" disconnect in existing systems, and realizes closed-loop control from cloud-based early warning to proactive intervention at the vehicle end.

[0087] This invention, through a closed-loop design of "perception-fusion-prediction-response-feedback", can continuously learn new patterns in real traffic scenarios, constantly optimize models and strategies, and adapt to complex and ever-changing traffic environments.

[0088] In some embodiments, such as Figure 2 As shown, step S2, which involves inputting the vehicle behavior data and driver emotion data into the traffic emotion big data model, and the traffic emotion big data model outputting emotion-based road condition indicators, includes the following steps:

[0089] Step S21: Preprocess the vehicle behavior data and driver emotion data;

[0090] Step S22: Input the preprocessed vehicle behavior data and driver emotion data into the traffic emotion big data model for feature fusion;

[0091] Step S23: Generate two types of emotional road condition indicators based on the fused feature vector: Anxiety Index (AI) and Conflict Probability (CP).

[0092] In some embodiments, step S21, the preprocessing step of the vehicle behavior data and driver emotion data, includes:

[0093] Data alignment: Based on GPS timestamps, the vehicle behavior data and driver emotion data are synchronized to a unified timeline. By associating vehicle IDs, the vehicle behavior data of the target vehicle is bound with the driver emotion data of the target vehicle, forming a "one vehicle, one emotion" associated dataset.

[0094] Data cleaning: For the vehicle behavior data, incomplete trajectories caused by sensor obstruction (such as when the missing rate is >30%) are removed, and for the driver emotion data, abrupt changes (such as a sudden increase in HRV heart rate caused by coughing) are smoothed using Kalman filtering.

[0095] Standardization: The vehicle behavior data and driver emotion data are mapped to the [0,1] interval using the min-max normalization method to eliminate the impact of dimensional differences on the fusion calculation;

[0096] Step S22, which involves inputting the preprocessed vehicle behavior data and driver emotion data into the traffic emotion big data model for feature fusion, includes the following steps:

[0097] Scene recognition: Based on preprocessed vehicle behavior data, determine the current scene (traffic status), which is either smooth, slow, or congested. Output the scene confidence score through a random forest model, such as congestion confidence score = 0.85.

[0098] Weighting: The weights of vehicle behavior features and driver emotion features are calculated using the following formulas: Driver emotion feature weight = 0.5 + 0.1 × (Congestion confidence - Smooth traffic confidence), Vehicle behavior feature weight = 1 - Driver emotion feature weight; For example, when the scenario is congested, the driver emotion feature weight is 0.6 and the vehicle behavior feature weight is 0.4; when the scenario is smooth traffic, the driver emotion feature weight is 0.4 and the vehicle behavior feature weight is 0.6; in slow-moving scenarios, the driver emotion feature weight and the vehicle behavior feature weight are both 0.5.

[0099] Feature fusion: The weighted feature vectors of vehicle behavior features and driver emotion features are concatenated by a multilayer perceptron (MLP), and then processed by a 3-layer fully connected network to output a fused feature vector (64 dimensions).

[0100] Step S23, which generates two types of emotional road condition indicators based on the fused feature vector: anxiety index and conflict probability, includes the following steps:

[0101] Anxiety index is calculated using a weighted summation formula: Anxiety index = 0.3 × Vehicle behavior abnormality + 0.7 × Group emotion abnormality, where vehicle behavior abnormality is generated by fusing abnormal vehicle behavior data, including the frequency of sudden braking, horn intensity, etc.; and group emotion abnormality is the average of the anxiety probability of all drivers within a preset area.

[0102] The probability of conflict is calculated using a logistic regression model: Conflict probability = 1 / [1 + exp (-(w1×anxiety index + w2×traffic density + b))], where w1 and w2 are model parameters (obtained through training with historical data), and b is the bias term.

[0103] In some embodiments, step S3, which involves inputting the sentiment traffic condition index, real-time traffic flow data, and real-time environmental factor data into a time-series fusion model, and the time-series fusion model outputting a trend prediction result for the sentiment traffic condition index, includes the following steps:

[0104] Input the emotional road condition index, real-time traffic flow data, and real-time environmental factor data from the first preset time (e.g., 30 minutes) into the time series fusion model;

[0105] The time-series fusion model uses a sliding window method (e.g., a window size of approximately 10 minutes) to learn short-term evolution patterns;

[0106] The time-series fusion model outputs the trend prediction results of anxiety index and conflict probability at a second preset time in the future (e.g., 15 minutes).

[0107] In some embodiments, step S4, generating early warning information based on the trend prediction result of the sentiment road condition index, includes:

[0108] A red alert (mandatory intervention) is triggered when the conflict probability is greater than the first probability threshold (e.g., CP > 85%). An orange alert (recommended detour) is triggered when the conflict probability is greater than or equal to the second probability threshold and less than or equal to the first probability threshold (e.g., CP ∈ [70%, 85%]). A yellow alert (caution warning) is triggered when the conflict probability is greater than or equal to the third probability threshold and less than the second probability threshold (e.g., CP ∈ [50%, 70%)).

[0109] The early warning information includes: the boundary of the risk area, the expected duration, and the scope of impact;

[0110] Step S5, which involves pushing warning information to vehicles in the corresponding target area based on the warning information level, includes:

[0111] When the warning information level is red, the warning information is pushed to all vehicles within a preset distance threshold (e.g., 2 kilometers) around the boundary of the risk area via V2X broadcast; when the warning information level is orange or yellow, the warning information is pushed to vehicles about to enter the boundary of the risk area (based on navigation path prediction), wherein the push protocol adopts the ITS-G5 standard to ensure reliable transmission in a weak network environment.

[0112] In some embodiments, in step S6, after the vehicle-mounted AI agent performs differentiated response measures based on the level of the warning information, real-time driving behavior change data and driver emotion change data collected by the vehicle-mounted AI agent are acquired and input into the traffic emotion big data model to update the model parameters:

[0113] The vehicle-mounted AI agent matches differentiated response measures based on the level of the warning information, including at least one of the following:

[0114] Upon receiving a red alert, the vehicle AI agent activates a defensive driving mode, which includes at least one of the following modes: increasing following distance, reducing maximum speed, and enhancing lane keeping assist. It also triggers assisted braking when dangerous behavior (such as sudden steering wheel turns) is detected in real time.

[0115] The in-vehicle AI agent performs personalized interventions based on the driver's emotional characteristics and historical preferences: for anxious drivers, measures include: playing alpha wave music, dimming the interior lights, or turning on at least one of the following: for angry drivers, initiating voice reassurance;

[0116] The in-vehicle AI agent calculates risk-avoidance routes by combining a 3D emotional road condition map with real-time navigation data. The 3D emotional road condition map is generated based on the anxiety index and conflict probability. The route score formula for the risk-avoidance route is: Route Score = 0.6 × Emotional Risk Value + 0.3 × Distance Length + 0.1 × Time Taken, where the emotional risk value is the average conflict probability (CP) of the area traversed by the route. Risk-avoidance routes are recommended in order of increasing score, with priority given to routes with the lowest scores.

[0117] Based on the same inventive concept, embodiments of the present invention also provide an emotion-based traffic condition early warning system for an AI agent and a large traffic model, configured to implement any of the methods described in the above embodiments, the system comprising:

[0118] The acquisition module is used to acquire vehicle behavior data in real time from the roadside edge AI agent and driver emotion data in real time from the in-vehicle AI agent.

[0119] The emotional road condition index module is used to input the vehicle behavior data and driver emotional data into the traffic emotional big model, and the traffic emotional big model outputs the emotional road condition index.

[0120] The prediction module is used to input the emotional traffic condition index, real-time traffic flow data, and real-time environmental factor data into the time series fusion model, and the time series fusion model outputs the trend prediction result of the emotional traffic condition index.

[0121] The early warning module is used to generate early warning information based on the trend prediction results of the emotional road condition index.

[0122] Based on the same inventive concept, embodiments of the present invention also provide an electronic device. Figure 3 This is a structural block diagram of an electronic device provided in an embodiment of the present invention. Figure 3As shown, an embodiment of the present invention provides an electronic device including: one or more processors 101, a memory 102, and one or more I / O interfaces 103. The memory 102 stores one or more programs, which, when executed by the one or more processors, cause the one or more processors to implement any of the methods described in the above embodiments; the one or more I / O interfaces 103 are connected between the processor and the memory, configured to enable information interaction between the processor and the memory.

[0123] The processor 101 is a device with data processing capabilities, including but not limited to a central processing unit (CPU); the memory 102 is a device with data storage capabilities, including but not limited to random access memory (RAM, more specifically SDRAM, DDR, etc.), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), and flash memory (FLASH); the I / O interface (read / write interface) 103 is connected between the processor 101 and the memory 102, and can realize information interaction between the processor 101 and the memory 102, including but not limited to a data bus (Bus).

[0124] In some embodiments, the processor 101, memory 102, and I / O interface 103 are interconnected via bus 104, and thus connected to other components of the computing device.

[0125] In some embodiments, the one or more processors 101 include a field-programmable gate array.

[0126] Based on the same inventive concept, embodiments of the present invention also provide a computer-readable medium. This computer-readable medium stores a computer program, wherein, when executed by a processor, the program implements the steps of any of the methods described in the above embodiments. The computer-readable storage medium may be a volatile or non-volatile computer-readable storage medium.

[0127] Figure 4 , Figure 5 , Figure 6 The following are the overall process framework diagram, traffic emotion big model fusion flowchart, and early warning response flowchart of another AI intelligent agent and traffic big model emotion traffic condition early warning system provided in the embodiments of the present invention.

[0128] 1. System Overall Architecture

[0129] This invention provides another AI agent and traffic big data model-based emotion-based traffic condition early warning system, which adopts a three-level collaborative architecture of "edge perception - cloud fusion - vehicle execution," achieving intelligent linkage throughout the entire process through a distributed AI agent cluster. The overall architecture comprises five core modules: a multi-source data acquisition layer, a data transmission layer, a vehicle-road-cloud fusion layer, a prediction and early warning layer, and a vehicle response layer. Each module interacts with data through standardized interfaces (based on the ISO 21434 protocol), forming a complete dynamic closed loop. See the overall process framework diagram below. Figure 4 .

[0130] 1. Multi-source data acquisition layer: As the "perception end" of the system, it integrates two major nodes: roadside edge AI agents and vehicle-mounted AI agents. Roadside nodes are deployed in key areas such as urban main roads and commercial districts to achieve full-area monitoring of vehicle behavior within a 500-meter radius; vehicle-mounted nodes are embedded in the smart cockpit to capture the physiological and behavioral characteristics of the driver in real time. The two ensure data consistency through a spatiotemporal synchronization mechanism (time error <10ms), laying the foundation for subsequent fusion analysis.

[0131] 2. Data Transmission Layer: A "5G+V2X" dual-link transmission network is constructed, employing differentiated transmission strategies. High-frequency roadside data, such as vehicle trajectories, are uploaded with low latency through 5G slicing technology, achieving an end-to-end latency of <50ms. Sensitive vehicle-mounted data, such as heart rate, is processed using an encryption protocol (AES-256) and then prioritized for transmission to roadside units via V2X direct communication before being forwarded to the cloud, balancing security and real-time performance.

[0132] 3. Vehicle-Road-Cloud Fusion Layer: Serving as the system's "data hub," this layer comprises a distributed storage cluster and the TEGM model inference engine. The storage cluster employs a hybrid architecture, combining an in-memory database with a distributed file system, enabling hierarchical management of hot data (nearly one hour) and cold data (historical data). The inference engine is deployed on a GPU cluster, supporting parallel processing of approximately 100,000 data points per second to complete the fusion calculation of multi-source features.

[0133] 4. Prediction and Early Warning Layer: Composed of a cloud-based AI intelligent swarm, forming the "decision brain," it includes a trend prediction module and an early warning generation module. The prediction module learns the coupling patterns of sentiment and traffic based on a temporal neural network; the early warning module generates differentiated information based on the risk level (low / medium / high), including elements such as risk area, duration of impact, and avoidance suggestions.

[0134] 5. Vehicle-side Response Layer: As the system's "execution terminal," it integrates vehicle control AI and interactive AI. The control AI interfaces with the vehicle's CAN bus to achieve dynamic adjustment of driving modes; the interactive AI interacts with the driver through multimodal methods such as voice and vision, providing personalized intervention solutions.

[0135] The multi-source data acquisition layer (multi-source data acquisition module) adopts a roadside edge AI intelligent agent acquisition mechanism and a multimodal biosensor cockpit acquisition mechanism. The multi-source data acquisition module breaks through the traditional single-dimensional data acquisition mode, constructs a "vehicle-driver" dual-dimensional perception system, and achieves holographic perception of traffic scenarios through multi-sensor collaboration and multimodal fusion technology.

[0136] Roadside edge AI agent acquisition mechanism: The roadside edge AI agent is deployed in the roadside edge unit (RSEU) and adopts a "multi-sensor heterogeneous fusion" technical architecture, which includes a three-level structure of perception layer, preprocessing layer and transmission layer;

[0137] Perception Layer: Integrates three types of devices: LiDAR, high-definition cameras, and acoustic arrays. LiDAR uses point cloud clustering algorithms to identify vehicle outlines and motion trajectories, accurately capturing abnormal behaviors such as sudden braking and continuous lane changes; high-definition cameras use a target detection network (YOLOv8) to extract vehicle lighting status (such as misuse of high beams and abnormal turn signals); the acoustic array uses beamforming technology to locate the source of horn sounds and filter out environmental noise interference.

[0138] Preprocessing layer: Equipped with an edge computing chip (computing power ≥20TOPS), it performs three core tasks: First, spatiotemporal registration, mapping different sensor data to a unified spatiotemporal coordinate system based on GPS timestamps and road coordinate systems; second, feature extraction, extracting vehicle behavior feature vectors from raw data, such as emergency braking intensity and horn frequency; and third, anomaly filtering, removing outliers caused by sensor malfunctions through the 3σ criterion to ensure data validity.

[0139] Transport layer: The "edge computing + edge storage" model is adopted to cache the collected data locally (cache for about 10 minutes) and upload it according to the "critical event priority" principle. When high-risk behavior is detected, such as continuous sudden braking, the immediate upload mechanism is triggered; regular data is uploaded in batches at 1-second intervals to reduce network load.

[0140] Multimodal Biosensor Cockpit Acquisition Mechanism: The multimodal biosensor cockpit (MBC) focuses on driver emotion perception and adopts a multi-dimensional acquisition strategy of "physiology-behavior-voice", which includes a sensing layer, a fusion layer and an output layer;

[0141] Sensing layer: Deploys three types of core devices: a photoelectric heart rate sensor attached to the steering wheel to collect heart rate variability (HRV) data; an infrared camera installed on the dashboard to capture facial micro-expressions, such as frowning and clenching teeth; and a microphone array integrated into the roof to collect voice signals and suppress ambient noise.

[0142] Fusion Layer: Multimodal data fusion is achieved based on the cockpit domain controller (computing power ≥15 TOPS). First, feature extraction is performed on the single-modal data: temporal indicators, such as SDNN, are calculated from HRV; dynamic changes of 68 feature points are extracted from facial images; and Mel-frequency cepstral series (MFCC) is extracted from speech. Then, the three types of features are fused through an attention mechanism to generate a driver emotion feature vector, achieving a three-level emotion classification of "calm-anxiety-anger".

[0143] Output layer: Using a standardized data interface, the fused emotional features are packaged with metadata such as vehicle ID and timestamp, and transmitted to the T-BOX via in-vehicle Ethernet. After encryption, the data is uploaded to the roadside unit. To protect privacy, the raw physiological data is stored locally only (with a storage period of 24 hours), and the uploaded data only includes feature vectors and classification results.

[0144] The Traffic Emotion Big Data Model Fusion Module is the core component of the system. It achieves deep fusion of cross-domain data on "vehicle behavior and driver emotion" through the Traffic Emotion Big Data Model (TEGM), generating accurate emotional traffic condition indicators. See the flowchart for the Traffic Emotion Big Data Model fusion process. Figure 5 As shown.

[0145] The traffic sentiment big data model fusion process includes: data preprocessing stage, feature fusion stage, and sentiment-based traffic condition generation stage.

[0146] Data preprocessing stage: Data preprocessing is the foundation of fusion computing and includes three steps: data alignment, cleaning, and standardization.

[0147] Data alignment: A "spatiotemporal dual-dimensional matching" strategy is adopted: In the time dimension, roadside and vehicle data are synchronized to a unified timeline based on GPS timestamps with an accuracy of about 10ms; in the spatial dimension, vehicle ID association technology is used to bind the behavioral data of a specific vehicle with its driver's emotional data to form a "one vehicle, one emotion" associated dataset.

[0148] Data cleaning: Two types of outlier handling: For roadside data, remove incomplete trajectories caused by sensor obstruction (missing rate > 30%); For emotion data, smooth HRV abrupt changes, such as sudden increases in heart rate caused by coughing, using Kalman filtering.

[0149] Standardization: The min-max normalization method is adopted to map vehicle behavior characteristics (such as the frequency of emergency braking and the duration of horn honking) and emotional characteristics (such as the probability of anxiety) to the [0,1] interval, thereby eliminating the influence of dimensional differences on the fusion calculation.

[0150] Feature fusion stage: Feature fusion adopts a "hierarchical attention mechanism" to dynamically adjust feature weights according to the traffic scene, and is divided into three steps: scene recognition, weight allocation, and feature fusion.

[0151] Scene recognition: The module determines the current traffic status (smooth / slow / congested) based on roadside data and outputs the scene confidence score through a random forest model, such as congestion confidence score = 0.85;

[0152] Weighting: The rules are as follows: When the scenario is congested, the driver's emotional characteristics are weighted at 0.6, and the vehicle behavior characteristics are weighted at 0.4; when the scenario is free-flowing, the weights are reversed; in slow-moving scenarios, each weight is 0.5. The calculation formula is as follows:

[0153] ① Emotional feature weight = 0.5 + 0.1 × (Congestion confidence score - Smooth traffic confidence score),

[0154] ② Vehicle behavior feature weight = 1 - Emotion feature weight;

[0155] Feature fusion: This is achieved through a multilayer perceptron (MLP), which concatenates the weighted feature vectors of the two classes, processes them through a 3-layer fully connected network, and outputs a fused feature vector (64 dimensions).

[0156] Emotional Roadmap Generation Stage: Based on the fused feature vectors, the TEGM model generates two core indicators: Anxiety Index (AI) and Conflict Probability (CP), and constructs a three-dimensional emotional roadmap.

[0157] The anxiety index is calculated using a weighted summation formula:

[0158] AI = 0.3 × Vehicle Behavior Anomaly Degree + 0.7 × Group Emotion Anomaly Degree

[0159] Among them, the abnormality of vehicle behavior is generated by integrating indicators such as the frequency of emergency braking and the intensity of horn honking; the abnormality of group emotions is the average of the anxiety probability of all drivers in the area.

[0160] The probability of conflict is calculated using a logistic regression model:

[0161] CP = 1 / [1 + exp (-(w1×AI + w2×flow density+ b))]

[0162] Where w1 and w2 are model parameters (obtained through training with historical data), and b is the bias term. When CP > 70%, the area is marked as a red high-risk zone.

[0163] Example of emotional traffic map data format:

[0164] {

[0165] "timestamp": "2023-10-01T08:30:00Z",

[0166] "region_id": "R10086",

[0167] "grid_size": 10, / / Grid size 10 meters

[0168] "emotion_grid": [

[0169] {

[0170] "grid_x": 120.0,

[0171] "grid_y": 30.0,

[0172] "anxiety_index": 0.82,

[0173] "conflict_prob": 0.75,

[0174] "risk_level": "high"

[0175] },

[0176] / / ... Other grid data ]

[0178] }

[0179] The predictive warning layer and the vehicle-side response layer (prediction and warning and vehicle-side response modules) construct a proactive safety closed loop of "prediction-warning-response." Through the synergy of cloud-based prediction and vehicle-side intervention, traffic risks can be mitigated in advance. See the warning and response flowchart. Figure 6 As shown.

[0180] Prediction and Early Warning Mechanism: The prediction and early warning mechanism comprises three stages: trend prediction, early warning generation, and precise push notification, forming a complete decision-making chain.

[0181] Trend prediction: Based on a time-series fusion model (LSTM+Transformer), the input includes three parts: the past 30 minutes of emotional traffic conditions (AI and CP), real-time traffic flow data (volume, density), and environmental factors (weather, time of day). The model uses a sliding window method (window size approximately 10 minutes) to learn short-term evolution patterns and outputs the anxiety index and conflict probability change curves for the next 15 minutes, with a prediction error of <10%.

[0182] Warning Generation: Based on the prediction results, a three-level warning system is triggered: when CP∈[50%,70%), a yellow warning is triggered (caution advised); when CP∈[70%,85%], an orange warning is triggered (alternative routes recommended); and when CP>85%, a red warning is triggered (mandatory intervention required). The warning information includes elements such as the risk area boundary, expected duration, and scope of impact, along with an analysis of the emotional transmission path, such as "anxiety from eastbound traffic has spread to intersections."

[0183] Precise push notifications: A "regional broadcast + precise targeting" strategy is employed: for red alert areas, notifications are sent to all vehicles within a 2-kilometer radius via V2X broadcast; for orange / yellow alerts, notifications are only sent to vehicles about to enter the area (based on navigation route prediction). The push protocol adopts the ITS-G5 standard to ensure reliable transmission in weak network environments.

[0184] Vehicle-side response strategy: Vehicle-side response is based on the principle of "tiered intervention," matching differentiated response measures according to the warning level, including three dimensions: driving control, cabin adjustment, and route planning.

[0185] Driving control dimension: Upon receiving a red alert, the vehicle control AI automatically activates a defensive driving mode—increasing following distance (by 20% compared to the normal mode), reducing the maximum speed (limiting it to 80% of the road speed limit), and enhancing lane keeping assist (reducing the steering correction threshold by 30%). Simultaneously, the system monitors driver actions in real time, triggering assisted braking when dangerous behavior is detected, such as sudden steering wheel movements.

[0186] Cockpit adjustment dimension: Based on the driver's emotional characteristics and historical preferences (stored in a local knowledge base), interactive AI performs personalized interventions. For anxious drivers, it automatically plays alpha wave music (60-80 BPM), dims the interior lights (brightness reduced to 50%), and activates the fragrance system (lavender essential oil); for angry drivers, it initiates voice reassurance, such as "Vehicles in the area ahead are agitated, it is recommended to maintain steady driving."

[0187] Route planning dimension: Combining a 3D emotional roadmap with real-time navigation data, three risk-avoidance routes are recommended. The route scoring formula is:

[0188] Route score = 0.6 × Emotional risk score + 0.3 × Distance length + 0.1 × Time spent.

[0189] The emotional risk value is the average CP (Constant Risk) of the areas traversed by the route. The system prioritizes recommending the routes with the lowest ratings and marks the risk distribution of each route on the AR navigation screen with different colors: red for high risk and green for low risk.

[0190] The vehicle-side response strategy also includes a response effect feedback mechanism: after the vehicle executes the response strategy, it collects changes in driving behavior (such as braking frequency and steering angle) and changes in driver emotions (such as the magnitude of heart rate drop) in real time, and feeds them back to the cloud through an encrypted channel for dynamic optimization of the TEGM model, updating the model parameters once an hour.

[0191] Those skilled in the art will understand that all or some of the steps, systems, and apparatuses disclosed above, and their functional modules / units, can be implemented as software, firmware, hardware, or suitable combinations thereof. In hardware implementations, the division between functional modules / units mentioned above does not necessarily correspond to the division of physical components; for example, a physical component may have multiple functions, or a function or step may be performed collaboratively by several physical components. Some or all physical components may be implemented as software executed by a processor, such as a central processing unit, digital signal processor, or microprocessor, or as hardware, or as an integrated circuit, such as an application-specific integrated circuit (ASIC). Such software can be distributed on a computer-readable storage medium, which may include computer storage media (or non-transitory media) and communication media (or transient media).

[0192] As is known to those skilled in the art, the term computer storage medium includes volatile and non-volatile, removable and non-removable media implemented in any method or technology for storing information, such as computer-readable program instructions, data structures, program modules, or other data. Computer storage media includes, but is not limited to, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM), static random access memory (SRAM), flash memory or other memory technologies, portable compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical disc storage, magnetic cartridges, magnetic tape, disk storage or other magnetic storage devices, or any other medium that can be used to store desired information and is accessible to a computer. Furthermore, it is known to those skilled in the art that communication media typically contain computer-readable program instructions, data structures, program modules, or other data in modulated data signals such as carrier waves or other transmission mechanisms, and may include any information delivery medium.

[0193] The computer-readable program instructions described herein can be downloaded from computer-readable storage media to various computing / processing devices, or downloaded via a network, such as the Internet, local area network, wide area network, and / or wireless network, to an external computer or external storage device. The network may include copper transmission cables, fiber optic transmission, wireless transmission, routers, firewalls, switches, gateway computers, and / or edge servers. A network adapter card or network interface in each computing / processing device receives the computer-readable program instructions from the network and forwards them to the computer-readable storage media in the respective computing / processing device.

[0194] The computer program instructions used to perform the operations of this invention may be assembly instructions, instruction set architecture (ISA) instructions, machine instructions, machine-dependent instructions, microcode, firmware instructions, state setting data, or source code or object code written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Smalltalk, C++, etc., and conventional procedural programming languages ​​such as the "C" language or similar programming languages. The computer-readable program instructions may be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving a remote computer, the remote computer may be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or may be connected to an external computer (e.g., via the Internet using an Internet service provider). In some embodiments, electronic circuitry, such as programmable logic circuitry, field-programmable gate arrays (FPGAs), or programmable logic arrays (PLAs), is personalized by utilizing state information from the computer-readable program instructions. This electronic circuitry can execute the computer-readable program instructions to implement various aspects of the invention.

[0195] The computer program product described herein can be implemented specifically through hardware, software, or a combination thereof. In one alternative embodiment, the computer program product is specifically embodied in a computer storage medium; in another alternative embodiment, the computer program product is specifically embodied in a software product, such as a software development kit (SDK), etc.

[0196] Various aspects of the present invention are described herein with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It should be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer-readable program instructions.

[0197] These computer-readable program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing apparatus to produce a machine such that, when executed by the processor of the computer or other programmable data processing apparatus, they create means for implementing the functions / actions specified in one or more blocks of the flowchart and / or block diagram. These computer-readable program instructions can also be stored in a computer-readable storage medium that causes a computer, programmable data processing apparatus, and / or other device to operate in a particular manner; thus, the computer-readable medium storing the instructions comprises an article of manufacture that includes instructions for implementing aspects of the functions / actions specified in one or more blocks of the flowchart and / or block diagram.

[0198] Computer-readable program instructions may also be loaded onto a computer, other programmable data processing apparatus, or other device to cause a series of operational steps to be performed on the computer, other programmable data processing apparatus, or other device to produce a computer-implemented process, thereby causing the instructions executed on the computer, other programmable data processing apparatus, or other device to perform the functions / actions specified in one or more boxes of a flowchart and / or block diagram.

[0199] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of the present invention. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of an instruction, which contains one or more executable instructions for implementing a specified logical function. In some alternative implementations, the functions marked in the blocks may occur in a different order than those shown in the drawings. For example, two consecutive blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, may be implemented using a dedicated hardware-based system that performs the specified function or action, or using a combination of dedicated hardware and computer instructions.

[0200] Example embodiments have been disclosed herein, and while specific terminology has been used, it is for illustrative purposes only and should be construed as such, and is not intended to be limiting. In some instances, it will be apparent to those skilled in the art that features, characteristics, and / or elements described in conjunction with particular embodiments may be used alone, or in combination with features, characteristics, and / or elements described in conjunction with other embodiments, unless otherwise expressly indicated. Therefore, those skilled in the art will understand that various changes in form and detail may be made without departing from the scope of the invention as set forth in the appended claims.

Claims

1. A method for emotion-based traffic condition early warning based on an AI agent and a large traffic model, characterized in that, include: Real-time vehicle behavior data is obtained from roadside edge AI agents, and real-time driver emotion data is obtained from in-vehicle AI agents. The vehicle behavior data and driver emotion data are input into the traffic emotion big data model, and the traffic emotion big data model outputs the emotion road condition index. The emotional traffic condition index, real-time traffic flow data, and real-time environmental factor data are input into the time series fusion model, and the time series fusion model outputs the trend prediction result of the emotional traffic condition index. Early warning information is generated based on the trend prediction results of the aforementioned emotional traffic conditions index.

2. The method according to claim 1, characterized in that, Also includes: Warning information is pushed to vehicles in the corresponding target area based on the warning level.

3. The method according to claim 1, characterized in that, Also includes: After the vehicle-mounted AI agent implements differentiated response measures based on the level of the warning information, it acquires real-time data on changes in driving behavior and driver emotions collected by the vehicle and inputs it into the traffic emotion big data model to update the model parameters.

4. The method according to claim 1, 2 or 3, characterized in that, The step of inputting the vehicle behavior data and driver emotion data into the traffic emotion model, and having the traffic emotion model output emotion-based road condition indicators, includes: The vehicle behavior data and driver emotion data are preprocessed; The preprocessed vehicle behavior data and driver emotion data are input into the traffic emotion big data model for feature fusion. Two types of emotional road condition indicators are generated based on the fusion feature vector: anxiety index and conflict probability.

5. The method according to claim 4, characterized in that, The preprocessing steps for the vehicle behavior data and driver emotion data include: Data alignment: Based on GPS timestamps, the vehicle behavior data and driver emotion data are synchronized to a unified timeline, and the vehicle behavior data of the target vehicle is bound to the driver emotion data of the target vehicle through vehicle ID association; Data cleaning: For the vehicle behavior data, incomplete trajectories caused by sensor obstruction are removed; for the driver emotion data, abrupt changes are smoothed using Kalman filtering. Standardization: The vehicle behavior data and driver emotion data are mapped to the [0,1] interval using the min-max normalization method; The step of inputting the preprocessed vehicle behavior data and driver emotion data into the traffic emotion big data model for feature fusion includes: Scene recognition: Based on preprocessed vehicle behavior data, the current scene is determined, which is either smooth traffic, slow traffic, or congestion. The scene confidence score is output through a random forest model. Weighting: The weights of vehicle behavior features and driver emotion features are calculated using the following formulas: Driver emotion feature weight = 0.5 + 0.1 × (congestion confidence - smooth traffic confidence), Vehicle behavior feature weight = 1 - Driver emotion feature weight. Feature fusion: The weighted feature vectors of vehicle behavior features and driver emotion features are concatenated by a multilayer perceptron, processed by a 3-layer fully connected network, and output as a fused feature vector. The steps for generating two types of emotional road condition indicators—anxiety index and conflict probability—based on fused feature vectors include: The anxiety index is calculated using a weighted summation formula: Anxiety Index = 0.3 × Vehicle Behavior Abnormality + 0.7 × Group Emotion Abnormality, where the Vehicle Behavior Abnormality is generated by fusing abnormal vehicle behavior data from vehicle behavior data; and the Group Emotion Abnormality is the average of the anxiety probabilities of all drivers within a preset area. The probability of conflict is calculated using a logistic regression model: Conflict probability = 1 / [1+exp (-(w1×anxiety index+w2×traffic density+b))], where w1 and w2 are model parameters, and b is the bias term.

6. The method according to claim 4, characterized in that, The step of inputting the sentiment traffic condition index, real-time traffic flow data, and real-time environmental factor data into the time-series fusion model, and having the time-series fusion model output the trend prediction result of the sentiment traffic condition index, includes: The emotional road condition index, real-time traffic flow data, and real-time environmental factor data from the first preset time period are input into the time series fusion model; The time-series fusion model uses the sliding window method to learn short-term evolution patterns; The time-series fusion model outputs the trend prediction results of anxiety index and conflict probability at a second preset time in the future.

7. The method according to claim 4, characterized in that, The step of generating early warning information based on the trend prediction results of the sentiment traffic indicator includes: A red alert is triggered when the probability of conflict is greater than the first probability threshold; an orange alert is triggered when the probability of conflict is greater than or equal to the second probability threshold and less than or equal to the first probability threshold; and a yellow alert is triggered when the probability of conflict is greater than or equal to the third probability threshold and less than the second probability threshold. The early warning information includes: the boundary of the risk area, the expected duration, and the scope of impact; The step of pushing early warning information to vehicles in the corresponding target area according to the level of the early warning information includes: When the warning information level is red, the warning information is pushed to all vehicles within a preset distance threshold around the boundary of the risk area via V2X broadcast; when the warning information level is orange or yellow, the warning information is pushed to vehicles about to enter the boundary of the risk area, wherein the push protocol adopts the ITS-G5 standard.

8. An emotion-based traffic condition early warning system based on an AI agent and a large-scale traffic model, characterized in that: The system, configured to implement the method as described in any one of claims 1 to 7, comprises: The acquisition module is used to acquire vehicle behavior data in real time from the roadside edge AI agent and driver emotion data in real time from the in-vehicle AI agent. The emotional road condition index module is used to input the vehicle behavior data and driver emotional data into the traffic emotional big model, and the traffic emotional big model outputs the emotional road condition index. The prediction module is used to input the emotional traffic condition index, real-time traffic flow data, and real-time environmental factor data into the time series fusion model, and the time series fusion model outputs the trend prediction result of the emotional traffic condition index. The early warning module is used to generate early warning information based on the trend prediction results of the emotional road condition index.

9. An electronic device, characterized in that, include: One or more processors; Memory, used to store one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the method as described in any one of claims 1 to 7.

10. A computer-readable medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method as described in any one of claims 1 to 7.

Images