Road condition early warning method, vehicle and medium
By combining multimodal data collaborative analysis and cloud-based credibility fusion, the problems of information isolation and navigation fragmentation in vehicle identification and early warning in extreme road conditions have been solved, achieving efficient and accurate sharing of early warning information and navigation support, thereby improving road safety and traffic efficiency.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-12-31
- Publication Date
- 2026-03-10
Smart Images
Figure CN121640718A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present disclosure relates to the technical field of vehicle safety, in particular to a road condition early warning method, a vehicle and a medium. BACKGROUND
[0002] With the development of intelligent transportation systems, vehicles gradually have environment perception capabilities based on sensors such as cameras and radars. However, under extreme weather or sudden geological disaster conditions, the extreme road conditions such as collapse, deep pit and short-time waterlogging that may occur on the road have local, sudden and hidden characteristics. The current widely relied on traditional navigation maps and roadside fixed monitoring systems are difficult to discover and feedback such sudden dangers in time due to long update cycle and sparse monitoring points, and there is a serious early warning lag problem.
[0003] Most of the existing vehicles with perception capabilities adopt a single-vehicle intelligent mode, and the front-view camera and radar system of the vehicle mainly serve the auxiliary driving or emergency braking of the vehicle. Even if the front vehicle identifies a dangerous road condition, there is a lack of effective cooperative communication mechanism between vehicles and the cloud, and the specific road condition type, accurate position and other information cannot be shared in real time with the vehicles that will enter the same road section, resulting in the risk being continuously transmitted and amplified in the vehicle flow.
[0004] Therefore, how to realize real-time and accurate perception of extreme road conditions and realize instant sharing of information through an efficient cooperative mechanism to improve road traffic safety and efficiency has become a technical problem that needs to be solved by those skilled in the art. SUMMARY
[0005] In view of the above problems, the present disclosure provides a road condition early warning method, a vehicle and a medium that overcome the above problems or at least partially solve the above problems, and the technical solutions are as follows: A road condition early warning method, the method comprising: acquiring multi-modal perception data of a driving vehicle on a current road section, and identifying whether an extreme road condition exists on the current road section; if an extreme road condition exists, uploading the identification result to a cloud service platform, and generating a reporting and statistical task corresponding to the current road section in the cloud service platform based on the identification result; after the reporting and statistical task is executed, performing a credibility fusion calculation based on the statistical result to determine whether to generate an extreme road condition early warning event; if an extreme road condition early warning event is generated, pushing the extreme road condition early warning event to a navigation system to provide the extreme road condition early warning event for a subsequent vehicle whose planned path includes the current road section through the navigation system.
[0006] The road condition warning method disclosed herein uploads the initial judgment results of extreme road conditions completed by a single vehicle at the edge to the cloud for aggregation and fusion, and finally pushes them to the navigation system. This realizes a complete link from individual vehicle perception to group-shared warning, which not only effectively breaks down the information silos of single-vehicle intelligence, enabling the danger perceived by the vehicle in front to serve the driving safety of subsequent vehicles in real time, but also provides drivers with proactive and clear risk avoidance decision support by deeply embedding the warning information into the navigation route planning and guidance process, thereby significantly improving road traffic safety and efficiency as a whole.
[0007] Optionally, the multimodal perception data includes road surface image data, vehicle vibration data, and vehicle speed data; identifying whether the current road segment has extreme road conditions specifically includes: performing an initial road condition category determination on the road surface image data based on the image recognition model in the vehicle edge computing module to obtain an initial road condition determination result; wherein, the initial road condition determination result includes an initial road condition determination category and a category confidence level; determining whether the initial road condition determination category is a preset extreme road condition type and whether the category confidence level exceeds a preset image confidence level threshold; if so, triggering a collaborative analysis of the vehicle vibration data and the vehicle speed data, and performing data fusion determination based on the collaborative analysis result and the category confidence level to determine whether the current road condition is an extreme road condition and the road condition type.
[0008] In this embodiment, by using high-confidence image recognition results as triggering conditions, and then collaboratively analyzing vibration and vehicle speed data, a hierarchical, cross-validated judgment logic is constructed. This ensures that the identification of extreme road conditions does not rely on a single, potentially distorted signal, but is based on the fusion decision of multi-source evidence. Thus, under the condition of limited on-board edge computing resources, the accuracy and robustness of the initial identification are significantly improved, providing a high-quality data source for subsequent cloud processing.
[0009] Optionally, the triggering of the collaborative analysis of the vehicle vibration data and the vehicle speed data specifically includes: processing the vertical acceleration time-series signal in the vehicle vibration data based on a preset peak impact detection algorithm to determine the peak acceleration within a first preset time window from the time the initial road condition determination result is generated, and determining whether the peak acceleration exceeds a preset vibration intensity threshold and whether the duration is shorter than a preset short-time threshold, so as to generate a vibration event detection result; extracting a continuous vehicle speed frame sequence within a second preset time window from the time the initial road condition determination result is generated in the vehicle speed data, and calculating the average vehicle speed decrease rate of the driving vehicle within the second preset time window based on the continuous vehicle speed frame sequence, so as to generate a vehicle speed decrease detection result; integrating the vibration event detection result and the vehicle speed decrease detection result to obtain the collaborative analysis result.
[0010] In this embodiment, by precisely aligning the time window of data analysis with the trigger time of image recognition, and extracting two types of physical features that are highly correlated with different extreme road conditions, namely "severe bumps" and "abnormal deceleration", the fusion judgment of multi-source data has clear physical meaning and scene specificity, avoiding the simple stacking or generalization of sensor data. This allows the fusion judgment rules to be mapped more accurately to specific road condition types such as "potholes" and "collapses", thus improving the precision of classification.
[0011] Optionally, the data fusion determination based on the collaborative analysis results and the category confidence specifically includes: matching the corresponding preset fusion determination rule based on the initial road condition determination category; wherein, the preset fusion determination rule includes the collaborative analysis conditions required for the corresponding extreme road condition type; inputting the collaborative analysis results into the selected fusion determination rule to determine whether the peak vertical acceleration meets the vibration threshold condition and / or whether the vehicle speed decrease rate meets the vehicle speed decrease threshold condition; if they meet, it is determined that the current road segment has an extreme road condition type corresponding to the initial road condition determination category.
[0012] In this embodiment, by pre-setting differentiated fusion judgment rules for different extreme road condition types, the system can intelligently call the most relevant collaborative analysis conditions for verification based on the current suspected road condition. This rule-based data fusion method not only makes the judgment process transparent and interpretable, but also facilitates flexible adjustment or addition of new judgment rules as the actual road condition type library expands, enhancing the system's adaptability and maintainability.
[0013] Optionally, generating a reporting statistics task corresponding to the current road segment in the cloud service platform based on the recognition result specifically includes: extracting the location information of the current road segment contained in the recognition result; searching the task management database of the cloud service platform based on the location information to determine whether there is an open target statistics task that matches the location information; if not, creating a new statistics task as the target statistics task and configuring a preset duration task execution time window for the target statistics task; recording the recognition result as a reporting result and storing it in association with the target statistics task.
[0014] In this embodiment, by dynamically creating and managing a time-limited reporting and statistical task for each suspected extreme road condition location, the cloud service platform can automatically collect and correlate multiple reporting records from different vehicles over a period of time. This mechanism effectively organizes massive and scattered vehicle-side reporting data, preparing data for subsequent group credibility calculations, and is a key organizational link in realizing the transformation from "single-point reporting" to "group confidence".
[0015] Optionally, the step of performing credibility fusion calculation based on statistical results to determine whether to generate an extreme road condition warning event specifically includes: receiving multiple reporting results from multiple vehicles on the current road segment within the task execution time window of the reporting statistics task; performing spatiotemporal clustering analysis on the multiple reporting results to aggregate reporting results that are geographically adjacent and have the same type of extreme road conditions into a group event; calculating the comprehensive credibility score of the group event based on the number of reporting results in the group event, the confidence level of the categories contained in each reporting result, and the reporting time of the reporting results, and determining whether the comprehensive credibility score exceeds a preset credibility threshold; if so, generating the extreme road condition warning event.
[0016] In this embodiment, by comprehensively scoring the credibility of spatiotemporally clustered group events, the system cleverly integrates three dimensions: "number of reporting vehicles," "individual identification confidence level," and "information freshness." This allows the system to effectively filter out noise data caused by misjudgments or brief interference from individual vehicle sensors. Only when a sufficient number of vehicles report similar events with high confidence within a similar spatiotemporal range are they confirmed as credible warning events. This mechanism significantly reduces the system's false alarm rate and ensures the reliability of the pushed warning information.
[0017] Optionally, pushing the extreme road condition warning event to the navigation system specifically includes: encapsulating the location information and road condition type information of the current road segment contained in the extreme road condition warning event into a data packet conforming to a preset navigation application programming interface protocol; and sending the data packet to the cloud server of the navigation system, so that the cloud server associates the extreme road condition warning event with the corresponding current road segment in the electronic map database.
[0018] In this embodiment, by encapsulating warning events into standard data packets and interfacing with the navigation system, the "last mile" of danger information delivery from the perception system to the user terminal is bridged. This deep integration ensures that warning events can be seamlessly embedded into navigation map data and business logic, providing the necessary technical pathway for accurate push and display of warnings.
[0019] Optionally, providing the extreme road condition warning event to subsequent vehicles whose planned routes include the current road segment through the navigation system specifically includes: responding to the route planning request or real-time navigation request initiated by the subsequent vehicle to the navigation system, determining whether the subsequent vehicle will approach the current road segment within a preset warning distance; if so, rendering a warning icon and text corresponding to the road condition type information on the vehicle's in-vehicle display interface, and broadcasting a voice warning content corresponding to the road condition type information through the vehicle's audio system.
[0020] In this embodiment, the navigation system proactively triggers warning prompts during route planning and real-time navigation, providing users with timely, intuitive, and guiding warning services. The system not only provides visual warnings through dedicated icons but also reinforces reminders through voice announcements and offers detour suggestions when necessary. This provides drivers with ample cognitive and decision-making support at critical moments, forming a complete warning service loop and effectively ensuring driving safety.
[0021] A vehicle includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform: acquiring multimodal perception data of a vehicle traveling on a current road segment, and identifying whether extreme road conditions exist on the current road segment; if so, uploading the identification result to a cloud service platform, and generating a reporting statistical task corresponding to the current road segment on the cloud service platform based on the identification result; after the reporting statistical task is completed, performing a confidence fusion calculation based on the statistical result to determine whether to generate an extreme road condition warning event; if generated, pushing the extreme road condition warning event to a navigation system, so as to provide the extreme road condition warning event to subsequent vehicles whose planned paths include the current road segment through the navigation system.
[0022] A computer-readable storage medium stores computer-executable instructions, the computer-executable instructions being configured to: acquire multimodal perception data of a vehicle traveling on a current road segment, and identify whether extreme road conditions exist on the current road segment; if so, upload the identification result to a cloud service platform, and generate a reporting and statistical task corresponding to the current road segment on the cloud service platform based on the identification result; after the reporting and statistical task is executed, perform a credibility fusion calculation based on the statistical result to determine whether to generate an extreme road condition warning event; if generated, push the extreme road condition warning event to a navigation system, so as to provide the extreme road condition warning event to subsequent vehicles whose planned paths include the current road segment through the navigation system. Attached Figure Description
[0023] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings: Figure 1 A flowchart illustrating a traffic condition warning method according to an embodiment of this disclosure is shown; Figure 2 A schematic diagram of the structure of a vehicle according to an embodiment of the present disclosure is shown. Detailed Implementation
[0024] Exemplary embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.
[0025] With the in-depth development of intelligent transportation systems and vehicle-to-everything (V2X) technology, the real-time perception and risk warning capabilities of vehicles regarding the driving environment have become crucial for improving road safety. Currently, in the identification and warning of extreme road conditions (such as road collapses, severe potholes, and short-term water accumulation), existing technologies mainly rely on single-vehicle perception systems and periodically updated navigation map data. However, this approach has significant drawbacks: First, the system's ability to perceive sudden and localized extreme road conditions heavily depends on single-vehicle sensors, lacking collaborative fusion and cross-validation of multi-source data, resulting in high false alarm and false negative rates. Second, the perceived information is limited to the vehicle itself, lacking an effective real-time collaborative sharing mechanism between vehicles and between vehicles and the cloud, thus failing to form a closed-loop warning system. Third, the warning information is disconnected from navigation guidance; the prompts received by users are vague and simplistic, making it difficult for them to make timely and clear risk avoidance decisions.
[0026] Specifically, traditional solutions have the following technical shortcomings: 1. Insufficient Reliability of Single-Vehicle Perception and Judgment: Existing technologies typically rely on single sensors such as front-view cameras or radar for obstacle recognition, failing to fully integrate multimodal data such as vehicle vibration and speed changes for collaborative verification of image recognition results. For example, in scenarios with sudden changes in lighting or road surface reflections, image recognition alone can easily misjudge shadows or water stains as potholes; conversely, when a vehicle is normally passing over speed bumps or minor bumps, the lack of supporting image evidence may cause it to overlook actual pothole risks. This single-modal judgment method, lacking a cross-validation mechanism, results in low reliability of recognition results, making it difficult to serve as a reliable basis for issuing warnings.
[0027] 2. Isolated and Delayed Information Sharing: Even if a single vehicle successfully identifies a hazard, the existing architecture lacks a low-latency, standardized mechanism for uploading and aggregating information. Identification results often remain on the vehicle's end or are uploaded only as unstructured data, preventing the cloud from effectively correlating, statistically analyzing, and merging the credibility of multiple reports from different vehicles targeting the same location. This hinders the rapid dissemination of hazard information within traffic flow, leaving following vehicles continuously exposed to the same risk, thus limiting the timeliness and coverage of warnings.
[0028] 3. Disconnect between early warning and navigation services: Mainstream navigation systems receive limited types of road condition event information, and the update delays are significant, especially lacking specific support for extreme road conditions such as road collapses. Even when relevant data is received, the lack of deep integration with map data and standardized early warning interfaces makes it difficult to proactively avoid hazards during the route planning stage, or to provide strong voice prompts, icons, and detour route guidance in real-time navigation. Users often only receive general warnings when they are close to danger, losing the decision-making window for proactive avoidance.
[0029] Therefore, this application provides a road condition warning method. Figure 1 This is a schematic flowchart illustrating a road condition warning method provided in one or more embodiments of this specification. The method can be applied to different types of vehicles, and the process can be executed by computing devices in the relevant field (e.g., controllers installed in the vehicle, in-vehicle systems, or servers located in the cloud). Certain input parameters or intermediate results in the process can be manually adjusted to help improve accuracy.
[0030] The early warning method disclosed in this embodiment can be implemented using a terminal device or a server, and this application does not impose any special limitations on it. For ease of understanding and description, the following embodiments are all described in detail using a server as an example. It should be noted that the server can be a single device or a system composed of multiple devices, i.e., a distributed server, and this application does not impose any specific limitations on it.
[0031] like Figure 1 As shown in the figure, this disclosure provides a road condition warning method, including: Step 101: Obtain multimodal perception data of the vehicle in the current road segment and identify whether there are extreme road conditions in the current road segment.
[0032] In this embodiment, the "multimodal perception data" refers to a set of data collected synchronously or quasi-synchronously from multiple physical sensors mounted on the vehicle, which can reflect the interaction state between the vehicle and the road in multiple dimensions. For example, this data includes, but is not limited to, road surface image data captured by the forward-looking camera, vehicle vibration data collected by the inertial measurement unit (IMU), and vehicle speed data provided by wheel speed sensors or the vehicle CAN bus.
[0033] Understandably, these data types correspond to visual perception, the physical impact experience of the vehicle chassis, and changes in the vehicle's motion state, respectively, collectively forming the composite information basis for judging abnormal road conditions. It should be noted that the "current road segment" is not a fixed-length road unit, but rather refers to the road area continuously traversed by the vehicle within a specific time window (e.g., the time for perception and calculation), its range dynamically defined by vehicle position, speed, and data processing latency. Extreme road conditions refer to a sudden, localized, and highly potentially risky abnormality in the physical state of the road surface. This abnormality exceeds normal road maintenance standards and the range of normal wear and tear. Without advance warning and avoidance or deceleration measures, it is highly likely to pose a direct and serious threat to the safety of vehicles (causing accidents), mechanical integrity (causing vehicle damage), and traffic efficiency (leading to congestion or stagnation).
[0034] In this embodiment, after acquiring the multimodal perception data of the vehicle in the current road segment, the multimodal perception data is subjected to localized, real-time fusion analysis and decision-making.
[0035] In a specific embodiment of this disclosure, identifying whether the current road segment has extreme road conditions specifically includes: performing an initial road condition category determination on the road surface image data based on the image recognition model in the vehicle edge computing module to obtain an initial road condition determination result; wherein, the initial road condition determination result includes an initial road condition determination category and a category confidence level; determining whether the initial road condition determination category is a preset extreme road condition type and whether the category confidence level exceeds a preset image confidence level threshold; if so, triggering a collaborative analysis of the vehicle vibration data and the vehicle speed data, and performing a data fusion determination based on the collaborative analysis result and the category confidence level to determine whether the current road condition is an extreme road condition and the road condition type.
[0036] In a specific embodiment of this disclosure, triggering the collaborative analysis of the vehicle vibration data and the vehicle speed data specifically includes: processing the vertical acceleration time-series signal in the vehicle vibration data based on a preset peak impact detection algorithm to determine the peak acceleration within a first preset time window from the time the initial road condition determination result is generated, and determining whether the peak acceleration exceeds a preset vibration intensity threshold and whether the duration is shorter than a preset short-time threshold, so as to generate a vibration event detection result; extracting a continuous vehicle speed frame sequence within a second preset time window from the time the initial road condition determination result is generated in the vehicle speed data, and calculating the average vehicle speed decrease rate of the driving vehicle within the second preset time window based on the continuous vehicle speed frame sequence, so as to generate a vehicle speed decrease detection result; and integrating the vibration event detection result and the vehicle speed decrease detection result to obtain the collaborative analysis result.
[0037] In a specific embodiment of this disclosure, the data fusion determination based on the collaborative analysis results and the category confidence specifically includes: matching a corresponding preset fusion determination rule based on the initial road condition determination category; wherein, the preset fusion determination rule includes the collaborative analysis conditions required for the corresponding extreme road condition type; inputting the collaborative analysis results into the selected fusion determination rule to determine whether the peak vertical acceleration meets the vibration threshold condition and / or whether the vehicle speed decrease rate meets the vehicle speed decrease threshold condition; if so, determining that the current road segment has an extreme road condition type corresponding to the initial road condition determination category.
[0038] First, it should be noted that the in-vehicle edge computing module is a computing unit with a certain computing power integrated inside the vehicle. It deploys a lightweight algorithm model and can complete preliminary intelligent identification at the data source, thereby reducing the dependence on cloud computing power and network transmission latency.
[0039] Based on the above recognition process, it can be understood that, to achieve highly reliable recognition, the system first performs real-time analysis on continuously acquired road surface image data using an image recognition model (such as an optimized lightweight convolutional neural network) deployed in the edge computing module. This analysis performs semantic segmentation or object detection tasks to complete the initial road condition category determination. This determination outputs an initial road condition category (e.g., "collapse," "pothole," "water accumulation," or "normal") and the corresponding category confidence level. This confidence level quantifies the model's confidence in the judgment result of the current frame image.
[0040] Furthermore, the system performs a crucial logical judgment: determining whether the initial road condition classification is a preset extreme road condition type and whether the confidence level of the classification exceeds a preset image confidence threshold. Here, the "preset extreme road condition type" is a predefined set of high-risk road conditions requiring warnings, such as the aforementioned collapses and potholes; while the "image confidence threshold" is a set empirical threshold used to filter out suspected events with unclear image evidence and low judgment certainty. Only when both conditions are met—"it is an extreme road condition type" and "the confidence level is high enough"—will the subsequent, more in-depth collaborative verification process be triggered. For example, if the image model identifies a regular asphalt repair as a "pothole" with only a very low confidence level, the system will ignore this event to avoid generating a large number of invalid reports due to image misjudgment.
[0041] If the above conditions are met, it signifies that a valid "suspected event" has been triggered, and the system immediately initiates a collaborative analysis of the vehicle vibration data and the vehicle speed data. The collaborative analysis aims to cross-validate the image recognition results using data from different physical modes.
[0042] Specifically, collaborative analysis includes two aspects. Firstly, it analyzes vehicle vibration data: the system extracts the vertical acceleration time-series signal within a first preset time window (e.g., within 10 seconds after triggering) from the moment image recognition is triggered. This signal is then processed using a preset impact peak detection algorithm (e.g., finding pulses in the signal that exceed a specific amplitude and have an extremely short duration). The aim is to determine whether there exists a "severe jolt event" where the peak acceleration exceeds a preset vibration intensity threshold (e.g., twice the acceleration due to gravity) and the duration is shorter than a preset short-time threshold (e.g., 0.3 seconds). For example, when a vehicle drives over a real pothole, a short and strong upward or downward impact signal is typically generated within a very short time after the image detects the anomaly.
[0043] On the other hand, the system analyzes vehicle speed data: it extracts a continuous sequence of vehicle speed frames within a second preset time window (e.g., from 10 seconds before triggering to the triggering time) starting from the triggering moment. By calculating the rate of change between the initial and final speeds within this window, the system obtains the average rate of decrease in vehicle speed during that time period. For example, when a driver sees what appears to be a collapse ahead, they may instinctively brake suddenly, causing a significant drop in vehicle speed within a short period.
[0044] Further, the system executes a data fusion and determination process. This process is based on a set of preset fusion and determination rules associated with road condition types. Specifically, the system matches the corresponding rules according to the category determined by the initial image (such as "potholes"). The fusion and determination rules in this embodiment define the collaborative analysis conditions required to determine such extreme road conditions; for example, it may require the simultaneous detection of "severe bump events." Subsequently, the system inputs the vibration event detection results and vehicle speed reduction detection results obtained from the aforementioned collaborative analysis into the matching rules for logical judgment. If the conditions specified in the rules are met, it is finally determined that the current road segment has an extreme road condition type corresponding to the category determined by the initial image.
[0045] For example, the criteria for determining "potholes" might be set as follows: both "image confidence exceeds a threshold" and "severe bumps exceeding the threshold" must be met simultaneously. If the system initially determines "potholes" with a high confidence level, but the vibration sensor does not detect any impact matching the characteristics during the same period, the fusion determination result will be "no," and this event will not be recognized as a reliable extreme road condition.
[0046] The multimodal fusion judgment method in this embodiment greatly reduces the probability of triggering a warning due to false alarms from a single sensor (such as image misjudgment caused by light and shadow), ensuring that the recognition results of the vehicle end have high reliability and providing a high-quality data source for subsequent cloud aggregation and decision-making.
[0047] Step 102: If it exists, upload the identification result to the cloud service platform, and generate a reporting and statistical task corresponding to the current road segment in the cloud service platform based on the identification result.
[0048] In this embodiment, once the vehicle-mounted edge computing module completes the fusion determination and confirms that there are extreme road conditions in the current road segment, the system immediately initiates the information collaboration and aggregation process.
[0049] First, the key information of this identified event needs to be packaged into a structured identification result and uploaded to the cloud service platform via the vehicle's built-in V2X communication module (supporting 4G / 5G or dedicated vehicle-to-everything communication). It can be understood that the "cloud service platform" here is a remote server cluster with high availability, elastic computing, and massive data management capabilities, serving as the central node for vehicle group data aggregation, analysis, and decision-making. The identification result includes at least the following core fields: the type of extreme road condition (e.g., a pothole), the identification location (usually GPS coordinates, with optional road number), the timestamp of the event, and the category confidence score derived from the vehicle-side fusion judgment. For example, after a car identifies a deep pothole, its vehicle system immediately generates a data packet containing "Type: Pothole," "Location: Longitude X, Latitude Y," "Time: 2023-10-27 14:30:05," and "Confidence Score: 0.85," and sends it to the cloud.
[0050] It should be noted that simple data uploads are insufficient to create effective collaborative early warning systems. If the cloud processes each reported data independently and issues an early warning immediately, it is highly susceptible to accidental misjudgments by individual vehicles, leading to a flood of warnings, interfering with users, and reducing the system's credibility. Therefore, in this embodiment, after uploading the identification results to the cloud service platform, a reporting and statistical task corresponding to the current road segment is generated on the cloud service platform based on the identification results.
[0051] In one specific embodiment of this disclosure, the step of generating a reporting statistics task corresponding to the current road segment in the cloud service platform based on the identification result specifically includes: extracting the location information of the current road segment contained in the identification result; searching the task management database of the cloud service platform based on the location information to determine whether there is an open target statistics task that matches the location information; if not, creating a new statistics task as the target statistics task and configuring a preset duration task execution time window for the target statistics task; recording the identification result as a reporting result and storing it in association with the target statistics task.
[0052] Based on the above task generation process, it can be understood that after receiving a recognition result, the cloud service platform first extracts the location information of the current road segment contained in the recognition result. This location information serves as the spatial anchor point for all subsequent collaborative processing. Next, the cloud platform searches its task management database based on this location information. In this embodiment, the "task management database" is a dedicated data storage area used by the cloud to track and manage all active reported statistical tasks. The purpose of the search is to determine whether there exists an open target statistical task that matches the location information. In this embodiment, "matching" typically refers to a location distance within a preset spatiotemporal clustering radius (e.g., within 100 meters) and the task not yet closed.
[0053] If no existing task meets the criteria, it means the current report is likely a "first report" of extreme road conditions at that location. In this case, the cloud system will create a new statistical task as the target statistical task. This new task will be assigned a unique task ID and the location of the first report will be recorded as the task anchor point. A crucial step is that the system will configure a preset execution time window for the target statistical task (e.g., 30 minutes from task creation). This time window design has profound practical considerations: based on the assumptions of the suddenness and timeliness of handling extreme road conditions. On the one hand, it provides sufficient time for other vehicles that may subsequently pass through the area to report the same event, thus collecting collective evidence; on the other hand, it avoids the task being open indefinitely, allowing for timely decisions based on the collected evidence (generating an alert or closing the task). For example, on a main urban road, the first vehicle to report an obstacle on the road triggers the generated task, and its 30-minute window is sufficient for dozens or even hundreds of subsequent vehicles passing through the area to report their perception data.
[0054] Finally, regardless of whether the task is newly created or already exists, the cloud will record the identification result as a reporting result and store it in association with the target statistical task. This means that reported data from different vehicles and at different times, but pointing to the same or similar road conditions near the same geographical location, will be categorized under the same "reporting statistical task" for unified management.
[0055] For example, within five minutes of the first sedan reporting the sinkhole and creating a task, a second SUV and a third van also drove through the area. Their onboard systems might produce slightly different confidence levels (e.g., 0.78, 0.90) due to differences in sensor perspective and lighting conditions, but the reported locations were close and of the same type. The cloud system would associate these subsequent reports with the original task ID created for that sinkhole, serving as multiple pieces of evidence for that task.
[0056] This embodiment transforms fragmented and asynchronous vehicle-side reporting into a structured investigation process with a clear objective (verifying specific road conditions), timeliness, and the ability to continuously accumulate evidence by organizing tasks. This lays a solid data foundation for the next step of group credibility fusion calculation.
[0057] Step 103: After the reporting statistics task is completed, a credibility fusion calculation is performed based on the statistical results to determine whether an extreme road condition warning event should be generated.
[0058] In this embodiment, the "completion of reporting statistical tasks" is a key decision trigger point, signifying that the preset task execution time window for this task has expired, and the cloud service platform stops receiving new reported results for this statistical analysis. It is understood that setting a fixed execution window is a strategy to balance timeliness and the sufficiency of evidence. For example, a statistical task created at 14:30 targeting a suspected pothole will have its 30-minute time window close precisely at 15:00.
[0059] Furthermore, the cloud platform will make the final decision on all evidence collected within this window period. It should be noted that in some other implementations, "execution completion" may also be triggered by conditions such as reaching a preset reporting threshold, but the time window is the primary basis, which ensures that a system conclusion can be given within a predictable time for any sudden road conditions, avoiding indefinite waiting.
[0060] In a specific embodiment of this disclosure, the step of performing credibility fusion calculation based on statistical results to determine whether to generate an extreme road condition warning event specifically includes: receiving multiple reporting results from multiple vehicles on the current road segment within the task execution time window of the reporting statistics task; performing spatiotemporal clustering analysis on the multiple reporting results to aggregate reporting results that are geographically adjacent and have the same type of extreme road conditions into a group event; calculating the comprehensive credibility score of the group event based on the number of reporting results in the group event, the category confidence level contained in each reporting result, and the reporting time of the reporting results, and determining whether the comprehensive credibility score exceeds a preset credibility threshold; if so, generating the extreme road condition warning event.
[0061] Based on the aforementioned credibility fusion calculation, it is understandable that the purpose of this calculation is not simply to accumulate the reported results, but rather to assess the credibility of the road condition anomaly from the perspective of collective perception through a multi-factor fusion intelligent model, thereby deciding whether to generate an authoritative extreme road condition warning event. The process first requires receiving multiple reports from various vehicles regarding the current road segment within the task execution time window of the reporting statistics task. These results constitute the original evidence set for collective perception. For example, within a 30-minute window, 5 sedans, 3 SUVs, and 2 trucks may have passed through the suspected road segment. Of these, 8 vehicles reported a "pothole" event, but with varying confidence levels. The other 2 vehicles may not have reported it due to sensor obstruction or the judgment threshold not being met, or they may have reported other types of incidents.
[0062] Furthermore, faced with these reported data that may be slightly discrete in time and space, the cloud first performs spatiotemporal clustering analysis. The purpose of this analysis is to intelligently merge scattered reported points. The logic is to aggregate reported results that are geographically close (e.g., within 50 meters of each other) and have the same extreme road condition type into a group event. This effectively solves the problem that the reported coordinates of the same pothole may differ slightly due to vehicle trajectory and GPS accuracy errors, identifying them as multiple reports for the same physical entity, rather than multiple isolated small events. For example, although the coordinates of 8 reported "potholes" do not completely overlap, they are all distributed within a 50-meter range east of an intersection. They will be clustered into a group event representing "a pothole exists east of this intersection".
[0063] Furthermore, the system will perform a core credibility assessment on the clustered group events, namely, calculate the comprehensive credibility score of the group events. In this embodiment, the scoring model is a carefully designed function aimed at quantifying the strength of group evidence. It comprehensively considers three key dimensions: first, the number of reported results, which reflects how many independent sources of perception have confirmed the event; the more reports, the lower the possibility of accidental false alarms; second, the category confidence level contained in each reported result, which reflects the degree of confidence each source of perception has in its own judgment; the system usually takes the average of these confidence levels as a quality indicator; and third, the reporting time of the reported results, by introducing a time decay coefficient (e.g., ...). The formula is: Δt is the difference between the current time and the reporting time, and τ is the decay time constant (e.g., 30 minutes), giving higher weight to recently reported evidence. This calculation method has profound rationale: multiple vehicles reporting (numerous) but with low average confidence (poor quality) will not receive high scores; similarly, a single vehicle's high-confidence report without corroboration from other vehicles (few) will also have limited scores; and earlier reports, even if of decent quantity and quality, will be suppressed due to time decay. For example, an event reported with high confidence by a single vehicle at the beginning of the window may fail to trigger an alert if no subsequent vehicles confirm it over a long period.
[0064] Furthermore, the system will determine whether the overall credibility score exceeds a preset credibility threshold. This threshold is the final hurdle for the system to determine whether the information is reliable enough to enter the warning issuance process. If so, the extreme road condition warning event will be generated. This warning event is a structured information object that has been authoritatively certified by the cloud, containing the exact location, type, confidence score, and generation time. For example, after aggregating the aforementioned 8 reports, the calculated overall score may be as high as 0.8, far exceeding the preset threshold of 0.6, and the cloud will officially generate a warning event about "a pothole 50 meters eastbound on XX Road". Conversely, if a suspected waterlogged area only receives 1-2 low-confidence reports near the end of the window period, its score may only be 0.3, and the system will not generate a warning event, thus effectively filtering out occasional false alarms caused by momentary glare or individual vehicle sensor malfunctions.
[0065] This embodiment, based on the computational mechanism of group credibility fusion, is the key to moving from "individual vehicle perception" to "collective intelligent decision-making," which greatly enhances the authority and reliability of the warning information output by the system and provides a solid guarantee for the final safety service.
[0066] Step 104: If generated, push the extreme road condition warning event to the navigation system so that the navigation system can provide the extreme road condition warning event to subsequent vehicles whose planned routes include the current road segment.
[0067] In this embodiment, after the extreme road condition warning event is generated through credibility fusion calculation and confirmation, the decision-making process shifts from "cloud-based decision-making" to "service delivery," aiming to bridge the "last mile" in delivering warning information to end users. It is understandable that if warning events merely remain in the cloud database, they cannot provide any protective value to vehicles actually in motion. Therefore, this embodiment establishes an efficient and standardized information channel, seamlessly integrating authoritative warnings certified by the cloud into the navigation services that drivers rely on daily.
[0068] In one specific embodiment of this disclosure, pushing the extreme road condition warning event to the navigation system specifically includes: encapsulating the location information and road condition type information of the current road segment contained in the extreme road condition warning event into a data packet conforming to a preset navigation application programming interface protocol; sending the data packet to the cloud server of the navigation system, so that the cloud server associates the extreme road condition warning event with the corresponding current road segment in the electronic map database.
[0069] Based on the above push process, it can be understood that the primary step is data encapsulation and interface adaptation. The cloud service platform will encapsulate the key information contained in the extreme road condition warning event, mainly the location information of the current road segment (such as precise GPS coordinates and road chain ID) and road condition type information (such as "road collapse"), according to a preset navigation application programming interface protocol, generating a structured data packet. For example, this protocol may follow the real-time traffic event reporting interface specifications opened by mainstream navigation platforms (such as Gaode Maps and Baidu Maps) to ensure that the data format, field definitions, and communication methods can be accurately parsed by the navigation system. Subsequently, the cloud will send the data packet to the cloud server of the navigation system. It should be noted that this push is proactive and event-driven, triggered immediately once the warning event is generated, aiming to minimize information flow delays. After receiving the data packet, the navigation system cloud server will parse it and associate the warning event as a high-priority dynamic layer information with the corresponding current road segment in the electronic map database. This means that the warning information is "pinned" to a specific location on the electronic map, becoming part of the road's real-time attributes, and can be accessed by the path planning engine and real-time navigation module.
[0070] Furthermore, once the warning event is successfully injected into the navigation system, the key becomes how to provide warning services to subsequent vehicles. Here, "subsequent vehicles" refers to all vehicles whose travel plans may pass through the warning section.
[0071] In one specific embodiment of this disclosure, the step of providing the extreme road condition warning event to subsequent vehicles whose planned routes include the current road segment through the navigation system specifically includes: in response to a route planning request or real-time navigation request initiated by the subsequent vehicle to the navigation system, determining whether the subsequent vehicle will approach the current road segment within a preset warning distance; if so, rendering a warning icon and text corresponding to the road condition type information on the vehicle's in-vehicle display interface, and broadcasting a voice warning content corresponding to the road condition type information through the vehicle's audio system.
[0072] Based on the above process, it can be understood that this process is triggered by the subsequent vehicle's route planning request or real-time navigation request to the navigation system. During the route planning phase, the navigation engine evaluates whether the planned route includes or is too close to the road segment where the warning event is located when calculating the route. In real-time navigation, the system continuously monitors the vehicle's position. When the system determines whether the subsequent vehicle will approach the current road segment associated with the extreme road condition warning event within a preset warning distance (for example, the vehicle is traveling along the planned route and is less than 2 kilometers from the warning point), it will proactively trigger a warning.
[0073] In this embodiment, the warning is presented in a multimodal manner, aiming to ensure that the driver can clearly and promptly be aware of the risk through multiple sensory channels. First, the warning icon and text corresponding to the road condition type information are rendered on the in-vehicle display interface of the following vehicle. For example, a prominent flashing icon representing "collapse" or "pothole" will be superimposed on the map interface at the corresponding location, and a text prompt box may pop up in a prominent position on the screen, stating "Road collapse detected 800 meters ahead". Second, the voice warning content corresponding to the road condition type information is broadcast through the in-vehicle audio system. This is usually a clear and concise synthesized voice, such as "Please note that there is a road collapse ahead, please drive carefully", with a tone and content of urgency far exceeding ordinary congestion warnings. In some preferred embodiments, the navigation system can also directly provide detour routes as alternative options for the driver to switch with one click.
[0074] Understandably, this precise warning, delivered at the right time (before approach) and through the right means (visual and auditory), can effectively break the "information cocoon" created by drivers' familiarity with the route, fatigue, or poor visibility, giving them valuable reaction time to take evasive actions such as slowing down, changing lanes, or taking detours. Ultimately, this transforms collective perception intelligence into tangible driving safety.
[0075] The above are embodiments of the method proposed in this application. Based on the same inventive concept, embodiments of this application also provide a vehicle, the structure of which is as follows: Figure 2 As shown.
[0076] Regarding the apparatus in the above embodiments, the specific manner in which each unit performs its operation has been described in detail in the embodiments related to the method, and will not be elaborated upon here.
[0077] Figure 2 This is a schematic diagram of the structure of a vehicle provided in an embodiment of this application.
[0078] For example, such as Figure 2As shown, the vehicle includes a memory 201 and a processor 202. The memory 201 stores executable program code 2011, and the processor 202 is used to call and execute the executable program code 2011 to perform a road condition warning method.
[0079] This embodiment can divide the vehicle into functional modules according to the above method example. For example, each function can be assigned to a separate module, or two or more functions can be integrated into one processing module. The integrated module can be implemented in hardware. It should be noted that the module division in this embodiment is illustrative and only represents one logical functional division. In actual implementation, there may be other division methods.
[0080] When each functional module is divided according to its corresponding function, the vehicle may include: Acquire multimodal perception data of the vehicle in the current road segment and identify whether there are extreme road conditions in the current road segment; If it exists, the identification result will be uploaded to the cloud service platform, and a reporting and statistical task corresponding to the current road segment will be generated in the cloud service platform based on the identification result; After the reporting and statistics task is completed, a credibility fusion calculation is performed based on the statistical results to determine whether an extreme road condition warning event should be generated. If generated, the extreme road condition warning event is pushed to the navigation system so that the navigation system can provide the extreme road condition warning event to subsequent vehicles whose planned routes include the current road segment.
[0081] Some embodiments of this application provide corresponding to Figure 1 A computer-readable storage medium storing computer-executable instructions, wherein the computer-executable instructions are configured as follows: Acquire multimodal perception data of the vehicle in the current road segment and identify whether there are extreme road conditions in the current road segment; If it exists, the identification result will be uploaded to the cloud service platform, and a reporting and statistical task corresponding to the current road segment will be generated in the cloud service platform based on the identification result; After the reporting and statistics task is completed, a credibility fusion calculation is performed based on the statistical results to determine whether an extreme road condition warning event should be generated. If generated, the extreme road condition warning event is pushed to the navigation system so that the navigation system can provide the extreme road condition warning event to subsequent vehicles whose planned routes include the current road segment.
[0082] The various embodiments in this application are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the embodiments for IoT devices and media are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions of the method embodiments.
[0083] The systems, media, and methods provided in this application are one-to-one correspondences. Therefore, the systems and media also have similar beneficial technical effects as their corresponding methods. Since the beneficial technical effects of the methods have been described in detail above, the beneficial technical effects of the systems and media will not be repeated here.
[0084] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.
[0085] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.
[0086] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.
[0087] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.
[0088] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.
[0089] Memory may include non-persistent storage in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.
[0090] Computer-readable media include both permanent and non-permanent, removable and non-removable media that can store information by any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.
[0091] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.
[0092] The above description is merely an embodiment of this application and is not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of the claims of this application.
Claims
1. A road condition warning method characterized by comprising: The method comprises: obtaining multi-modal perception data of a driving vehicle on a current road section, and identifying whether an extreme road condition exists on the current road section; if so, uploading the identification result to a cloud service platform, and generating a reporting and statistical task corresponding to the current road section in the cloud service platform based on the identification result; after the reporting and statistical task is executed, performing a credibility fusion calculation based on the statistical result to determine whether to generate an extreme road condition warning event; if so, pushing the extreme road condition warning event to a navigation system to provide the extreme road condition warning event to a subsequent vehicle planning a path including the current road section through the navigation system.
2. The road condition warning method according to claim 1, wherein The multi-modal perception data comprises road surface image data, vehicle vibration data, and vehicle speed data; The identification of whether an extreme road condition exists on the current road section specifically comprises: performing initial road condition category determination on the road surface image data based on an image recognition model in a vehicle-mounted edge computing module to obtain an initial road condition determination result; wherein the initial road condition determination result comprises an initial road condition determination category and a category confidence level; determining whether the initial road condition determination category is a preset extreme road condition type and whether the category confidence level exceeds a preset image confidence level threshold; if so, triggering a collaborative analysis of the vehicle vibration data and the vehicle speed data, and performing data fusion determination based on a collaborative analysis result and the category confidence level to determine whether the current road condition is an extreme road condition and a road condition type.
3. The road condition warning method according to claim 2, wherein The triggering of the collaborative analysis of the vehicle vibration data and the vehicle speed data specifically comprises: processing a vertical acceleration time series signal in the vehicle vibration data based on a preset impact peak value detection algorithm to determine a peak acceleration within a first preset time window from a time when the initial road condition determination result is generated, and determining whether the peak acceleration exceeds a preset vibration intensity threshold and a duration thereof is shorter than a preset short time threshold to generate a vibration event detection result; extracting a continuous vehicle speed frame sequence within a second preset time window from the time when the initial road condition determination result is generated in the vehicle speed data, and calculating an average vehicle speed drop rate of the driving vehicle within the second preset time window based on the continuous vehicle speed frame sequence to generate a vehicle speed drop detection result; integrating the vibration event detection result and the vehicle speed drop detection result to obtain the collaborative analysis result.
4. The road condition warning method according to claim 3, wherein The data fusion determination based on the collaborative analysis result and the category confidence level specifically comprises: matching a preset fusion determination rule corresponding to the initial road condition determination category based on the initial road condition determination category; wherein the preset fusion determination rule comprises a collaborative analysis condition required for a corresponding extreme road condition type; inputting the collaborative analysis result into the selected fusion determination rule to determine whether the vertical acceleration peak value satisfies a vibration threshold condition and / or whether the vehicle speed drop rate satisfies a vehicle speed drop threshold condition; if so, determining that the current road section has the extreme road condition type corresponding to the initial road condition determination category.
5. The road condition warning method of claim 1, wherein, The generation of the reporting and statistical task corresponding to the current road section in the cloud service platform based on the identification result specifically comprises: extracting position information of the current road segment contained in the recognition result; based on the position information, searching in a task management database of the cloud service platform to determine whether there is a target statistical task in an open state and matching the position information; if not, creating a statistical task as the target statistical task and configuring a preset time length of task execution time window for the target statistical task; storing the recognition result as a reporting result record and associating with the target statistical task.
6. The road condition warning method of claim 1, wherein, based on the statistical result, performing credibility fusion calculation to determine whether to generate an extreme road condition warning event, specifically including: receiving multiple reporting results reported by multiple vehicles on the current road segment within the reporting statistical task execution time window; performing spatio-temporal clustering analysis on the multiple reporting results to aggregate reporting results with the same extreme road condition type and adjacent geographical position into a group event; based on the number of reporting results in the group event, the category confidence contained in each reporting result, and the reporting time of the reporting result, calculating the comprehensive credibility score of the group event and determining whether the comprehensive credibility score exceeds a preset credibility threshold; if yes, generating the extreme road condition warning event.
7. The road condition warning method of claim 1, wherein, pushing the extreme road condition warning event to the navigation system, specifically including: encapsulating the position information and road condition type information of the current road segment contained in the extreme road condition warning event into a data packet conforming to a preset navigation application programming interface protocol; sending the data packet to the cloud server of the navigation system to enable the cloud server to associate the extreme road condition warning event with the corresponding current road segment in the electronic map database.
8. The road condition warning method according to claim 7, wherein providing the extreme road condition warning event for subsequent vehicles planning a path containing the current road segment through the navigation system, specifically including: in response to a path planning request or real-time navigation request initiated by the subsequent vehicle to the navigation system, determining whether the subsequent vehicle will approach the current road segment within a preset warning distance; if yes, rendering a warning icon and text corresponding to the road condition type information in the vehicle-mounted display interface of the subsequent vehicle and playing voice warning content corresponding to the road condition type information through the vehicle-mounted audio.
9. A vehicle characterized by comprising: The vehicle includes: at least one processor; and a memory in communication connection with the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the road condition warning method of any one of claims 1-8.
10. A computer storage medium storing computer-executable instructions, which, when executed by a processor, cause the processor to perform acts comprising: The computer executable instructions are executed to implement the road condition warning method of any one of claims 1-8. The computer executable instructions are executed to implement the road condition warning method of any one of claims 1-8.