Different-intelligence traffic subject interaction information model construction method based on three-dimensional digital model

Abstract

The invention relates to a different intelligence traffic subject interaction information model construction method based on a three-dimensional digital model. The method comprises the following steps: S1, constructing a multi-level scene based on spatial levels of a road traffic system and designing functions of the multi-level scene; s2, defining autonomous levels of the traffic system, and describing the perception capability, decision logic and execution precision of each terminal device of the vehicle road cloud of each level; s3, according to different attributes of traffic entities, designing a perception-transmission-decision-control four-stage traffic information circulation process and realizing unambiguous collaboration among different intelligence subjects; and S4, converting the standardized and defined information model into a plurality of instantiation units, and deploying the instantiation units into computing entities of various end-edge-cloud traffic subjects to support actual operation of a traffic service scene. According to the method, the problems of difficult cross-domain collaboration, non-uniform information models and insufficient global optimization caused by non-uniform intelligent level of traffic subjects in the prior art are effectively solved, and methodological support is provided for design, verification and implementation of a vehicle-road cloud integrated system.

Description

Technical Field

[0001] This invention belongs to the field of intelligent transportation system technology, specifically relating to a method for constructing a heterogeneous intelligent transportation subject interactive information model based on a three-dimensional digital model. Background Technology

[0002] The Vehicle-Road-Cloud Integration System (VRoad-Cloud Integration System) integrates vehicle-side perception, roadside infrastructure, and cloud computing capabilities. Through multi-dimensional information interaction, it enables real-time and efficient collaboration between vehicles, roads, and the cloud, achieving global perception, real-time sharing, collaborative decision-making, and efficient management of traffic elements such as traffic data, providing global optimization support for intelligent driving. The information flow between vehicles, roads, and the cloud involves four stages: perception, transmission, decision-making, and control. Specifically, vehicles primarily rely on autonomous driving perception and control capabilities; roads mainly use roadside perception, edge computing, and V2X communication for environmental supplementation and safety assurance; and the cloud is primarily responsible for global information fusion, traffic prediction, and optimized scheduling. However, on the one hand, the data from vehicles, roads, and the cloud suffer from inconsistencies in spatiotemporal benchmarks and heterogeneous communication protocols, leading to low efficiency in multi-source information fusion. On the other hand, different intelligent agents (such as L1-L4 vehicles, roadside equipment, and cloud platforms) have varying levels of understanding of traffic semantics, which can easily lead to instruction conflicts or execution deviations during collaborative decision-making. Furthermore, the increase in information entropy and the accumulation of latency in the perception-transmission-decision-control link will further restrict the response capability of the vehicle-road-cloud integrated system in highly dynamic scenarios. Moreover, existing technologies rely too much on centralized cloud processing, making it difficult to balance real-time performance, security, and cross-domain collaborative efficiency.

[0003] Heterogeneous Autonomy refers to the phenomenon of varying levels of intelligence among vehicle-side, roadside, and cloud-based devices in the current traffic environment. It encompasses heterogeneous traffic flow, heterogeneous road infrastructure, and heterogeneous cloud processing. Heterogeneous traffic flow refers to the coexistence of vehicles with different levels of automation and human drivers in the same space-time environment. Heterogeneous road infrastructure refers to the varying levels of intelligence in roadside sensing facilities, communication equipment, and edge computing units. Heterogeneous cloud processing refers to the differences in intelligence at the global level between cloud computing and data centers.

[0004] The harmonious coexistence of vehicle-side, road-side, and cloud-side intelligence is crucial for the transition from traditional transportation to intelligent transportation. Existing transportation systems are also transforming into autonomous transportation systems. However, there are significant semantic interaction barriers among existing intelligent entities. Transportation entities with different levels of intelligence exhibit orders of magnitude differences in perception capabilities, decision-making logic, and execution accuracy, and face difficulties in integrating multi-source heterogeneous data. Furthermore, existing related models have not yet established a complete semantic interoperability framework, which can easily lead to semantic ambiguity during collaborative decision-making, making it difficult to guarantee system reliability.

[0005] As can be seen from the above, the existing vehicle-road-cloud integrated system mainly has the following technical defects: 1. Limitations of scene design scale: Existing scene-based design has a small time span and spatial scale, mainly focusing on the interaction within a single vehicle or a limited vehicle range. It lacks the ability to describe the integration and matching relationship of all elements of "vehicle-road-cloud" at a large scale of road network, and cannot support the realization of global optimization tasks. 2. One-sidedness and lack of coordination in the system perspective: Existing technical solutions involving roadside, cloud and vehicle lack a holistic system perspective to coordinate the functional interaction and collaborative integration among vehicles, roads and cloud. Data barriers between subsystems have not been broken down, and basic data is difficult to integrate. 3. Neglecting heterogeneous characteristics and poor compatibility: Traditional hierarchical control models and most existing technical solutions assume that the controlled objects are homogeneous, failing to fully consider and define the huge differences in the level of autonomy of traffic entities, as well as the fundamental differences in perception capabilities, decision-making logic and execution accuracy. This makes it difficult for existing technical solutions to adapt to the complex environment of "hybrid traffic with different intelligences" that will exist for a long time in the future. 4. Lack of information flow model: Existing technologies fail to clearly depict the standardized flow process of data in the entire "perception-transmission-decision-control" chain. In particular, there is a lack of attention to semantic unification processing and interface standardization design, which makes it easy for traffic information to be ambiguous when transmitted between subjects with different levels of intelligence, resulting in low collaborative efficiency.

[0006] The closest existing technology to this invention is the hierarchical control model in traditional ITS. For example, in the common "center-area-intersection" three-level signal control architecture, the center level is responsible for strategic decision-making, the area level is responsible for coordination, and the intersection level is responsible for execution. Although this three-level control model embodies the hierarchical concept, its information flow is usually unidirectional and centralized (from top to bottom), and the type of information processed is limited (mainly signal timing parameters). It does not consider the capabilities of connected vehicles as mobile sensing sources and collaborative control nodes, nor can it adapt to the interaction needs between entities with different levels of intelligence, from manual driving to fully autonomous driving. It is evident that this three-level signal control architecture is essentially a relatively fixed control model oriented towards traditional traffic infrastructure, rather than an information model construction method for vehicle-road-cloud integrated systems. Therefore, this three-level control model is not suitable for vehicle-road-cloud integrated systems. Summary of the Invention

[0007] This invention is applicable to multi-entity information interaction and collaborative decision-making in scenarios such as vehicle-road cooperation, autonomous driving, and traffic management. It aims to address the core technical challenges of cross-level, cross-domain, and cross-task collaboration arising from significant differences in the intelligence levels, functional responsibilities, and information perception and processing capabilities of various transportation entities (vehicles, roads, and cloud computing) during the development of intelligent connected transportation systems. Specifically, this invention addresses the following three key issues: 1. Lack of a unified information model hinders cross-domain collaboration: Current technologies lack a unified and standardized information model to formally describe and express information at different spatial scales (lane level, intersection level, road network level, etc.) and at different functional stages (perception, transmission, decision-making, control, etc.). Therefore, traffic information from different sources and in different formats is difficult for various intelligent traffic entities (from A1-level manually driven vehicles to A4-level fully automated driving systems) to understand and share ambiguously, forming "information silos" and hindering the effective implementation of large-scale vehicle-road-cloud integrated collaborative applications. 2. Lack of system design methodology, making it difficult to adapt to heterogeneous intelligent traffic scenarios: Current system design and verification are mostly aimed at homogeneous, high-level intelligent transportation entities, lacking a systematic design methodology that can simultaneously cover the functions, performance, and interaction interfaces of traffic entities (vehicles, roadside equipment, and cloud platforms) at different levels of autonomy (A1-A4). Therefore, in the inevitable transitional phase of "heterogeneous intelligent traffic," it is difficult to design and verify complex application scenarios that require close collaboration between entities with different levels of intelligence (such as high-level intelligent connected vehicles using roadside information to assist decision-making while sharing roads with human-driven vehicles). 3. Unclear information flow process hinders global optimization: In vehicle-road-cloud collaborative systems, the efficient and reliable flow of information between entities at different levels and intelligence levels, and its eventual transformation into control commands, is a complex process. Existing technologies lack a clear definition and hierarchical analysis of the complete information flow chain of "perception-transmission-decision-control," resulting in system optimization often being limited to local aspects. This prevents the realization of global collaborative optimization from micro-level vehicle control to macro-level road network scheduling, hindering the full realization of the potential effectiveness of the integrated vehicle-road-cloud system.

[0008] To address the aforementioned technical problems, this invention provides a three-dimensional digital intelligent traffic subject interactive information model and its modeling method, which is a "vehicle-road-cloud" three-dimensional collaborative architecture scheme built based on high-precision digital twin technology. The aim is to realize the intelligent flow and interaction of traffic information throughout the four stages of "perception-transmission-decision-control". This traffic information model covers traffic subjects with different levels of intelligence in four levels of autonomy, including A1-A4. It can adapt to the differentiated needs of manual driving, low, medium and high-level autonomous driving, intelligent infrastructure and traffic cloud platforms, and realize micro, meso and macro traffic optimization control at the lane level, intersection level and road network level, respectively.

[0009] This invention specifically provides a method for constructing an interactive information model of intelligent transportation subjects based on a three-dimensional digital model, which can realize the representation of the state knowledge of intelligent transportation subjects in cross-domain and cross-task scenarios, including the following steps: S1: Multi-level scene and its functional design: Based on the spatial hierarchy of the road traffic system, construct a multi-level scene scale information model including lane level, intersection level and road network level, and describe the functional characteristics, logical connection and interaction between the information models at each level. S2: Intelligent Multi-Granularity Information Analysis: Defines the autonomy levels of the traffic system from A1 to A4, and describes the perception capabilities, decision-making logic, and execution accuracy of the vehicle-road-cloud devices at each autonomy level; S3: Multi-stage migration and evolution information analysis: Based on the different attributes of traffic entities and the information analysis process of the vehicle-road-cloud integrated system of intelligent traffic flow, a four-stage traffic information flow process including perception, transmission, decision-making and control is designed, and the standardized processing and interaction logic of the traffic information between devices with different levels of intelligence is described. S4: Information Model Instantiation and Deployment: The standardized information model is transformed into several instantiated units, namely the data structures, interface protocols and configuration files for specific execution, and deployed to the computing entities of various transportation entities in the end-edge-cloud to support the actual operation of transportation service scenarios.

[0010] The term "heterogeneous intelligence" refers to differences in intelligence level. This invention uses autonomy levels (A1-A4) for quantitative characterization.

[0011] The three-dimensional meaning of the three-dimensional digital model refers not only to the three dimensions of geometric space, but also to the three dimensions of the interactive information model constructed by this invention, namely spatial scale (hierarchy), intelligence level (grade), and information flow (stage), thereby providing a three-dimensional and structured framework for the interactive information model.

[0012] The "end-edge-cloud" architecture involves different levels of transportation entities in the transportation sector, including the following three categories: The term "terminal" refers to the terminal device responsible for data collection, including intelligent traffic cameras, in-vehicle terminals, and mobile devices such as smartphones. For example, intelligent cameras use the YOLOv8 algorithm to identify vehicle outlines; smartphone sensors capture vibration signals to assist in determining vehicle type.

[0013] An edge refers to an edge computing node, which is responsible for local data processing and real-time response, including gateway devices and edge servers. For example, traffic sensing facilities use edge computing nodes to perform data preprocessing, filter irrelevant information, and transmit critical data to the cloud.

[0014] The cloud refers to a cloud-based data center, which is responsible for global decision-making and long-term storage, including a dynamic traffic database and a cloud processing engine. For example, a navigation system can predict vehicle trajectories and generate avoidance routes 15 minutes in advance by fusing multi-dimensional data in the cloud.

[0015] The collaborative work of the "device-edge-cloud" system involves data collection at the terminal, preprocessing at the edge, and decision-making in the cloud, thus forming a complete intelligent traffic perception system.

[0016] Further, step S1 includes the following steps: S11: Construct the multi-level scene scale information model, including: Lane-level scale information model: Taking a single road segment or a single lane as the spatiotemporal scope, it considers the information interaction between intelligent vehicles and between vehicles and roadside equipment to achieve refined control of vehicles at the micro level. Intersection-level scale information model: Taking one or more intersections as the spatiotemporal range, it considers the information interaction between intelligent vehicles, between vehicles and roadside sensing devices, and between intersection signal control devices to realize multi-segment traffic connection and conflict management. Road network-scale information model: Taking the overall road network of a city or region as the spatiotemporal scope, it considers the aggregation of intelligent vehicle information, collaborative computing of roadside edge computing units, and collaborative decision-making of cloud decision-making systems to achieve global traffic optimization and scheduling across road segments and regions; S12: Describe the functions and interrelationships of the multi-level scene scale information model: The function of the lane-level scale information model is to describe the vehicle-to-vehicle and vehicle-to-infrastructure cooperative functions at the micro-local level, including car-following control, lane-changing cooperation, local obstacle avoidance, and traffic flow stability control within the lane. The function of the intersection-level scale information model is to describe the vehicle-road cooperation and conflict resolution functions within the scope of the one or more intersections at the intersection-level scale, including signal control optimization, intersection traffic strategy generation, and emergency vehicle priority passage. The function of the road network-scale information model is to describe the global optimization and scheduling functions of the intelligent connected vehicle system at the macro level at the road network-scale, including road network-level traffic flow prediction, cross-segment path planning, regional traffic balance control, and multi-source information fusion prediction. The interrelationships between the functions of each level of the scenario are described as follows: This describes the interaction and coordination between the functions of the lane-level information model, the intersection-level information model, and the road network-level information model. Under global scheduling, the function of the intersection-level information model interacts with the function of the road network-level information model as a meso-level coordination unit. Under local decision-making, the functions of the lane-level information model coordinate with the functions of the intersection-level information model to achieve multi-level linkage from macro to micro.

[0017] Further, step S2 includes the following steps: S21: Define the autonomy levels from A1 to A4: Level A1: Vehicles are primarily driven manually, roadside equipment operates according to a fixed plan, static information services are provided by the cloud, and humans are highly or completely involved in each specific process. Level A2: The vehicle has partial autonomous driving capabilities but requires human supervision. The roadside has moderate dynamic scheduling capabilities, the cloud can optimize a single subsystem, and human monitoring only intervenes in specific links when necessary. Level A3: The vehicle has conditional autonomous driving capabilities, the roadside has cross-system collaborative control capabilities, the cloud can form a global optimization scheduling plan, and humans can observe the whole process without intervening in specific aspects. A4 level: The vehicle achieves highly or fully automated driving, the roadside achieves fully automatic dynamic adjustment, and the cloud uses artificial intelligence algorithms to perform real-time global optimization and prediction, with no human intervention in any aspect. S22: Based on the different intelligence levels of traffic entities in the various levels of traffic systems, describe the traffic entities and their functions involved in the autonomous traffic systems from level A1 to level A4: Traffic entities and their functions in a Class A1 transportation system: The vehicles operating in the A1-level transportation system are mainly driven by humans. The perception and decision-making tasks mainly rely on the driver's real-time reaction. Basic status information, including vehicle speed and position, is obtained through basic sensors and uploaded to the cloud through V2X technology to realize information interaction and vehicle status perception. It can receive simple instructions from the roadside and the cloud, but its response capability is limited. Decision-making is still mainly done manually. The roadside signal control equipment in the A1-level traffic system operates on a fixed cycle or a preset scheme, which makes it impossible to respond to traffic conditions in real time; there are few traffic monitoring devices and the roadside perception capability is weak. The fixed period includes a fixed signal period, and the preset scheme includes scheduled bus departures; The cloud-based system in the A1-level transportation system only provides static information services and does not have the ability to analyze real-time data or perform dynamic scheduling. The static information service includes the publication of fixed routes and timetables; Traffic entities and functions of a Level A2 transportation system: The vehicles operating in the A2-level transportation system are mainly low-level autonomous driving. The vehicles can handle most perception and control tasks, but key decisions still require human intervention. The system obtains vehicle speed, position and operating mode through basic sensors, obtains road channelization information and vehicle-road relationship information through radar, and then uploads it to the cloud through V2X technology to achieve information interaction. It can receive and respond to instructions issued from the roadside and the cloud, but still requires human supervision. The roadside equipment in the A2-level traffic system has a medium level of dynamic scheduling capability. The roadside equipment acquires real-time traffic data through various sensors, but the amount of information acquired and interacted is small and the interaction is slow, and manual intervention is still required in critical situations. The roadside equipment can acquire real-time data including traffic flow and traffic events through various sensors, but the amount of information acquired and interacted is small and the interaction is slow, and manual intervention is still required in critical situations. The medium-level dynamic scheduling capability includes dynamic bus departure and taxi reservation scheduling. The cloud in the A2-level transportation system can provide static and partially dynamic information services, providing data support for dynamic scheduling of the system. However, the scheduling capability is limited to a single subsystem and lacks cross-system and cross-level collaborative scheduling capabilities. The single subsystem includes optimizations only for bus departure scheduling or signal control; Traffic entities and functions of a Level A3 transportation system: The vehicles operating in the A3-level transportation system have medium-to-high level autonomous driving functions, strong environmental perception and autonomous decision-making capabilities, and can maintain real-time interaction with the roadside and cloud. They can perceive and deeply integrate their own operating status and environmental factors, including external weather, lighting and obstacles, and thus can assess road segment delays in real time and dynamically identify and avoid potential driving conflicts in multi-vehicle cooperative scenarios. The roadside signal control equipment in the A3-level traffic system has high dynamic regulation and collaborative control capabilities and can perform cross-system linkage control, but there are still time delays and accuracy deviations in the optimization results. The cross-system linkage control includes dynamic optimization of bus routes and online dispatching of taxis; The cloud-based system in the A3-level transportation system can provide static and dynamic information services and can form a global optimized scheduling scheme. The global optimization scheduling scheme involves the interactive iteration of signal control strategy and path planning decision to achieve the optimal result; Traffic entities and functions of a Level A4 transportation system: The vehicles operating in the A4-level transportation system are equipped with high-level autonomous driving capabilities, capable of fully taking over driving tasks and operating according to the globally optimal strategy issued by the cloud, relying on high-speed, low-latency communication. The vehicle-side equipment can achieve full-dimensional, high-precision monitoring and fusion of operating status, fully grasping its own operating conditions; it can model and predict high-precision road network status in real time, and then achieve dynamic optimization by combining delay monitoring; and it can proactively predict and resolve potential driving conflicts in multi-vehicle collaborative scenarios. The roadside signal control equipment in the A4-level traffic system has fully automatic dynamic adjustment capabilities, enabling real-time optimization and control of the traffic system and supporting seamless coordination between normal and emergency situations. The transportation system includes a traffic light system, a public transportation system, a taxi system, and a road guidance system; The cloud-based A4-level transportation system can provide real-time, low-latency static and dynamic traffic information, and use artificial intelligence-driven global optimization algorithms to predict upcoming congestion sections, thereby generating real-time optimal scheduling and control schemes.

[0018] Furthermore, the information parsing process of the intelligent vehicle flow vehicle-road-cloud integrated system in step S3 is a phased model, which includes: Perception Phase: Vehicle-side and roadside equipment effectively filter out some noise and extract key feature information from multi-source heterogeneous data through feature learning methods, ensuring the accuracy and effectiveness of the input key feature information. The key feature information includes, but is not limited to, vehicle status data, road condition data, driving environment data, conflict situation data, and delay situation data. After processing the key feature information, clear and reliable traffic perception information can be provided, laying the foundation for the subsequent decision-making and control phases. The traffic perception information includes traffic event information, traffic participant information, traffic status information, and traffic signal information; Transmission Phase: To meet the need for efficient and unambiguous semantic interaction among intelligent transportation entities, an artificial intelligence technology, including natural language processing, is used to construct a semantically standardized expression knowledge graph that integrates multi-source information from vehicle, road, and cloud sources. This semantically standardized expression knowledge graph can uniformly transform multimodal information generated by various transportation objects into a semantically standardized expression form that adapts to the cognitive levels of intelligent transportation entities from A1 to A4, eliminating semantic differences that arise when the multi-source information is transmitted in various interaction scenarios, including vehicle-to-vehicle (V2V), vehicle-to-infrastructure (V2I), vehicle-to-network (V2N), road-to-network (I2N), and vehicle-to-human (V2P) communication. The knowledge graph is a graph-based data structure composed of nodes (entities, concepts) and edges (relationships) used for semantic reasoning. The knowledge graph in this invention is used to standardize traffic concepts (such as "red light" and "congestion"), enabling a consistent understanding of traffic concepts across different systems.

[0019] Decision-making stage: Providing scientific and reasonable decision support for traffic subjects. The intelligent traffic subjects deeply understand the multimodal information obtained in the perception stage and the semantic standardization information obtained in the transmission stage. After processing by edge or cloud computing, personalized decision response results of the intelligent multi-traffic subjects are generated.

[0020] The phased model utilizes advanced computational algorithms and data analysis techniques to deeply mine and analyze real-time information on the traffic environment, traffic flow, and road conditions, thereby providing scientific and reasonable decision support for each traffic entity. The key to the decision-making phase lies in accurately and quickly analyzing data and making intelligent decisions to adapt to the dynamically changing needs of the traffic system.

[0021] Control Phase: After semantic reconstruction by the information model, the decision response results are transformed into interactive operation instructions and control language adapted to traffic entities with different intelligence levels. This unified control language enables efficient collaboration between traffic entities with varying levels of intelligence. The control phase ensures that traffic entities can perform precise operations according to the decision results through accurate control instructions, such as adjusting driving routes, changing traffic signals, and optimizing public transportation scheduling.

[0022] Furthermore, the feature extraction in the perception stage employs deep learning models or digital signal processing algorithms to achieve feature extraction and noise reduction of multi-source heterogeneous data.

[0023] Furthermore, the deep learning model employs a convolutional neural network (CNN) for image feature extraction.

[0024] Furthermore, the deep learning model employs a point cloud network for LiDAR data processing.

[0025] Furthermore, the digital signal processing algorithm employs filtering or Fourier transform.

[0026] Furthermore, the optimization algorithm in the decision-making stage adopts a reinforcement learning algorithm according to the actual application scenario. The reinforcement learning algorithm enables the system to learn the optimal collaborative strategy autonomously through interaction with the environment.

[0027] Furthermore, the optimization algorithm in the decision-making stage adopts a swarm intelligence algorithm based on the actual application scenario, which can solve the distributed collaborative decision-making problem.

[0028] Furthermore, when dealing with low-level intelligent entities such as A1 level, the control commands issued during the control phase are transformed into more intuitive human-machine interface (HMI) prompts, rather than directly controlling the vehicle.

[0029] Furthermore, the human-machine interface (HMI) prompts include providing suggested operations to the driver through in-vehicle screen graphics and / or voice reminders.

[0030] Further, each instantiation unit in step S4 includes: a. Instantiate the unit name, number, and description; b. Corresponding instantiation process, target deployment entity, and the corresponding level of autonomy of the transportation system; c. Core configuration parameters, parameter descriptions, generation tools, and output formats; d. Traffic service scenario of instantiated unit.

[0031] Furthermore, the traffic service scenarios include vehicle cooperative lane changing scenarios, emergency vehicle priority passage scenarios, and cross-road segment route planning scenarios.

[0032] This invention relates to a wide range of technical fields, particularly those closely related to systems engineering (MBSE), the Internet of Things (IoT), edge computing (MEC), artificial intelligence (AI), and autonomous driving. The core solution of this invention represents the deep integration and innovative application of these technical fields in the field of intelligent transportation. Compared with existing hierarchical control models and fragmented system design schemes, this invention has the following significant advantages: 1. High degree of uniformity and standardization: The unified information model framework and design method provided by this invention eliminates the fragmented information models of various subsystems in the prior art. It ensures unambiguous understanding of information between different intelligent subjects through semantic standardization, realizes the expression of state knowledge of different intelligent transportation subjects in cross-domain and cross-task scenarios, and provides key support for breaking down data barriers and realizing the integration of basic data. 2. Strong systematicity and forward-looking perspective: This invention adopts the idea and framework of systematically constructing information models, that is, through multi-scale, hierarchical and full-process systems engineering methods, it coordinates the design of the collaborative relationship of all elements of vehicles, roads and cloud, overcoming the one-sided perspective of existing technologies; 3. Further enhance the overall optimization effect: By clearly defining the information closed-loop flow process from micro-perception to macro-decision and then to reverse control, this invention enables the vehicle-road-cloud integrated system to comprehensively utilize information across the entire domain for collaborative decision-making, thereby achieving a further leap from single-node optimization to global road network optimization. This technical effect is difficult to achieve in existing technologies that optimize local or single functions. 4. Provide core methodologies for multiple anticipated application scenarios: The information model constructed by this invention can be directly applied to multiple fields, including guiding the architectural design of a new generation of vehicle-road-cloud collaborative systems; serving as a basic model for simulation testing platforms to generate test scenarios and evaluate system performance; and serving as a theoretical basis and reference standard for the design of various interface protocols in actual vehicle-road-cloud systems. Attached Figure Description

[0033] Figure 1 This is a schematic diagram illustrating the workflow of the method for constructing an interactive information model of intelligent transportation subjects based on a three-dimensional digital model according to the present invention. Figure 2 This is a schematic diagram of the framework structure of the method for constructing a heterogeneous intelligent transportation subject interaction information model based on a three-dimensional digital model according to the present invention; Figure 3 This is a UML diagram illustrating the interaction of a lane-scale information model for vehicle following and cooperative control scenarios in the intelligent transportation subject interaction information model construction method based on a three-dimensional digital model of the present invention. Figure 4 This is a UML diagram illustrating the interaction of a bus priority collaborative control scenario in the intersection-level scale information model of the intelligent transportation subject interaction information model based on a three-dimensional digital model in the present invention. Figure 5 This is a UML diagram illustrating the interaction of a road network-level information model in the road network-coordinated signal control scenario in the intelligent traffic subject interaction information model construction method based on a three-dimensional digital model of the present invention. Detailed Implementation

[0034] The preferred embodiments of the present invention will now be described with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0035] To address the aforementioned technical problems, this embodiment provides a method for constructing an interactive information model of intelligent transportation entities based on a three-dimensional digital model, such as... Figure 1 and Figure 2 As shown, this technical solution achieves multi-level and cross-domain multi-entity collaborative decision-making and control by classifying and digitally representing traffic entity information in multiple dimensions and at multiple levels, including the following steps: S1: Multi-level scene and its functional design: Based on the spatial hierarchy of the road traffic system, construct a multi-level scene scale information model including lane level, intersection level and road network level, and describe the functional characteristics, logical connection and interaction between the information models at each level. S2: Intelligent Multi-Granularity Information Analysis: Defines the autonomy levels of the traffic system from A1 to A4, and describes the perception capabilities, decision-making logic, and execution accuracy of the vehicle-road-cloud devices at each autonomy level; S3: Multi-stage migration and evolution information analysis: Based on the different attributes of traffic entities, design a four-stage traffic information flow process including perception, transmission, decision-making and control, and describe the standardized processing and interaction logic of the traffic information between devices with different levels of intelligence. S4: Information Model Instantiation and Deployment: The standardized information model is transformed into several instantiated units, namely the data structures, interface protocols and configuration files for specific execution, and deployed to the computing entities of various transportation entities in the end-edge-cloud to support the actual operation of transportation service scenarios.

[0036] Optionally, step S1 includes the following steps: S11: Construct the multi-level scene scale information model, including: Lane-level scale information model: Taking a single road segment or a single lane as the spatiotemporal scope, it considers the information interaction between intelligent vehicles and between vehicles and roadside equipment to achieve refined control of vehicles at the micro level. Intersection-level scale information model: Taking one or more intersections as the spatiotemporal range, it considers the information interaction between intelligent vehicles, between vehicles and roadside sensing devices, and between intersection signal control devices to realize multi-segment traffic connection and conflict management. Road network-scale information model: Taking the overall road network of a city or region as the spatiotemporal scope, it considers the aggregation of intelligent vehicle information, collaborative computing of roadside edge computing units, and collaborative decision-making of cloud decision-making systems to achieve global traffic optimization and scheduling across road segments and regions; S12: Describe the functions and interrelationships of the multi-level scene scale information model: The function of the lane-level scale information model is to describe the vehicle-to-vehicle and vehicle-to-infrastructure cooperative functions at the micro-local level, including car-following control, lane-changing cooperation, local obstacle avoidance, and traffic flow stability control within the lane. The function of the intersection-level scale information model is to describe the vehicle-road cooperation and conflict resolution functions within the scope of the one or more intersections at the intersection level, including signal control optimization, intersection traffic strategy generation, and emergency vehicle priority passage. The function of the road network-scale information model is to describe the global optimization and scheduling functions of the intelligent connected vehicle system at the macro level at the road network-scale, including road network-level traffic flow prediction, cross-segment path planning, regional traffic balance control, and multi-source information fusion prediction. The interrelationships between the functions of each level of the scenario are described as follows: the interaction and coordination between the functions of the lane-level scale information model, the intersection-level scale information model, and the road network-level scale information model are described; under global scheduling, the function of the intersection-level scale information model interacts with the function of the road network-level scale information model as a meso-level coordination unit; under local decision-making, the function of the lane-level scale information model coordinates with the function of the intersection-level scale information model.

[0037] Optionally, step S2 includes the following steps: S21: Define the autonomy levels from A1 to A4: Level A1: Vehicles are primarily driven manually, roadside equipment operates according to a fixed plan, static information services are provided by the cloud, and humans are highly or completely involved in each specific process. Level A2: The vehicle has partial autonomous driving capabilities but requires human supervision. The roadside has moderate dynamic scheduling capabilities, the cloud can optimize a single subsystem, and human monitoring only intervenes in specific links when necessary. Level A3: The vehicle has conditional autonomous driving capabilities, the roadside has cross-system collaborative control capabilities, the cloud can form a global optimization scheduling plan, and humans can observe the whole process without intervening in specific aspects. A4 level: The vehicle achieves highly or fully automated driving, the roadside achieves fully automatic dynamic adjustment, and the cloud uses artificial intelligence algorithms to perform real-time global optimization and prediction, with no human intervention in any aspect. S22: Based on the different intelligence levels of traffic entities in the various levels of traffic systems, describe the traffic entities and their functions involved in the autonomous traffic systems from level A1 to level A4: Traffic entities and their functions in a Class A1 transportation system: The vehicles operating in the A1-level transportation system are mainly driven by humans. The perception and decision-making tasks mainly rely on the driver's real-time reaction. Basic status information, including vehicle speed and position, is obtained through basic sensors and uploaded to the cloud through V2X technology to realize information interaction and vehicle status perception. It can receive simple instructions from the roadside and the cloud, but its response capability is limited. Decision-making is still mainly done manually. The roadside signal control equipment in the A1-level traffic system operates on a fixed cycle or a preset scheme, which makes it impossible to respond to traffic conditions in real time; there are few traffic monitoring devices and the roadside perception capability is weak. The fixed period includes a fixed signal period, and the preset scheme includes scheduled bus departures; The cloud-based system in the A1-level transportation system only provides static information services and does not have the ability to analyze real-time data or perform dynamic scheduling. The static information service includes the publication of fixed routes and timetables; Traffic entities and functions of a Level A2 transportation system: The vehicles operating in the A2-level transportation system are mainly low-level autonomous driving. The vehicles can handle most perception and control tasks, but key decisions still require human intervention. The system obtains vehicle speed, position and operating mode through basic sensors, obtains road channelization information and vehicle-road relationship information through radar, and then uploads it to the cloud through V2X technology to achieve information interaction. It can receive and respond to instructions issued from the roadside and the cloud, but still requires human supervision. The roadside equipment in the A2-level traffic system has a medium level of dynamic scheduling capability. The roadside equipment acquires real-time traffic data through various sensors, but the amount of information acquired and interacted is small and the interaction is slow, and manual intervention is still required in critical situations. The roadside equipment can acquire real-time data including traffic flow and traffic events through various sensors, but the amount of information acquired and interacted is small and the interaction is slow, and manual intervention is still required in critical situations. The medium-level dynamic scheduling capability includes dynamic bus departure and taxi reservation scheduling. The cloud in the A2-level transportation system can provide static and partially dynamic information services, providing data support for dynamic scheduling of the system. However, the scheduling capability is limited to a single subsystem and lacks cross-system and cross-level collaborative scheduling capabilities. The single subsystem includes optimizations only for bus departure scheduling or signal control; Traffic entities and functions of a Level A3 transportation system: The vehicles operating in the A3-level transportation system have medium-to-high level autonomous driving functions, strong environmental perception and autonomous decision-making capabilities, and can maintain real-time interaction with the roadside and cloud. They can perceive and deeply integrate their own operating status and environmental factors, including external weather, lighting and obstacles, and thus can assess road segment delays in real time and dynamically identify and avoid potential driving conflicts in multi-vehicle cooperative scenarios. The roadside signal control equipment in the A3-level traffic system has high dynamic regulation and collaborative control capabilities and can perform cross-system linkage control, but there are still time delays and accuracy deviations in the optimization results. The cross-system linkage control includes dynamic optimization of bus routes and online dispatching of taxis; The cloud-based system in the A3-level transportation system can provide static and dynamic information services and can form a global optimized scheduling scheme. The global optimization scheduling scheme involves the interactive iteration of signal control strategy and path planning decision to achieve the optimal result; Traffic entities and functions of a Level A4 transportation system: The vehicles operating in the A4-level transportation system are equipped with high-level autonomous driving capabilities, capable of fully taking over driving tasks and operating according to the globally optimal strategy issued by the cloud, relying on high-speed, low-latency communication. The vehicle-side equipment can achieve full-dimensional, high-precision monitoring and fusion of operating status, fully grasping its own operating conditions; it can model and predict high-precision road network status in real time, and then achieve dynamic optimization by combining delay monitoring; it can predict and proactively resolve potential driving conflicts in multi-vehicle collaborative scenarios. The roadside signal control equipment in the A4-level traffic system has fully automatic dynamic adjustment capabilities, enabling real-time optimization and control of the traffic system and supporting seamless coordination between normal and emergency situations. The transportation system includes a traffic light system, a public transportation system, a taxi system, and a road guidance system; The cloud-based A4-level transportation system can provide real-time, low-latency static and dynamic traffic information, and use artificial intelligence-driven global optimization algorithms to predict upcoming congestion sections, thereby generating real-time optimal scheduling and control schemes.

[0038] Optionally, the information parsing process of the intelligent vehicle flow vehicle-road-cloud integrated system in step S3 is a phased model, which includes: Perception phase: Vehicle-side and roadside equipment filter noise and extract key feature information from multi-source heterogeneous data to generate traffic perception information, which includes traffic event information, traffic participant information, traffic status information and traffic signal information; Transmission phase: Natural language processing and knowledge graph technologies are used to transform the traffic perception information into a semantically standardized expression, thereby eliminating semantic differences between different traffic subjects; Decision-making stage: The semantically standardized information is deeply mined and analyzed by the edge computing unit, and personalized decision response results for each traffic subject are generated in collaboration with the traffic management cloud platform according to the actual needs of the multi-level scene scale model. Control phase: The semantics of the decision response results are reconstructed into control commands that are adapted to different intelligent entities, and precise control is achieved through vehicle-side or road-side execution units.

[0039] Optionally, each instantiation unit in step S4 includes: a. Instantiate the unit name, number, and description; b. Corresponding instantiation process, target deployment entity, and the corresponding level of autonomy of the transportation system; c. Core configuration parameters, parameter descriptions, generation tools, and output formats; d. Traffic service scenario of instantiated unit.

[0040] Optionally, the traffic service scenarios include vehicle cooperative lane changing scenarios, emergency vehicle priority passage scenarios, and cross-road segment route planning scenarios.

[0041] Based on the above embodiments and in combination Figures 2 to 5 A preferred embodiment of the present invention is described in further detail as follows: A method for constructing an interactive information model of intelligent transportation subjects based on a three-dimensional digital model, comprising the following steps: S100: Constructing a dynamic and static information attribute system for traffic entities: Based on a three-dimensional digital model, various types of information involved in the intelligent transportation subject are systematically divided and classified, forming two major categories: dynamic attributes and static attributes. Dynamic attributes characterize the real-time changes of the transportation subject during operation, while static attributes describe the inherent characteristics of the transportation subject that remain stable over a longer timescale. Together, they constitute the data foundation of the transportation subject interaction information model, specifically including the following steps: S1001: Classify and define information attributes: Divide traffic subject information into static attributes and dynamic attributes, and clarify the specific content of each type of attribute: The static attribute refers to an attribute that remains unchanged within a certain time range, including: Road structural attributes: number of lanes, lane width, curvature, slope, intersection geometry, spatial location of traffic signs, and traffic light layout; Traffic facility attributes: Roadside unit (RSU) deployment location and coverage area, camera installation location and viewing angle, radar detection area, communication base station distribution and communication protocol support; Vehicle inherent attributes: vehicle dimensions, wheelbase, curb weight, maximum load, communication module type and version, autonomous driving level (L0 to L5) and sensor configuration list; Inherent environmental attributes: road paving materials, height and weight limits for bridges and tunnels, climate zones, locations of historically frequent accidents, and the location of fixed obstacles; The dynamic attributes refer to attributes that change with time and traffic conditions, including: Vehicle dynamic attributes: real-time latitude and longitude coordinates, speed, acceleration, heading angle, lateral and longitudinal control commands, communication link status, perception data confidence level, and energy system status; Traffic flow dynamics: lane flow, time-averaged speed, spatial density, queue length, average intersection delay, number and location of conflict points; Event dynamic attributes: traffic event type, event impact range, estimated duration, emergency response level, and evacuation route planning; Dynamic environmental attributes: real-time weather conditions, visibility, road surface adhesion coefficient, light intensity, and wind force and direction.

[0042] S1002: Establish the mapping relationship between the dynamic and static attributes and multi-level scenes, that is, systematically map the dynamic and static attributes to lane-level, intersection-level, and road network-level three-dimensional scenes to form a structured hierarchical information model: Lane-level scene mapping: Static attributes: lane line geometric coordinates, lane function type, speed limit sign value, and physical boundary constraints; Dynamic attributes: vehicle trajectory sequence, time and space distance between vehicles, lane change decision markers, and local event status; Intersection-level scene mapping: Static attributes: spatial coordinates of traffic light groups, stop line positions, functional division of turning lanes, and layout of pedestrian crossing facilities; Dynamic attributes: current phase and remaining time of traffic lights, vehicle queue detection data, real-time occupancy status of conflict points, and priority passage request signals; Road network-level scene mapping: Static attributes: road network topology connections, regional functional attributes, and infrastructure distribution map; Dynamic attributes: regional traffic state index, route selection behavior characteristics, scope of event propagation impact, and parameters of coordination and control strategies.

[0043] S1003: Constructing a standardized attribute system: The dynamic and static attributes are standardized by defining data format, unit, precision, update frequency and communication protocol to form a unified information description specification, which supports the fusion and interaction of multi-source heterogeneous data.

[0044] S200: Constructing a multi-stage collaborative information model for intelligent transportation entities: Based on a three-dimensional digital architecture, the system constructs a multi-stage information model covering the entire process of "perception-transmission-decision-control" for traffic entities with different intelligence levels at the vehicle, road, and cloud levels. This enables multi-granularity, cross-level, and semantically unified collaborative information expression and interaction, specifically including the following steps: S2001: Information Model Construction in the Perception Phase The perception stage information model aims to collect, fuse, and extract features from multimodal, multi-source heterogeneous data on traffic environment and subject status, and form standardized perception output to provide accurate and reliable input for subsequent stages. Its construction includes the following: Sensing content and data sources: The sensing content includes vehicle status (such as position, speed and heading angle), traffic flow parameters (such as flow rate, density and speed), event information (such as accidents and congestion) and environmental information (such as weather, lighting and road conditions). The data sources include on-board sensors (such as cameras, radar and LiDAR), roadside equipment (such as detectors, cameras and RSUs) and cloud systems (such as traffic management platforms and meteorological information systems). Multi-level perception modeling method: Lane-level scenarios: Focusing on vehicle behavior and micro-interactions, a high-precision local environment perception model with the vehicle as the core is constructed to support the control requirements of following, lane changing and obstacle avoidance. Intersection-level scenario: Integrate multi-directional traffic flow, signal status, pedestrian and non-motorized vehicle information to build a conflict detection and collaborative perception model; Road network level scenario: Integrate multi-source macro data to build traffic situation awareness and event propagation models, supporting regional-level monitoring and dispatching; Feature extraction and fusion methods: Deep learning, filtering algorithms and multi-sensor fusion technology are used to denoise, align and extract features from the raw perception data to generate perception results in a unified format, such as target lists, traffic event labels and environmental status assessments; Output and Interface: The perception results are output in a standardized format and an API interface is provided for the transmission stage to ensure data consistency and interactivity.

[0045] S2002: Information Model Construction in the Transmission Phase The information model in the transmission phase is responsible for achieving efficient, reliable, and low-latency transmission of perceived data and control commands, and for eliminating differences in information understanding between different intelligent agents through semantic standardization. Its construction includes the following: Communication protocols and mechanisms: Supports multiple communication methods such as V2V, V2I, V2N and cloud-edge-device, and adopts multiple protocols such as C-V2X, 5G, MQTT and DDS to ensure the real-time performance and reliability of data in complex environments; Semantic standardization and knowledge graph construction: Utilizing natural language processing (NLP) and knowledge graph technologies, multimodal perception data is transformed into a unified semantic expression, establishing relationships between traffic entities, events, and states to form a machine-understandable semantic network that supports barrier-free interaction between devices of different intelligence levels; Multi-level transmission modeling method: Lane-level scenarios: Focus on ensuring low-latency communication between vehicles and between vehicles and the road, and support rapid interaction of collaborative control commands; Intersection-level scenarios: Enable efficient uploading and distribution of structured data such as signal status and priority requests; Road network level scenario: Supports cloud-edge synchronization and cross-domain transmission of massive data to meet global optimization needs; Data security and privacy protection mechanisms: Encrypted transmission, identity authentication, and data anonymization technologies are employed to ensure the security and compliance of information during transmission.

[0046] S2003: Information Model Construction in the Decision-Making and Control Phase The decision-making and control stage information model is based on information provided by the perception and transmission stages. It generates control commands through a collaborative decision-making algorithm and implements them to specific execution units, forming a closed-loop control. Its construction includes the following: Decision-making mechanism and algorithm: Based on the intelligence level (A1-A4) of traffic subjects and the needs of the scenario, rule engine, optimization algorithm and reinforcement learning methods are used to realize multi-level decision-making function from local behavior decision to global collaborative scheduling; Lane-level scenarios: The decision-making focus is on micro-behavioral planning such as following, lane changing, and obstacle avoidance; Intersection-level scenarios: Decision-making includes signal timing optimization, conflict resolution, and bus priority. Road network level scenario: The decision-making objective is macro-level optimization, such as route guidance, regional coordination, and emergency dispatch; Control command generation and semantic reconstruction: The decision results are converted into control commands that conform to the device's cognitive level through the semantic reconstruction module, such as speed suggestions, traffic light control commands and path adjustment instructions, to ensure that the commands are understandable and executable; Control execution and feedback mechanism: Control commands are executed through vehicle-mounted actuators, roadside equipment (such as traffic signals and information boards) or cloud-based dispatching systems. At the same time, the execution effect is collected in real time through sensors and communication systems to form a closed-loop feedback and support dynamic adjustment and optimization of decisions. Cross-level collaborative control method: Establish a hierarchical control architecture of "cloud-road-vehicle". The cloud is responsible for global policy generation, the road is responsible for regional coordination, and the vehicle is responsible for local execution. Through information sharing and command coordination, the overall system efficiency is improved.

[0047] S300: Constructing Information Object Interaction Diagrams in Multi-Level Scenarios: This section uses Unified Modeling Language (UML) sequence diagrams to visualize the dynamic interaction relationships and information flow processes between various information objects and functional objects in lane-level, intersection-level, and road network-level collaborative scenarios. This clarifies the temporal logic and collaborative mechanisms of the system during operation, providing a basis for system implementation and verification. The specific implementation method is as follows: S3001: Interactive modeling of vehicle-following cooperative control scenarios based on lane-scale information models: like Figure 3As shown, this scenario establishes a collaborative interaction model between intelligent connected vehicles, roadside units, and edge computing units for micro-vehicle following behavior, ensuring that vehicles improve road traffic efficiency while maintaining a safe following distance; Intelligent connected vehicles collect their own status information at a period of 100ms based on the vehicle dynamic attributes (such as position, speed, acceleration and heading angle) defined by S1001, and upload the status information to the roadside unit through V2I communication. After receiving the status information of multiple vehicles, the roadside unit performs data fusion based on the traffic flow dynamic attributes in S1001 to generate traffic flow information including the status set of queued vehicles, average headway, and lane occupancy rate, and transmits it to the edge computing unit. The edge computing unit uses the car-following control algorithm, combined with the decision logic defined in S2003, to calculate the optimized control command information (such as target acceleration, desired velocity, and desired distance). Control commands are sent to the vehicle via the RSU, and the vehicle control system executes longitudinal control operations to achieve precise adjustment of the throttle and brakes; The vehicle sends control confirmation information to the RSU, completing the information interaction loop and providing execution status feedback to the system. In this interaction model, the intelligent connected vehicle (edge ​​side) is responsible for high-frequency perception and command execution; The roadside unit (edge ​​access layer) is responsible for low-latency communication and data aggregation of the local network; The edge computing unit (edge ​​computing layer) is responsible for running computationally intensive algorithms and generating cooperative strategies to maintain a stable and safe desired spacing within the lane, suppress the propagation of disturbances in the traffic flow, and thus improve single-lane traffic efficiency and driving safety.

[0048] S3002: Interactive Modeling of Bus Priority Cooperative Control Scenario Based on Intersection-Level Scale Information Model: like Figure 4 As shown, this scenario addresses the priority passage requirement for buses at signalized intersections, establishing an information interaction and collaborative decision-making model among intelligent buses, roadside units, signal controllers, edge computing units, and a traffic management cloud platform; Based on the event dynamic attributes defined by S1001, intelligent buses send priority request information to the RSU, which includes key decision parameters such as vehicle ID, route number, real-time passenger capacity and current delay. Based on the intersection-level dynamic and static attributes in S1001, RSU integrates multi-source information such as traffic flow, queue length and real-time signal phase to form a complete understanding of the intersection's operating status. After the edge computing unit generates a local priority plan, it applies for intersection coordination permission from the traffic management cloud platform to ensure that local optimization does not violate the road network-level static attribute constraints defined in S1001. Once approved by the traffic management cloud platform, the edge computing unit writes the signal control parameters into the signal controller, and at the same time, the RSU sends traffic guidance information to the buses. This interactive modeling scheme can realize a three-level collaborative architecture of "end-edge-cloud": the bus (end) initiates the request; the edge node (edge) makes a local quick decision; and the traffic management cloud platform (cloud) is responsible for global optimization decision-making, which can reflect the collaborative mechanism between traffic entities with different intelligence levels. This interactive modeling scheme can ultimately achieve intersection-level vehicle-road cooperative functions: proactively prioritizing bus signals based on real-time demand, thereby reducing bus travel time and parking delays, improving the punctuality and reliability of public transportation, and ultimately optimizing the overall traffic efficiency of the intersection.

[0049] S3003: Interactive Modeling of Road Network Cooperative Signal Control Scenarios Based on Road Network-Level Information Model: like Figure 5 As shown, this scenario takes the collaborative control of multiple intersections in an urban road network as an example to demonstrate the large-scale, multi-layered coordination and interaction mechanism between the traffic management cloud platform, edge computing unit, roadside unit, signal controller and intelligent connected vehicles under the A4 level system; Each edge computing unit collects the road network-level dynamic attributes (such as average vehicle speed in sub-areas, link travel time, and saturation of key intersections) defined in S1001 every 300 seconds and uploads them to the traffic management cloud platform. The traffic management cloud platform integrates information from across the region, runs regional coordination and optimization algorithms, and generates coordinated control strategies for the entire region, including parameters such as public cycle, green ratio, and phase difference. The optimization strategy is distributed to each edge computing unit, decomposed into executable signal timing parameters, configured to the signal controller, and at the same time, traffic guidance strategies are generated and published through the roadside unit. Connected vehicles perform path planning and speed adjustment based on the received global strategy, and feed back vehicle trajectory and status information to the system to evaluate the control effect and optimize the strategy for the next cycle. This interactive modeling solution enables a collaborative optimization architecture of "end-edge-cloud": the traffic management cloud platform performs centralized global optimization calculations; the edge computing unit is responsible for distributed execution of control strategies; and connected vehicles and roadside units achieve accurate information dissemination and collection, reflecting the logical relationships between devices with different levels of intelligence. This interactive modeling solution ultimately achieves road network-level vehicle-road-cloud collaborative functions, thereby optimizing regional road network traffic flow, forming efficient green wave traffic belts, reducing the overall average travel time and number of stops of the road network, and maximizing the traffic efficiency of the regional road network.

[0050] This invention employs a comprehensive information model construction method that combines multi-level, hierarchical, and multi-stage approaches. It organically integrates spatial scale, intelligence level, and information flow, ultimately forming a complete collaborative information model construction method to guide the design and analysis of integrated vehicle-road-cloud systems. This method generates collaborative information models for system design, simulation, and testing. This invention addresses the technical problems in existing technologies, such as difficulties in cross-domain collaboration, inconsistent information models, and insufficient global optimization caused by uneven intelligence levels among traffic entities. It also provides methodological support for the design, verification, and implementation of integrated vehicle-road-cloud systems.

[0051] The above-described technical solutions are merely preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for constructing an interaction information model of heterogeneous traffic subjects based on a three-dimensional digital model, characterized in that, Comprise the following steps: S1: multi-level scene and its function design: based on the spatial hierarchy of road traffic system, the multi-level scene scale information model including lane level, intersection level and road network level is constructed, and the function characteristics of each level scale information model, the logical connection and interaction between each other are described; S2: analysis of heterogeneous intelligent multi-granularity information: the autonomy level of traffic system A1 to A4 is defined, and the perception ability, decision logic and execution accuracy of the vehicle-end-road-end-cloud equipment under each autonomy level are described; S3: multi-stage migration evolution information analysis: according to the information analysis process of different attributes of traffic entities and vehicle-road-cloud integrated system of heterogeneous intelligent traffic flow, the four-stage traffic information flow process including perception, transmission, decision and control is designed, and the standardized processing and interaction logic of traffic information between devices with different intelligent levels are described; S4: information model instantiation and deployment: the information model defined by standardization is converted into several instantiation units, that is, specific data structure, interface protocol and configuration file, and is deployed to the computing entities of various traffic subjects in end-edge-cloud, which is used to support the actual operation of traffic service scene. 2.The method of claim 1, wherein, The step S1 comprises the following steps: S11: constructing the multi-level scene scale information model, comprising: Lane level scale information model: taking a single road section or a single lane as the time and space range, considering the information interaction between heterogeneous vehicles and vehicles and roadside devices, realizing fine control of vehicles at the micro level; Intersection level scale information model: taking one or more intersections as the time and space range, considering the information interaction among heterogeneous vehicles, vehicles and roadside sensing devices, intersection control devices, realizing multi-road section traffic connection and conflict management; Road network level scale information model: taking the whole road network of city or region as the time and space range, considering the information aggregation of heterogeneous vehicles, the collaborative calculation of roadside edge computing units and the collaborative decision of cloud decision system, realizing global traffic optimization and scheduling across road sections and regions; S12: describe the function and relationship of the multi-level scene scale information model: The function of the lane level scale information model: describing the vehicle-vehicle and vehicle-road cooperation function of vehicles at the micro local level at the lane level scale, including following control, lane changing cooperation, local obstacle avoidance and lane traffic flow stability control; The function of the intersection level scale information model: describing the vehicle-road cooperation and conflict resolution function in the range of one or more intersections at the intersection level scale, including signal control optimization, intersection passing strategy generation and emergency vehicle priority passing; The function of the road network level scale information model: describing the global optimization and scheduling function of intelligent connected vehicle system at the macro level at the road network level scale, including road network level traffic flow prediction, cross-road section path planning, regional traffic balance control and multi-source information fusion prediction; Interrelation between the hierarchical scene functions: describe the upper and lower interaction and coordination relationship between the functions of the lane-level scale information model, the functions of the intersection-level scale information model and the functions of the road network-level scale information model; under global scheduling, the functions of the intersection-level scale information model interact with the functions of the road network-level scale information model as a macro coordination unit; under local decision-making, the functions of the lane-level scale information model coordinate with the functions of the intersection-level scale information model. 3.The method of claim 1, wherein, The step S2 comprises the following steps: S21: defining the autonomy levels A1 to A4: A1 level: the vehicle is mainly manually driven, the roadside device operates according to a fixed scheme, the cloud provides static information services, and humans highly or completely intervene in each specific link; A2 level: the vehicle has partial automatic driving function and needs manual supervision, the roadside has medium dynamic scheduling capability, the cloud can realize single subsystem optimization, and humans only intervene in each specific link when necessary; A3 level: the vehicle has conditional automatic driving capability, the roadside has cross-system collaborative control capability, the cloud can form a global optimization scheduling scheme, and humans observe globally and no longer intervene in specific links; A4 level: the vehicle realizes high or complete automatic driving, the roadside realizes full-automatic dynamic adjustment, the cloud uses artificial intelligence algorithms for real-time global optimization and prediction, and humans completely do not intervene in any link; S22: according to the different intelligent levels of the traffic subjects in the traffic system, describe the traffic entities involved in the autonomous levels A1 to A4 traffic system and their functions: Traffic entities and their functions of the A1 level traffic system: The running vehicles in the A1 level traffic system are mainly manually driven, the perception and decision-making tasks mainly rely on the real-time reaction of the driver, the basic state information including vehicle speed and position is obtained through basic sensors, and is uploaded to the cloud through V2X technology to realize information interaction and vehicle state perception, can receive simple instructions issued by the roadside and the cloud and has limited response capability, and the decision is still mainly made by humans; The road side signal control device in the A1 level traffic system operates according to a fixed cycle or a preset scheme, and cannot realize real-time response to the traffic state; the traffic monitoring device is less, and the roadside perception capability is weak; The fixed cycle includes a fixed signal cycle, and the preset scheme includes a fixed bus departure time; The cloud in the A1 level traffic system only provides static information services and does not have the ability of real-time data analysis and dynamic scheduling; The static information services include the publication of fixed routes and time schedules; Traffic entities and their functions of the A2 level traffic system: The running vehicles in the A2 level traffic system are mainly low-level automatic driving, the vehicle can handle most of the perception and control tasks, but the key decision still needs human participation, the vehicle speed, position and running mode are obtained through basic sensors, the road channelization information and vehicle-road relationship information are obtained through radar, and then the information interaction is realized by uploading to the cloud through V2X technology, can accept the instructions issued from the roadside and the cloud and make responses, but still needs manual supervision; The roadside device in the A2-level traffic system has a medium-level dynamic scheduling capability. The roadside device obtains real-time traffic data through various sensors, but the amount of information obtained and interacted is small, the interaction is slow, and manual intervention is still required in critical situations. The roadside can obtain real-time data including traffic flow and traffic events through various sensors, but the amount of information obtained and interacted is small, the interaction is slow, and manual intervention is still required in critical situations. The medium-level dynamic scheduling capability includes dynamic departure of buses and pre-booking scheduling of taxis. The cloud in the A2-level traffic system can provide static and partial dynamic information services to support system dynamic scheduling, but the scheduling capability is limited to a single subsystem, lacking cross-system and cross-level collaborative scheduling capability. The single subsystem includes optimization for bus departure scheduling or signal control only. Traffic entities and their functions in the A3-level traffic system: The running vehicles in the A3-level traffic system have high-level automatic driving functions, strong environmental perception and autonomous decision-making capabilities, and can maintain real-time interaction with the roadside and the cloud. They can perceive and deeply integrate their own running state and environmental factors including external weather, lighting, and obstacles, and then assess real-time road segment delays and dynamically identify and avoid potential driving conflicts in multi-vehicle collaborative scenarios. The roadside signal control device in the A3-level traffic system has high dynamic regulation and collaborative control capability, and can control across systems, but there are still time delays and precision deviations in the optimization results. The cross-system linkage control includes dynamic optimization of bus routes and online scheduling of taxis. The cloud in the A3-level traffic system can provide static and dynamic information services and form a global optimization scheduling scheme. The global optimization scheduling scheme includes interactive iteration of signal control strategies and path planning decisions to achieve optimality. Traffic entities and their functions in the A4-level traffic system: The running vehicles in the A4-level traffic system have high-level automatic driving functions and can fully take over driving tasks. They rely on high-speed low-latency communication and operate according to the global optimal strategy issued by the cloud. The vehicle-side device can achieve full-dimensional, high-precision running state monitoring and integration, fully mastering its own running situation. It can model and predict high-precision road network state in real time, and then realize dynamic optimization combined with delay monitoring. It can predict and actively resolve potential driving conflicts in multi-vehicle collaborative scenarios. The roadside signal control device in the A4-level traffic system has full-automatic dynamic regulation capability and can perform real-time optimization control of the traffic system, supporting seamless collaboration in normal and emergency situations. The traffic system includes signal light systems, bus systems, taxi systems, and road guidance systems. The cloud in the A4-level traffic system can provide real-time low-latency static information and dynamic traffic information, and use artificial intelligence to drive global optimization algorithms to predict upcoming congested road segments, and then generate real-time optimal scheduling and control schemes. 4.The method of claim 1, wherein, The information analysis process of the vehicle-road cloud integrated system of the heterogeneous traffic flow in step S3 is a phased model, which includes: The perception stage: the vehicle-side and road-side devices generate traffic perception information including traffic event information, traffic participant information, traffic state information and traffic signal information after filtering noise and extracting key feature information from multi-source heterogeneous data; The transmission stage: the traffic perception information is converted into semantic standardized information through natural language processing technology and knowledge graph technology, so as to eliminate the semantic differences between different intelligent traffic subjects; The decision stage: the semantic standardized information is deeply mined and analyzed by the edge computing unit, and individualized decision response results for each traffic subject are generated according to the actual needs of the multi-level scene scale model in combination with the traffic management cloud platform; The control stage: the decision response results are reconstructed into control instructions adapted to different intelligent subjects, and precise control is realized through the vehicle-side or road-side execution unit. 5.The method of claim 1, wherein, Each instantiation unit in step S4 includes: a. instantiation unit name, number and description; b. corresponding instantiation link, target deployment subject and adapted traffic system autonomy level; c. core configuration parameters, parameter description, generation tool and output form; d. traffic service scene of the instantiation unit.

6. The method of claim 1 or 5, wherein, The traffic service scene includes vehicle cooperative lane changing scene, emergency vehicle priority passing scene and cross-section path planning scene.

Images