Intelligent driving environment abnormal state diagnosis method and system based on vehicle cloud cooperation

Abstract

The invention discloses an intelligent driving environment abnormal state diagnosis method and system based on vehicle cloud collaboration, and the method comprises the steps: receiving the full-sample trajectory and environment data of a target intelligent driving vehicle and the surrounding environment thereof, and building a full-sample trajectory and environment data platform; based on trajectory data and environment data related to the target intelligent driving vehicle in the full sample trajectory and environment data platform, calculating an environment risk index of a space-time position where the target intelligent driving vehicle is located; according to the numerical value of the environmental risk index, dividing the environmental risk level of the spatial-temporal position of the target intelligent driving vehicle; based on the environmental risk index and the environmental risk level, judging whether the current state is abnormal; if yes, early warning information is generated and issued to the corresponding intelligent driving vehicle. According to the invention, the vehicle cloud cooperation advantage is utilized, and through cloud full-sample data integration and analysis, accurate quantification, grade division and abnormity diagnosis of driving environment risks are realized, and differential early warning is provided.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent driving, and in particular to an intelligent driving driving environment abnormal state diagnosis method and system based on vehicle-cloud cooperation. BACKGROUND

[0002] With the rapid development of environmental perception, vehicle-mounted computing and vehicle-cloud communication technologies, intelligent driving vehicles have been able to obtain real-time driving environment information around themselves through vehicle-mounted sensors such as cameras, millimeter wave radars and laser radars, and upload part of the driving data to the cloud for centralized processing. Under the vehicle-cloud cooperation architecture, the cloud can converge the trajectory and environmental data of multiple intelligent driving vehicles and road infrastructure, so as to analyze the road traffic operation state in a larger spatial range and a longer time scale. If such full-sample data can be fully utilized to assess the risk of the driving environment of intelligent driving vehicles and diagnose abnormal states, the driving safety in complex traffic scenarios will be effectively improved.

[0003] However, the prior art still has many deficiencies: the driving safety risk assessment is mostly based on the single vehicle perspective, only uses local safety indicators at the vehicle end to judge the collision risk, and can only reflect the short-term and local safety conditions, and it is difficult to depict the spatio-temporal evolution characteristics of the overall road operation state and environmental risk; the cloud traffic monitoring or road condition service focuses on macro indicators such as average speed and congestion index, and the safety risk of the current environment of the intelligent driving vehicle is roughly depicted and lacks pertinence; some abnormal traffic state recognition methods rely on a single or a small number of statistical indicators, and do not systematically consider the comprehensive influence of road geometric structure, multi-vehicle dynamic interaction and environmental elements, and lack a complete technical chain from risk quantification, grade division to individualized early warning.

[0004] Therefore, the present application provides an intelligent driving driving environment abnormal state diagnosis scheme based on vehicle-cloud cooperation, which establishes a standardized full-sample data platform in the cloud, comprehensively considers static and dynamic traffic elements, realizes accurate risk quantification, scientific grade division and efficient abnormal diagnosis, and provides differentiated early warning support for intelligent driving vehicles. SUMMARY

[0005] To solve the above problems, the present application provides an intelligent driving driving environment abnormal state diagnosis method and system based on vehicle-cloud cooperation, which aims to realize accurate risk quantification, grade division and abnormal diagnosis of the driving environment by cloud full-sample data fusion analysis, comprehensively consider static road constraints and multi-vehicle dynamic interaction, issue differentiated early warning to the target vehicle, and improve the intelligent driving safety and environmental adaptability.

[0006] To solve the problems raised in the above background art and achieve the above technical purposes, the present application realizes the following technical solutions:

[0007] The first aspect of the application provides a vehicle-cloud collaborative intelligent driving environment abnormal state diagnosis method, comprising the following steps:

[0008] S1. Receiving full-sample trajectory and environment data of a target intelligent driving vehicle and its surrounding environment, and establishing a full-sample trajectory and environment data platform;

[0009] S2. Based on the trajectory data and environment data related to the target intelligent driving vehicle in the full-sample trajectory and environment data platform, calculating the environment risk index of the space-time position of the target intelligent driving vehicle;

[0010] S3. According to the value of the environment risk index, dividing the environment risk level of the space-time position of the target intelligent driving vehicle;

[0011] S4. Based on the environment risk index and the environment risk level, judging whether the current driving environment is in an abnormal state; if yes, generating corresponding warning information and delivering it to the corresponding intelligent driving vehicle.

[0012] Further, the full-sample trajectory and environment data are formed by time synchronization and coordinate alignment of the driving state information and road environment information of the target intelligent driving vehicle and vehicles on the same section. The driving state information includes vehicle position, speed, acceleration, heading angle, vehicle geometric size, etc.; the road environment information includes lane line type and geometric parameters, lane width, road boundary position, etc. static information, and dynamic information such as relative motion state of surrounding vehicles.

[0013] Further, the step S1 includes: performing latitude and longitude coordinate conversion on the full-sample trajectory and environment data, projecting the vehicle trajectory to the high-precision map coordinate system, performing road matching and lane matching on the projected trajectory, aligning to the road and lane topological structure of the high-precision map, and combining the corresponding road attribute (such as road number, lane number) and traffic facility information to construct a standardized full-sample trajectory and environment data platform containing fields such as spatial coordinates, time stamp and environment attribute.

[0014] Further, the step S2 includes: based on the target intelligent driving vehicle and its surrounding vehicle data in the full-sample trajectory and environment data platform, extracting static road elements (lane lines, road boundaries, speed limits, etc.) and dynamic traffic elements (relative position, speed, acceleration, etc.), calculating static risk index and dynamic risk index, and fusing to generate intelligent driving environment risk index.

[0015] Further, the static risk index risk value calculation method is as follows:

[0016]

[0017] In the formula, a sum of potential energy risk values generated by lane lines in the driving scene at point P (xj, yj), a lane marking weight (the stronger the lane marking constraint, the larger the value), a lane marking lateral coordinate, a lateral coordinate of a point in the scene, an exponential decay coefficient (controls the speed of risk decay with distance); the static risk index reflects the constraint of road structure on vehicle lateral deviation and driving space, and the risk value is higher near the lane boundary line and the road outside boundary.

[0018] Further, the dynamic risk index risk value is calculated as follows:

[0019]

[0020] wherein G is a total control constant (controls the overall numerical range of the risk field), a speed decay control coefficient, a road correction coefficient (related to road conditions), a virtual mass of the object (affected by factors such as object size, type, speed, etc.), d is the distance from the external point to the rectangular frame, and λ is a speed correction coefficient (constructed based on longitudinal and lateral time to collision and other interaction indicators). The dynamic risk index is obtained by superimposing the obstacle interaction potential fields of multiple surrounding vehicles, reflecting the dynamic interaction risk of multiple vehicles.

[0021] Further, the step S3 comprises: in the offline stage or system initialization stage, selecting representative driving scenes (such as urban roads, highways, congested road sections, accident-prone road sections, etc.) from the full-sample data platform, collecting environmental risk index samples to form an environmental risk sample set; performing clustering analysis on the environmental risk sample set using a clustering method (such as a K-means clustering method combining genetic algorithm and simulated annealing optimization of initial clustering center) to divide the samples into several risk clusters; determining the corresponding environmental risk level and risk value interval according to the cluster center and numerical distribution of each risk cluster, and establishing a mapping relationship table between the environmental risk index and the risk level; in online operation, the current environmental risk index is converted into the corresponding environmental risk level according to the mapping relationship.

[0022] Further, the environmental risk level is divided into at least low, medium and high levels, and each level is configured with corresponding diagnostic meaning and warning level; wherein the low level corresponds to a normal or slight risk state, no special warning is needed, and the vehicle can drive according to the normal strategy; the medium level corresponds to a risk state that needs attention, a prompt warning needs to be issued to the vehicle to remind the driver or the vehicle-mounted system to pay attention to environmental changes; the high level corresponds to a high risk or dangerous state that needs to take intervention measures, a warning warning or emergency warning needs to be issued, and the vehicle is suggested to slow down, avoid or stop driving.

[0023] Further, the step S4 comprises: in each diagnosis cycle, judging whether at least one of the following abnormal conditions is met based on the environment risk index and the environment risk level of the target intelligent driving vehicle at the current time: one of the environment risk index or the environment risk level exceeds a preset threshold; two of continuously staying at a high risk level for multiple cycles; when any condition is met, generating differentiated early warning information (including risk level, abnormal type, recommended driving behavior, and abnormal impact range) matched with the abnormal type (such as lane deviation risk, collision risk, road structure risk, and multi-vehicle interaction risk) and severity, and delivering the same to the intelligent driving vehicle causing the abnormal state or affected by the abnormal state.

[0024] The second aspect of the application also provides an intelligent driving driving environment abnormal state diagnosis system based on vehicle-cloud cooperation, comprising a vehicle-end system and a cloud-end system, wherein the vehicle-end system and the cloud-end system are connected through a vehicle-cloud communication network.

[0025] The vehicle-end system comprises:

[0026] A data acquisition module is configured to acquire target intelligent driving vehicle and its driving environment data, including obtaining driving state information of the target vehicle through vehicle-mounted sensors (cameras, millimeter wave radars, laser radars, GPS, etc.), and obtaining driving state information of other vehicles on the same road section and road environment information through vehicle-to-vehicle communication and vehicle-to-road communication;

[0027] A data preprocessing module is configured to perform time synchronization and coordinate alignment on the collected data, eliminate the spatio-temporal deviation of multi-source data, and ensure data consistency;

[0028] A data transmission module is configured to upload the preprocessed data to the cloud-end system, and receive early warning information and control instructions issued by the cloud-end system.

[0029] The cloud-end system comprises:

[0030] A data receiving module is configured to receive data uploaded by multiple vehicle-end systems;

[0031] A data processing and database building module is configured to perform coordinate conversion, road matching, lane matching, and other processing on the received data, combine road attributes and traffic facility information, and build a standardized full-sample trajectory and environment data platform;

[0032] A risk index calculation module is configured to extract static road elements and dynamic traffic elements based on the full-sample trajectory and environment data platform, calculate static risk indexes, dynamic risk indexes, and fused environment risk indexes;

[0033] a risk grade division module, configured to convert the current environmental risk index into a corresponding environmental risk grade based on the mapping relationship table established offline;

[0034] an abnormality diagnosis module, configured to determine whether the current driving environment is in an abnormal state according to the environmental risk index and the environmental risk grade in each diagnosis cycle;

[0035] a warning module, configured to generate differentiated warning information when the abnormal state is determined, and deliver the information to the corresponding vehicle-side system through the vehicle-cloud communication network.

[0036] The present application has the advantages that: the present application utilizes the vehicle-cloud collaborative advantage, integrates and analyzes the cloud-end full-sample data, constructs intelligent driving driving environment risk indexes from the aspects of road structure and multi-vehicle dynamic interaction, and realizes the fine representation of the driving environment safety level through the data-driven risk grade division; on this basis, the abnormal state diagnosis mechanism based on threshold overrun and high risk persistence is introduced, and differentiated warning information is delivered to the intelligent driving vehicle that leads to or is affected by the abnormality; the precise quantization, grade division and abnormality diagnosis of the driving environment risk are realized, differentiated warning is provided for the intelligent driving vehicle, the driving safety in the complex traffic scene is effectively improved, and the present application has a wide application prospect. BRIEF DESCRIPTION OF DRAWINGS

[0037] Figure 1 FIG. 1 is a flow chart of the intelligent driving driving environment abnormal state diagnosis method based on vehicle-cloud collaboration of the present application. DETAILED DESCRIPTION

[0038] In order for those skilled in the art to better understand the technical solutions of the present application, the technical solutions of the present application will be described clearly and completely below in conjunction with the drawings in the embodiments of the present application. It should be understood that the specific embodiments described herein are only used to explain the present application, and are not used to limit the present application.

[0039] As shown in FIG. 1, the intelligent driving driving environment abnormal state diagnosis method based on vehicle-cloud collaboration provided by the present application embodiment includes the following steps: Figure 1

[0040] S1: data acquisition and cloud-end database construction

[0041] ​The data acquisition module of the vehicle terminal system obtains the driving state information of the target intelligent driving vehicle through the vehicle-mounted sensors, including position (latitude and longitude), speed, acceleration, heading angle, and vehicle geometric size, etc. Meanwhile, the driving state information of other vehicles on the same road section and the road environment information (such as lane line type, lane width, road boundary position, etc.) are obtained through vehicle-to-vehicle communication and vehicle-to-road communication. The data preprocessing module performs time synchronization (alignment based on timestamp) and coordinate alignment (eliminate the installation deviation of different sensors) on the above-mentioned multi-source data, ensuring the spatio-temporal consistency of the data, and forming standardized full-sample trajectory and environment data. The data transmission module uploads the preprocessed data to the cloud system through the vehicle-to-cloud communication network.

[0042] After the data receiving module of the cloud system receives the full-sample trajectory and environment data uploaded by multiple vehicles, the data processing and database building module performs the following operations: converting the latitude and longitude information in the data into a high-precision map coordinate system, and projecting the vehicle trajectory onto the high-precision map coordinate system to realize accurate matching of the trajectory and the map; performing road matching and lane matching on the projected trajectory, aligning the vehicle trajectory to specific roads and lanes based on the road and lane topology structure of the high-precision map; storing the processed trajectory data together with road attributes (road number, lane number), traffic facility information, etc., and building a standardized full-sample trajectory and environment data platform containing fields such as road number, lane number, spatial coordinate, timestamp, and environment attribute.

[0043] S2: Environment risk index calculation

[0044] In this step, based on the trajectory information of the target intelligent driving vehicle and its surrounding vehicles, road lane line information, and road boundary information in the standardized full-sample trajectory and environment data platform, a static risk potential field and a dynamic risk field are constructed in the high-precision map coordinate system, and the static risk potential field and the dynamic risk field are superimposed in the space region near the target intelligent driving vehicle, obtaining an interaction risk field representing the road structure constraint and the influence of multi-vehicle dynamic interaction. The field value of the interaction risk field in the position of the target intelligent driving vehicle and the predetermined space region in front of and around it is taken as the intelligent driving driving environment risk index.

[0045] The interaction risk field specifically includes a static risk potential field and a dynamic risk field.

[0046] The static risk potential field is composed of static road information such as lane line type and geometric parameters, lane width, lane center line position and road boundary position, and is used to reflect the constraint of road structure on the lateral deviation and driving space of the target intelligent driving vehicle. The static risk potential field is constructed by calculating the lateral distance of any point in the space to each type of lane line and road boundary, and assigning different weights according to different lane line types and boundary types, so that the static risk value near the lane boundary line and the road outside boundary is higher than that in the lane center area.

[0047] The dynamic risk field is composed of dynamic information such as the relative distance, relative speed and relative acceleration of the target intelligent driving vehicle and the surrounding vehicles, and the surrounding vehicles are modeled as a rectangular vehicle body boundary with length and width. The interactive kinetic energy field of single vehicle obstacle is calculated according to the minimum distance of the space point to the rectangular boundary of the surrounding vehicle and the relative motion state in the longitudinal and lateral directions, and the interactive kinetic energy fields of multiple surrounding vehicle obstacles are superimposed to obtain the dynamic risk field.

[0048] Specifically, the risk index calculation module of the cloud system calculates the environmental risk index according to the following steps:

[0049] S21: Read the trajectory data and environmental data of the target vehicle and the surrounding vehicles from the full-sample trajectory and environmental data platform, including trajectory information, road lane line information and road boundary information;

[0050] S22: Coordinate system conversion and road / lane matching are performed in the high-precision map coordinate system;

[0051] S23: Extract static road elements, including lane line type and geometric parameters, lane width, road boundary position, speed limit information, etc.

[0052] S24: Extract dynamic traffic elements, including the relative position, relative speed and relative acceleration of the target vehicle and the surrounding vehicles;

[0053] S25: According to the static road elements, a static risk potential field is constructed and its risk value at each point in the space is calculated, and the risk value is used to represent the static risk index:

[0054] According to the static road elements, the lateral distance of any point in the space to each type of lane line and road boundary is calculated, and the static risk index risk value generated by each lane line is calculated and superimposed by combining the marking weight (different weights are assigned according to the lane line type of dashed line and solid line) and the exponential decay coefficient, to obtain the lane line risk potential field reflecting the constraint degree of the lane line to the vehicle. According to the lateral distance of the target vehicle to the left and right boundaries of the road, the static risk potential field is constructed, so that the static risk value increases when the vehicle approaches the road outside boundary, and the static risk value in the area near the lane boundary and the road outside boundary is higher.

[0055] The static risk indicator risk value is calculated in the following manner:

[0056]

[0057] wherein, represents the sum of the generated potential energy risk values of the lane lines at point P (xj, yj) in the driving scene, represents the marking line weight (the stronger the marking line constraint ability, the larger the value), is the lateral coordinate of the lane marking, is the lateral coordinate of a point in the scene, is an exponential decay coefficient (controls the speed of risk decay with distance); the static risk indicator reflects the constraint of the road structure on the vehicle lateral deviation and the driving space, and the risk value is higher near the lane boundary line and the road outside boundary.

[0058] S26: A dynamic risk field is constructed by superimposing the obstacle interaction dynamic energy fields of multiple surrounding vehicles, and the risk values of each point in the space are calculated, which are used to represent the dynamic risk indicator:

[0059] The vehicles surrounding the target vehicle are modeled as rectangular vehicle body boundaries with length and width in a preset coordinate system, the minimum distance d between any point in the space and the rectangular vehicle body boundary is calculated, representing the geometric proximity of the point to the surrounding vehicle; the relative distance, relative speed and relative acceleration of the target vehicle and the surrounding vehicle in the longitudinal and lateral directions are obtained, the corrected relative speed is calculated, and the directional weight and speed correction coefficient λ are constructed based on the time-to-collision and other interaction indicators in the longitudinal and lateral directions, so that the obstacle interaction dynamic energy field decays more slowly and the risk value is higher in the direction of larger relative speed and smaller time-to-collision; the minimum distance d and the speed correction coefficient λ are used as inputs, the obstacle interaction dynamic energy field corresponding to each surrounding vehicle is calculated by using a decay function that varies with distance and relative motion state, and the obstacle interaction dynamic energy fields of all surrounding vehicles are superimposed to obtain the dynamic risk indicator risk value in the interaction risk field.

[0060] The dynamic risk indicator risk value is calculated in the following manner:

[0061]

[0062] wherein G is a total control constant (controls the overall numerical range of the risk field), is a speed decay control coefficient, is a road correction coefficient (related to road conditions), where m is the virtual mass of the object (affected by the size, type, speed, etc. of the object), d is the distance from the external point to the rectangular frame, and l is a speed correction coefficient (based on longitudinal and lateral time-to-collision, etc. indicators). The dynamic risk indicator is obtained by superimposing the obstacle interaction dynamic energy fields of multiple surrounding vehicles, reflecting the dynamic interaction risk of multiple vehicles.

[0063] S27: Generating an intelligent driving environment risk indicator by fusing static and dynamic risks, superimposing the static risk potential field generated by the road marking line constraint in the spatial region and the dynamic risk field generated by the surrounding objects, and obtaining an interaction risk field to complete the calculation of the intelligent driving environment risk indicator.

[0064] S3: Risk level division

[0065] In the offline stage or system initialization stage, a plurality of representative driving scenes are selected from the full-sample trajectory and environment data platform, and sample values of the environmental risk indicators in the road and the space region in front of and around the target intelligent driving vehicle are collected to form an environmental risk sample set; the environmental risk sample set is clustered by using a clustering method, the samples are divided into a plurality of risk clusters, and the corresponding environmental risk level and risk value interval are determined according to the cluster center and numerical distribution of each risk cluster, thereby forming a mapping relationship between the environmental risk indicator and the environmental risk level; in online operation, the risk indicator of the current driving environment of the target intelligent driving vehicle is converted into the corresponding environmental risk level according to the mapping relationship.

[0066] The intelligent driving driving environment is divided into at least three environmental risk levels, and the corresponding diagnostic meanings and warning levels are configured for each environmental risk level. Among them, the low-level environmental risk corresponds to a normal or slight risk state, the medium-level environmental risk corresponds to a risk state that needs to be paid attention to, and the high-level environmental risk corresponds to a high risk or dangerous state that needs to be taken intervention measures; when the environmental risk indicator of the target intelligent driving vehicle is in different environmental risk level intervals, differential warning information is generated according to the corresponding warning level types.

[0067] Specifically, the risk level division module of the cloud system divides the risk level according to the following steps:

[0068] S31: In the offline stage or system initialization stage, a plurality of representative driving scenes such as straight driving on urban roads, overtaking on expressways, turning at intersections, and following on congested road sections are selected from the full-sample trajectory and environment data platform, and sample values of the environmental risk indicators in different scenes are collected in the road and the space region in front of and around the target intelligent driving vehicle to form an environmental risk sample set;

[0069] S32: Using the K-means clustering method combining genetic algorithm and simulated annealing optimization of initial clustering center, the environmental risk sample set is clustered and analyzed, and the samples are divided into low, medium and high risk clusters (the number of clusters can be adjusted according to actual needs), so as to improve the stability of the clustering results and the ability to distinguish different risk levels;

[0070] S33: According to the cluster center and numerical distribution of each risk cluster, the environmental risk value interval corresponding to each environmental risk level is determined, for example: the risk value interval of low risk level is [0, E1), the risk value interval of medium risk level is [E1, E2), and the risk value interval of high risk level is [E2, +∞), and a mapping relationship table between environmental risk indicators and environmental risk levels is established;

[0071] S34: In online operation, the environmental risk indicators of the target vehicle at the current time and in the front and surrounding warning areas are retrieved from the cloud sample trajectory and environmental data platform;

[0072] S35: The mapping relationship table is searched to determine the risk interval to which the current environmental risk indicator belongs, and the corresponding environmental risk level is output for calling by the abnormal state diagnosis and warning issuing module; thus, the intelligent driving driving environment risk level division is completed.

[0073] S4: Abnormal state diagnosis and warning issuing

[0074] In each diagnosis cycle, based on the environmental risk indicators and environmental risk levels of the target intelligent driving vehicle at the current time, it is judged whether at least one of the following conditions is met:

[0075] First, the current environmental risk indicators or environmental risk levels exceed the preset threshold;

[0076] Second, the environmental risk indicators or environmental risk levels are continuously at a preset high risk level in consecutive multiple diagnosis cycles;

[0077] When any of the above conditions is met, the current driving environment is diagnosed as an abnormal state. According to the corresponding environmental risk level and warning level in the diagnosis result, warning information matched with the abnormal type and severity is generated, and the warning information is issued to the target intelligent driving vehicle causing the abnormal state or affected by the abnormal state through the cloud.

[0078] Specifically, the abnormal diagnosis module and the warning module of the cloud system execute the following steps:

[0079] S41: Read the environmental risk indicators and risk levels of the target vehicle in the current diagnosis cycle;

[0080] S42: Determine if the abnormal triggering conditions are met: If the current risk indicator / level exceeds the preset threshold, or if it remains at a high risk level for three consecutive diagnostic cycles (the number of cycles can be customized), then the abnormal conditions are met; if none of these conditions are met, then no warning is triggered, only the diagnostic record is updated and the current cycle ends.

[0081] S43: When abnormal conditions are met, the warning level shall be determined according to the pre-established correspondence between "environmental risk level and warning level" and the risk level: no warning for low risk level, warning for medium risk level, and warning for high risk level; and one or more of warning, alert, or emergency warning shall be determined according to the warning level.

[0082] S44: Based on the determined warning level, and combined with the current environmental risk indicators, environmental risk level and its changing trend, generate warning information that matches the type and severity of the anomaly, including risk level (e.g., "high risk"), anomaly type (e.g., "risk of collision with vehicles ahead"), recommended driving behavior (e.g., "immediately reduce to 30km / h and give way to vehicles in the left lane") and anomaly impact range (e.g., "impact range 100 meters").

[0083] S45: Through the vehicle-cloud communication network, it issues warning information to intelligent driving vehicles that cause abnormal states (such as vehicles that change lanes illegally) or are affected by abnormalities (such as surrounding vehicles in the abnormal area). The vehicle-side system informs the driver through the vehicle display screen, voice prompts, light alarms, etc., or directly transmits the information to the vehicle control module to execute corresponding driving operations.

[0084] S46: Regardless of whether an alert is triggered, the environmental risk indicators, environmental risk level, and alert status of the current diagnostic cycle will be recorded in the cloud for subsequent statistical analysis and model optimization, and the next diagnostic cycle will be started at the same time.

[0085] In summary, this invention establishes a cloud-based platform for full-sample trajectory and environmental data, integrating multi-source sensing data to overcome the limitations of a single-vehicle perspective. This allows for the characterization of the spatiotemporal evolution of driving environment risks across a larger spatial range and over a longer timescale. By comprehensively considering the constraints of static road structure and the dynamic interaction of multiple vehicles, a static risk potential energy field and a dynamic risk field are constructed and fused to calculate environmental risk indicators, achieving precise risk quantification and solving the problem of existing technologies relying on single indicators and providing coarse risk characterization. A data-driven clustering method is used to classify risk levels and establish a mapping relationship between risk indicators and risk levels, achieving a refined representation of risk levels and providing a scientific basis for differentiated early warning. An anomaly diagnosis mechanism based on threshold exceeding limits and persistent high risk is introduced to generate early warning information matching the anomaly type and severity, and distributes it to relevant vehicles, improving the timeliness and relevance of early warnings and effectively supporting the safe driving of intelligent vehicles.

[0086] In the above embodiments, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0087] In the various embodiments provided in this application, it should be understood that the disclosed apparatus can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some service interface; the indirect coupling or communication connection between apparatuses or units may be electrical or other forms.

[0088] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0089] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0090] The above are merely exemplary embodiments of this disclosure and should not be construed as limiting the scope of this disclosure. Any equivalent changes and modifications made in accordance with the teachings of this disclosure shall still fall within the scope of this disclosure. Other embodiments of this disclosure will readily conceive of those skilled in the art upon consideration of the specification and the disclosure of practical truths.

[0091] This application is intended to cover any variations, uses, or adaptations of this disclosure that follow the general principles of this disclosure and include common knowledge or customary techniques in the art not described in this disclosure. The specification and embodiments are to be considered exemplary only.

Claims

1. A method for diagnosing abnormal states in an intelligent driving environment based on vehicle-cloud collaboration, characterized in that, Includes the following steps: S1. Receive full sample trajectory and environmental data of the target intelligent driving vehicle and its surrounding environment, and establish a full sample trajectory and environmental data platform; S2. Based on the trajectory data and environmental data related to the target intelligent driving vehicle in the full-sample trajectory and environmental data platform, calculate the environmental risk index of the spatiotemporal location of the target intelligent driving vehicle. S3. Based on the values ​​of environmental risk indicators, classify the environmental risk level of the target intelligent driving vehicle's location in time and space; S4. Based on environmental risk indicators and environmental risk levels, determine whether the current driving environment is in an abnormal state; If the determination is yes, then the corresponding warning information is generated and sent to the corresponding intelligent driving vehicle.

2. The method according to claim 1, characterized in that, The full sample trajectory and environmental data are formed by time synchronization and coordinate alignment of the driving status information and road environment information of the target intelligent driving vehicle and other vehicles on the same road segment.

3. The method according to claim 1, characterized in that, Step S1 includes: performing latitude and longitude coordinate transformation on the full sample trajectory and environmental data, projecting the vehicle trajectory onto the high-precision map coordinate system, performing road matching and lane matching, aligning to the road and lane topology structure of the high-precision map, and constructing a full sample trajectory and environmental data platform by combining the corresponding road attributes and traffic facility information.

4. The method according to claim 1, characterized in that, Step S2 includes: extracting static road elements and dynamic traffic elements based on the target intelligent driving vehicle and its surrounding vehicle data in the full sample trajectory and environmental data platform, calculating static risk indicators and dynamic risk indicators, and fusing them to generate intelligent driving environment risk indicators.

5. The method according to claim 4, characterized in that, The static risk indicator risk value is calculated as follows: In the formula, This represents the sum of potential energy risk values ​​generated by the lane lines at point P(xj, yj) in a driving scenario. Represented as datum weights, The lateral coordinates of the lane markings Let be the horizontal coordinate of a point in the scene. This is the exponential decay coefficient.

6. The method according to claim 4, characterized in that, The dynamic risk indicator risk value is calculated as follows: In the formula, G is the overall control constant. This is the speed decay control coefficient. This is the road correction factor. Let λ be the virtual mass of the object, d be the distance from the outer point to the rectangular border, and λ be the velocity correction coefficient.

7. The method according to claim 1, characterized in that, Step S3 includes: during the offline phase or system initialization phase, selecting representative driving scenarios from the full sample data platform, collecting environmental risk indicator samples and using clustering methods to divide risk clusters, determining risk levels and risk value ranges, and establishing mapping relationships; during online operation, converting the current environmental risk indicators into corresponding environmental risk levels according to the mapping relationships.

8. The method according to claim 7, characterized in that, The environmental risk level is divided into at least three levels: low, medium, and high, and each level is configured with a corresponding diagnostic meaning and warning level; among them, the low level corresponds to a normal or slightly risky state, the medium level corresponds to a risky state that requires attention, and the high level corresponds to a high-risk or dangerous state that requires intervention measures.

9. The method according to claim 1, characterized in that, Step S4 includes: within each diagnostic cycle, determining whether an abnormal condition is met based on environmental risk indicators and environmental risk levels: first, the environmental risk indicators or environmental risk levels exceed a preset threshold; second, the environmental risk level remains high for multiple consecutive cycles; when either condition is met, generating differentiated early warning information that matches the type and severity of the abnormality, and sending it to the intelligent driving vehicle that caused or was affected by the abnormal state.

10. An intelligent driving environment abnormal state diagnosis system based on vehicle-cloud collaboration, characterized in that, It includes a vehicle-side system and a cloud-based system, wherein the vehicle-side system and the cloud-based system are connected through a vehicle-cloud communication network; The vehicle-mounted system includes: The data acquisition module is used to collect data on the target intelligent driving vehicle and its driving environment. The data preprocessing module is used to synchronize the collected data in time and align the coordinates. The data transmission module is used to upload preprocessed data to the cloud system and receive information sent by the cloud system; The cloud system includes: The data receiving module is used to receive data uploaded by multiple vehicle-side systems; The data processing and database building module is used to construct a full-sample trajectory and environmental data platform; The risk indicator calculation module is used to calculate the environmental risk indicators of the target intelligent driving vehicle's location in time and space. The risk level classification module is used to classify the environmental risk level of the target intelligent driving vehicle's location in time and space; The anomaly diagnosis module is used to determine whether the current driving environment is in an abnormal state based on risk indicators and risk levels. The early warning module is used to generate differentiated early warning information that matches the type and severity of the anomaly and send it to the corresponding vehicle-side system.

Images