A vehicle intelligent monitoring and early warning method and system

Abstract

This application provides a vehicle intelligent monitoring and early warning method and system, relating to the field of vehicle intelligent monitoring and early warning technology. It continuously acquires the working status information of the vehicle's main sensing devices to identify perception-limited situations. When a perception-limited situation is identified, auxiliary sensing signals are collected to generate a perception reliability depreciation index. Sensing information and kinematic features are extracted from the auxiliary sensing signals and matched with preset high-risk entity behavior rules to identify core risk targets. Based on the perception reliability depreciation index, sensing information, and the vehicle's own operating status, a comprehensive collision risk level is calculated, vehicle safety operation guidance parameters are generated, core risk targets are tracked, and safe operation decisions are made. The perception-limited situations, perception reliability depreciation index, and safe operation guidance parameters are structured and stored, and a multi-dimensional index is constructed to enable post-event review and traceability, improving the vehicle's safe operation capability in complex dynamic environments.

Description

Technical Field

[0001] This application relates to the field of intelligent vehicle monitoring and early warning technology, and more specifically, to a method and system for intelligent vehicle monitoring and early warning. Background Technology

[0002] In modern industrial production and logistics, specialized vehicles, especially driverless logistics shuttles, are playing an increasingly important role in highly automated smart manufacturing parks. To ensure the safe and efficient operation of these vehicles, they are typically equipped with a variety of advanced sensing devices, such as lidar, high-definition cameras, millimeter-wave radar, and ultrasonic sensors, as well as global navigation satellite systems and inertial measurement units, to achieve comprehensive environmental perception and precise monitoring of their own status. The massive amounts of data collected by these devices are initially processed by the onboard computing platform and then transmitted to the park's central monitoring and dispatch center for storage and analysis, thereby supporting real-time vehicle operation monitoring and risk assessment. However, in actual operation, the park environment is not static. Sudden changes in the local environment, such as temporary obstructions caused by production line adjustments or steam clouds generated by accidental leaks, often have a significant impact on the vehicle's sensing system, leading to a series of complex safety challenges.

[0003] Current technology suffers from a critical flaw: when special vehicles face sudden local environmental changes (such as vapor cloud obstruction) that cause severe performance degradation or failure of some key sensing devices (such as cameras and lidar), existing monitoring and early warning systems, while faithfully recording all multi-source heterogeneous data (including low-confidence outputs from failed sensing devices, ambiguous signals from redundant sensing devices, and vehicle internal status data), cannot intelligently identify and quantify this "perception-limited" situation caused by environmental changes in real-time. Furthermore, the system cannot dynamically adjust the importance allocation and fusion strategies of different data sources based on this perception-limited situation, thus failing to rapidly and deeply logically correlate and infer fragmented information from different sensing devices (such as ambiguous target information from millimeter-wave radar and attitude changes from inertial measurement units), information that might normally be considered low-priority or irrelevant. This makes it difficult for the system to synthesize a high-confidence, unified "collision risk" assessment in real-time from these incomplete, uncertain, and interconnected data at critical moments, thereby affecting the vehicle's immediate decision-making capabilities.

[0004] Furthermore, in post-event debriefing, the lack of such contextualized data association and causal chain construction mechanisms makes it difficult for analysts to efficiently and accurately reconstruct the dynamic complementary and conflicting relationships between data from different sensing devices at the time of the event from massive amounts of fragmented data, and how these relationships affected the vehicle's perception, decision-making, and final behavior. This makes it difficult to effectively identify system defects and implement targeted improvements. Therefore, the current technical challenge lies in establishing a new data organization, fusion, and reasoning mechanism that can proactively perceive and understand complex operating scenarios (especially those with limited perception), and on this basis, intelligently associate and integrate multi-source heterogeneous, and even partially faulty, data. This would support special vehicles in conducting high-confidence real-time risk assessments and decisions in complex dynamic environments, and achieve efficient and accurate causal chain backtracking for complex events. Summary of the Invention

[0005] This application provides a vehicle intelligent monitoring and early warning method and system, which aims to solve the technical problems in the prior art where special vehicles in complex dynamic environments face sudden local environmental changes that cause the performance of the main sensing devices to decline or fail. These problems make it difficult to intelligently identify and quantify the "perception limitation" situation, dynamically adjust the importance allocation and fusion strategy of data sources, and thus make it difficult to synthesize a unified "collision risk" assessment with high confidence in real time, affecting the vehicle's immediate decision-making ability. Furthermore, it is difficult to efficiently reconstruct the causal chain of events after the fact.

[0006] On the one hand, this application provides a vehicle intelligent monitoring and early warning method, including: The system continuously acquires the working status information of the vehicle's main sensing devices, determines whether the environmental perception performance of the main sensing devices has dropped to a preset performance threshold, and thus identifies perception-restricted situations caused by sudden obstructions in the park. The perception-restricted situation is a situation where the main sensing devices cannot achieve accurate environmental perception due to obstructions. When the perception-limited situation is identified, auxiliary perception signals are collected through auxiliary perception devices, and the reliability of the auxiliary perception signals is quantitatively evaluated to generate a perception reliability loss index. The system extracts perception information related to vehicle safety operation and kinematic features of the detected targets from the auxiliary perception signals. It then matches the kinematic features with preset high-risk entity behavior rules and identifies detected targets with a matching degree higher than a preset matching degree threshold as core risk targets. The perception information related to vehicle safety operation includes the relative distance and relative speed between the detected targets and the vehicle, as well as the location of obstacles. Based on the perceived reliability loss index, the perceived information and the vehicle's own operating status, a comprehensive collision risk level is calculated. Based on the comprehensive collision risk level, vehicle safety operation guidance parameters are generated. The core risk targets are continuously tracked, and vehicle safety operation decisions are made. The safety operation decisions include at least one of deceleration, avoidance and emergency braking. The perception-limited situation, the auxiliary perception signal, the perception reliability loss index, the calculation process of the comprehensive collision risk level, and the safety operation guidance parameters are structured, recorded, packaged into a contextualized event package, and stored. At the same time, a multi-dimensional index is constructed with perception-limited situation type, perception equipment status, and safety operation decision as causal correlation dimensions to achieve post-event review and traceability.

[0007] Optionally, the step of performing a reliability quantification assessment of the auxiliary sensing signal to generate a sensing reliability depreciation index includes: The perception reliability loss index is generated by quantifying at least one of the following: signal-to-noise ratio of the auxiliary sensing signal, standard deviation of distance measurement, standard deviation of velocity measurement, and fluctuation amplitude of target features. The perception reliability loss index is used to calculate the comprehensive collision risk level.

[0008] Optionally, the target feature fluctuation amplitude is determined by tracking the changes in the movement speed, physical contour size, and position of the detected target relative to the vehicle in consecutive frames, and calculating the fluctuation range of each change within a preset time period.

[0009] Optionally, the safe operating guidance parameters include the maximum permissible driving speed and the minimum safe braking distance. The maximum permissible driving speed decreases as the comprehensive collision risk level increases, while the minimum safe braking distance increases as the comprehensive collision risk level increases.

[0010] Optionally, the multidimensional index includes at least one field among scenario type, primary failure sensor, auxiliary effective sensor, and target type; wherein, the auxiliary effective sensor refers to an auxiliary sensing device that works normally and collects auxiliary sensing signals in a perception-limited scenario, and the target type refers to the entity type of the detected target, including forklifts, heavy trucks, workers, and stacks of goods.

[0011] Optionally, the preset high-risk entity behavior rules include the typical kinematic parameter range of high-risk vehicles in the park. The typical kinematic parameter range includes at least one of the physical contour size range, speed range, acceleration range, and turning radius range. The matching degree is calculated by weighted sum of physical contour size matching degree, speed matching degree, acceleration matching degree, and trajectory matching degree.

[0012] Optionally, the comprehensive collision risk level is calculated by fusing the perception reliability loss index, the basic collision risk value corresponding to the perception information, the environmental risk factor, and the vehicle state risk factor. The basic collision risk value is calculated based on the relative distance and relative speed between the detected target and the vehicle. The environmental risk factor is determined based on the degree of danger of the area where the vehicle is currently located. The area is divided into high-risk, medium-risk, and low-risk areas according to the degree of danger. High-risk areas include intersections and loading / unloading areas, medium-risk areas include the main roads of the area, and low-risk areas include parking lots. The vehicle state risk factor is determined based on the vehicle's own operating state, which includes the vehicle's current speed, acceleration, and braking performance parameters. The braking performance parameters include braking response time and maximum braking deceleration. The comprehensive collision risk level is calculated by multiplying the perception reliability loss index, the basic collision risk value, the environmental risk factor, and the vehicle state risk factor by preset weighting coefficients and then summing them.

[0013] Optionally, the contextualized event package is stored in a serializable file format or a database that supports complex document structures. The serializable file format is JSON or Protocol Buffers, and the database that supports complex document structures is a NoSQL database.

[0014] Optionally, the step of continuously tracking the core risk target and making vehicle safety operation decisions includes: obtaining the instantaneous position, speed and trajectory of the core risk target in real time by fusing auxiliary sensing signals, vehicle self-positioning information and attitude information; Predict the subsequent movement trend of the core risk target within a preset time range, and formulate vehicle safety operation decisions based on the comprehensive collision risk level and the vehicle safety operation guidance parameters.

[0015] On the other hand, this application provides a vehicle intelligent monitoring and early warning system, which includes: The scenario judgment module is used to continuously acquire the working status information of the vehicle's main sensing devices, determine whether the environmental perception performance of the main sensing devices has dropped to a preset performance threshold, and thus identify the perception-restricted scenario caused by sudden obstructions in the park. The perception-restricted scenario is a scenario in which the main sensing devices cannot achieve accurate environmental perception due to obstructions. The index calculation module is used to collect auxiliary sensing signals through auxiliary sensing devices when the perception-limited situation is identified, and to perform a reliability quantification assessment of the auxiliary sensing signals to generate a perception reliability loss index. The core risk target identification module is used to extract perception information related to vehicle safe operation and kinematic features of detection targets from the auxiliary perception signals, match the kinematic features with preset high-risk entity behavior rules, and identify detection targets with a matching degree higher than a preset matching degree threshold as core risk targets. The risk assessment and decision-making module is used to calculate the comprehensive collision risk level based on the perceived reliability loss index, the perceived information and the vehicle's own operating status, generate vehicle safety operation guidance parameters based on the comprehensive collision risk level, continuously track the core risk targets, and formulate vehicle safety operation decisions, the safety operation decisions including at least one of deceleration, avoidance and emergency braking. The storage and retrieval module is used to structurally record and package the perception-limited situation, the auxiliary perception signal, the perception reliability loss index, the calculation process of the comprehensive collision risk level, and the safety operation guidance parameters into a contextualized event package and store it. At the same time, it constructs a multi-dimensional index with perception-limited situation type, perception equipment status, and safety operation decision as causal correlation dimensions to realize post-event review and traceability.

[0016] The vehicle intelligent monitoring and early warning method and system disclosed in this application continuously acquires the working status information of the vehicle's main sensing devices, intelligently determines whether their environmental perception performance has decreased to a preset threshold, and thus identifies perception-limited situations caused by sudden obstructions within the park. When a perception-limited situation is identified, the method can collect auxiliary sensing signals through auxiliary sensing devices and perform reliability quantification assessment to generate a perception reliability depreciation index. Based on this, it extracts perception information related to vehicle safe operation and kinematic features of detected targets from the auxiliary sensing signals, and matches them with preset high-risk entity behavior rules, identifying detected targets with a matching degree higher than a preset threshold as core risk targets. Subsequently, based on the perception reliability depreciation index, perception information, and the vehicle's own operating status, it calculates the comprehensive collision risk level and generates vehicle safety operation guidance parameters, continuously tracks core risk targets, and formulates safe operation decisions including deceleration, avoidance, and emergency braking. Furthermore, this method also structurally records and packages the perception-limited situation, auxiliary perception signals, perception reliability depreciation index, calculation process of comprehensive collision risk level, and safety operation guidance parameters into a contextualized event package for storage. Simultaneously, it constructs a multi-dimensional index with perception-limited situation type, perception equipment status, and safety operation decisions as causal correlation dimensions to enable post-event review and traceability. Through the above technical solution, this application effectively solves the technical problems in the prior art where special vehicles, facing sudden local environmental changes leading to performance degradation or failure of main perception equipment, struggle to intelligently identify and quantify "perception-limited" situations, cannot dynamically adjust data source importance allocation and fusion strategies, thus making it difficult to synthesize a high-confidence unified "collision risk" assessment in real time, affecting the vehicle's immediate decision-making ability, and making it difficult to efficiently reconstruct the causal chain of events during post-event review. This application, by introducing a perception reliability depreciation index, achieves a quantitative assessment of the reliability of auxiliary perception signals, enabling the system to more accurately utilize auxiliary information for risk assessment when main perception equipment is limited. Meanwhile, by identifying core risk targets and calculating the comprehensive collision risk level by combining multi-dimensional risk factors, this application can generate more targeted safety operation guidance parameters and operational decisions, significantly improving the vehicle's safe operation capability in complex dynamic environments. Furthermore, the storage of contextualized event packages and the construction of multi-dimensional indexes provide an efficient and accurate causal chain backtracking mechanism for post-event review, greatly facilitating the identification of system defects and targeted improvements. This overcomes the shortcomings of existing technologies, such as data fragmentation and difficulty in contextualized association and reasoning, achieving significant progress and excellent technical results. Attached Figure Description

[0017] To illustrate this application more clearly, the accompanying drawings used in the embodiments will be briefly described below. Obviously, those skilled in the art can obtain other drawings based on these drawings without any creative effort.

[0018] Figure 1 The diagram above illustrates a flowchart of a vehicle intelligent monitoring and early warning method. Figure 2 The diagram above illustrates a structural schematic of a vehicle intelligent monitoring and early warning system.

[0019] Attached reference numerals: 100, Intelligent vehicle monitoring and early warning system; 10, Context judgment module; 20, Index calculation module; 30, Core risk target identification module; 40, Risk assessment and decision-making module; 50, Storage and retrieval module. Detailed Implementation

[0020] The technical solutions of this application will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of this application, and not all embodiments. The components of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0021] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this application, terms such as "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0022] Traditional vehicle monitoring and early warning systems often struggle to intelligently identify and quantify the resulting perception limitations when faced with sudden environmental changes within a park (such as vapor clouds or temporary obstructions). They also cannot dynamically adjust data fusion strategies, making it difficult to synthesize high-confidence collision risk assessments in real time from incomplete, uncertain, and heterogeneous multi-source data, thus impacting the vehicle's immediate decision-making capabilities. Furthermore, in post-event analysis, the lack of contextualized data correlation and causal chain construction mechanisms makes it difficult for analysts to efficiently reconstruct the dynamic complementarity and conflict relationships between data from different sensing devices at the time of the event, hindering the effective identification of system defects and targeted improvements.

[0023] like Figure 1 The diagram illustrates a flowchart of a vehicle intelligent monitoring and early warning method. This application proposes a vehicle intelligent monitoring and early warning method, comprising: S10, continuously acquire the working status information of the vehicle's main sensing devices, determine whether the environmental perception performance of the main sensing devices has dropped to a preset performance threshold, thereby identifying the perception-restricted situation caused by sudden obstructions in the park, wherein the perception-restricted situation is the situation in which the main sensing devices cannot achieve accurate environmental perception due to obstructions. The primary sensing devices typically refer to the core sensors on a vehicle used for environmental perception, such as LiDAR, high-definition cameras, and millimeter-wave radar. Under normal operating conditions, they provide high-precision, high-confidence environmental perception data. The preset performance threshold refers to the minimum standard that the environmental perception performance (such as detection distance, accuracy, and resolution) of the primary sensing devices should achieve under normal operating conditions. When the actual performance falls below this threshold, its sensing capability is considered to be affected. Perception-limited scenarios refer to situations where sudden obstructions within the park (such as temporarily stored goods, sudden vapor clouds, dust, etc.) prevent the primary sensing devices from achieving accurate environmental perception.

[0024] S20, when the perception-limited situation is identified, auxiliary perception signals are collected through auxiliary perception devices, and the reliability of the auxiliary perception signals is quantitatively evaluated to generate a perception reliability loss index; Auxiliary sensing devices refer to sensors that can provide supplementary or alternative sensing capabilities when the performance of the main sensing devices is limited. Examples include ultrasonic sensors and inertial measurement units (IMUs). They may provide information that, while not comprehensive, is still valuable in specific situations. The sensing reliability degradation index is a quantitative indicator used to assess the reliability of auxiliary sensing signals in sensing-limited situations. The higher the index, the lower the reliability of the auxiliary sensing signal, and vice versa.

[0025] S30, extract perception information related to vehicle safe operation and kinematic features of the detected target from the auxiliary perception signal, match the kinematic features with preset high-risk entity behavior rules, and take the detected target with a matching degree higher than the preset matching degree threshold as the core risk target; wherein, the perception information related to vehicle safe operation includes the relative distance and relative speed between the detected target and the vehicle, and the position of obstacles; Among them, the core risk targets refer to entities that pose a high risk to the safe operation of vehicles by analyzing auxiliary sensing signals in situations where perception is limited, such as forklifts traveling at high speeds or workers suddenly appearing.

[0026] S40, based on the perceived reliability loss index, the perceived information and the vehicle's own operating status, calculate the comprehensive collision risk level, generate vehicle safety operation guidance parameters based on the comprehensive collision risk level, continuously track the core risk targets, and formulate vehicle safety operation decisions, the safety operation decisions including at least one of deceleration, avoidance and emergency braking; The comprehensive collision risk level is a quantitative assessment of the collision risk currently faced by a vehicle, taking into account factors such as the perceived reliability impairment index, perceived information, and the vehicle's own operating status. Vehicle safety operation guidance parameters are specific operational suggestions provided to the vehicle based on the comprehensive collision risk level, such as maximum permissible speed and minimum safe braking distance.

[0027] Safe operation decisions are the operational instructions made by the vehicle autonomously or in assistance to the driver based on the comprehensive collision risk level and safe operation guidance parameters, including deceleration, avoidance, and emergency braking.

[0028] S50, the perception-limited situation, the auxiliary perception signal, the perception reliability loss index, the calculation process of the comprehensive collision risk level, and the safety operation guidance parameters are structured, recorded, packaged into a contextualized event package, and stored. At the same time, a multi-dimensional index is constructed with perception-limited situation type, perception equipment status, and safety operation decision as causal correlation dimensions to achieve post-event review and traceability.

[0029] Among them, the contextualized event package is a structured record and package of relevant data (such as auxiliary sensing signals, sensing reliability loss index, risk assessment process, decision parameters, etc.) when a perception-limited situation occurs, so as to facilitate subsequent storage and analysis.

[0030] Multidimensional indexing is a data organization method that establishes causal relationships between multiple dimensions, such as the type of perception-limited context, the status of perception devices, and safe operation decisions, to achieve efficient retrieval and retrospective analysis of contextualized event packages.

[0031] The implementation environment of this application is typically a special vehicle within a smart manufacturing park, such as an unmanned logistics shuttle, which is equipped with various sensing devices and an onboard computing platform, and interacts with the park's central monitoring and dispatch center for data exchange.

[0032] This application provides a vehicle intelligent monitoring and early warning method, the main feature of which is that, firstly, the method continuously acquires the working status information of the vehicle's main sensing devices and determines whether their environmental perception performance has decreased to a preset performance threshold. This step can be achieved in several ways. For example, the working status can be evaluated by periodically reading the diagnostic data inside the main sensing devices, such as error codes and self-test status. Another way is to analyze the quality of the environmental perception data output by the main sensing devices, for example, by detecting the sparsity of the LiDAR point cloud, the clarity of the camera image, or the presence of large areas of blurriness, to determine whether their performance has degraded. When the environmental perception performance of the main sensing devices is detected to have decreased to the preset performance threshold, it is possible to identify a perception-limited situation caused by sudden obstructions within the area. For example, when a vehicle drives into an area where a large amount of goods are temporarily piled up, resulting in a large amount of missing LiDAR point cloud data, or when the camera image becomes blurry due to being obscured by steam clouds, it can be determined that the current situation is a perception-limited situation.

[0033] When a perception-limited situation is identified, this application acquires auxiliary sensing signals through auxiliary sensing devices and performs a reliability quantification assessment of the auxiliary sensing signals to generate a perception reliability depreciation index. The acquisition methods of the auxiliary sensing devices can be diverse. For example, ultrasonic sensors can be used to detect nearby obstacles, providing auxiliary sensing signals. Inertial measurement units (IMUs) can provide the vehicle's own attitude and motion information, indirectly assisting in judging environmental changes. Various methods can be used for the reliability quantification assessment of auxiliary sensing signals. For example, reliability can be evaluated by analyzing the stability, consistency, or cross-validation results with data from other vehicle sensors. For instance, when a millimeter-wave radar detects a target in a perception-limited situation, but its signal strength fluctuates significantly or the target's position changes frequently, its reliability can be considered low, thus generating a high perception reliability depreciation index.

[0034] Furthermore, this application extracts perception information related to vehicle safety operation and kinematic features of the detected target from the auxiliary perception signals. The perception information related to vehicle safety operation includes the relative distance and speed between the detected target and the vehicle, and the location of obstacles. This information can be obtained by processing and parsing the auxiliary perception signals. For example, millimeter-wave radar can provide the relative distance and speed information of the target. Ultrasonic sensors can provide the location information of nearby obstacles. Simultaneously, the kinematic features of the detected target, such as its speed, acceleration, and turning radius, can also be extracted from the auxiliary perception signals. For example, the instantaneous speed and acceleration of the target can be calculated using continuous frames of millimeter-wave radar data. Subsequently, these kinematic features are matched against preset high-risk entity behavior rules. These preset high-risk entity behavior rules can be predefined, for example, specifying the typical high-speed driving pattern of forklifts in a specific area of ​​the park. When the matching degree between the kinematic features of the detected target and these rules is higher than a preset matching degree threshold, the detected target is identified as a core risk target. For example, if a detection target is identified as a forklift, and its speed and acceleration both exceed the typical range for safe driving within the park, then the forklift will be marked as a core risk target.

[0035] Based on this, this application calculates the comprehensive collision risk level based on the perception reliability degradation index, perception information, and the vehicle's own operating status. The vehicle's own operating status includes the vehicle's current speed, acceleration, and braking performance parameters. The comprehensive collision risk level can be calculated using a weighted summation method, integrating various factors. For example, when the perception reliability degradation index is high (indicating unreliable auxiliary perception signals), the relative distance between the detected target and the vehicle is close, the relative speed is high, and the vehicle's own speed is high, the comprehensive collision risk level will increase accordingly. Based on the calculated comprehensive collision risk level, vehicle safety operation guidance parameters will be generated. For example, when the comprehensive collision risk level is high, guidance parameters such as "maximum permissible speed reduced to 5 km / h" or "minimum safe braking distance increased to 5 meters" can be generated. Simultaneously, this application continuously tracks core risk targets and formulates vehicle safety operation decisions. Safety operation decisions include at least one of deceleration, avoidance, and emergency braking. For example, if the core risk target suddenly changes its trajectory and the comprehensive collision risk level continues to increase, an "emergency braking" decision may be made.

[0036] Finally, this application structurally records the calculation process of the perception-limited situation, auxiliary perception signals, perception reliability depreciation index, comprehensive collision risk level, and safety operation guidance parameters, packages them into contextualized event packages, and stores them. The contextualized event packages can be stored in various ways; for example, they can be stored as serializable file formats, such as JSON, or in databases that support complex document structures, such as NoSQL databases. Simultaneously, this application constructs a multi-dimensional index with perception-limited situation type, perception device status, and safety operation decisions as causal correlation dimensions. The construction of the multi-dimensional index can employ relational databases or graph databases. For example, an index can be created that allows for quick retrieval of all event packages matching the criteria by querying events where "perception-limited situation type is 'vapor cloud obstruction,' 'the main failure sensor is lidar,' and the final 'safety operation decision is emergency braking,'" thereby enabling post-event review and traceability.

[0037] The vehicle intelligent monitoring and early warning method of this application addresses the challenges existing in the prior art through a series of closely related steps working together.

[0038] First, by continuously acquiring the operational status information of the vehicle's main sensing devices and assessing their environmental perception performance, the system can proactively and in real-time identify perception-limited situations caused by sudden obstructions within the park. This mechanism allows the vehicle to move beyond passively waiting for the main sensing devices to completely fail and instead provide early warnings and enter emergency response mode. For example, when the point cloud data from the LiDAR begins to show localized sparseness or the camera image becomes blurry, it can be determined that perception performance has deteriorated, thereby identifying potential perception-limited situations.

[0039] Once a perception-limited situation is identified, auxiliary sensing equipment is immediately activated to collect auxiliary sensing signals. While these auxiliary sensing signals may not be as comprehensive as the data from the primary sensing equipment under normal conditions, they provide crucial supplementary information in perception-limited situations. For example, millimeter-wave radar has good penetration in fog or steam, providing basic distance and velocity information of targets. Subsequently, the reliability of these auxiliary sensing signals is quantitatively assessed, generating a perception reliability degradation index. This index is one of the core innovations of this application; it quantifies the credibility of the auxiliary sensing signals in the current situation, providing an important weighting factor for subsequent risk assessment. For example, if the millimeter-wave radar signal is subject to strong interference, its perception reliability degradation index will increase, indicating a decrease in its data reference value.

[0040] Next, sensory information relevant to vehicle safety (such as relative distance, relative speed, and obstacle position) and kinematic characteristics of detected targets are extracted from the auxiliary sensing signals. This information forms the basis for collision risk assessment. By matching the kinematic characteristics with pre-defined high-risk entity behavior rules, core risk targets posing a threat to vehicle safety can be accurately identified. For example, a forklift approaching at high speed, whose kinematic characteristics match the high-risk rules, will be identified as a core risk target.

[0041] Based on the perceived reliability degradation index, perceived information, and the vehicle's own operating status, a comprehensive collision risk level is calculated. The introduction of the perceived reliability degradation index means that risk assessment no longer relies solely on the surface data of perceived information, but considers the reliability of the data itself, thereby improving the accuracy and robustness of risk assessment. For example, even if perceived information indicates a close target, a high perceived reliability degradation index would indicate a higher risk because the perceived data itself contains uncertainty. Based on the comprehensive collision risk level, vehicle safety operation guidance parameters are generated, and core risk targets are continuously tracked, ultimately leading to safe operating decisions, including deceleration, avoidance, or emergency braking. This decision-making process is dynamic and real-time, capable of adjusting to constantly changing perceived scenarios and risk levels.

[0042] Finally, this application structurally records and packages all relevant information, including perception-limited scenarios, auxiliary perception signals, perception reliability degradation index, calculation processes for comprehensive collision risk levels, and safe operation guidance parameters, into a contextualized event package for storage. Simultaneously, a multi-dimensional index is constructed with perception-limited scenario type, perception device status, and safe operation decisions as causal dimensions. This mechanism not only provides a detailed data foundation for post-event review and traceability, but more importantly, through the multi-dimensional index, analysts can efficiently reconstruct the dynamic complementary and conflicting relationships between data from different perception devices at the time of the event, and how these relationships affected the vehicle's perception, decision-making, and final behavior. This enables accurate identification of defects and targeted improvements.

[0043] The vehicle intelligent monitoring and early warning method of this application demonstrates significant innovation and advantages in solving existing technical problems.

[0044] Existing technologies often struggle to intelligently identify and quantify situations where perception is limited due to sudden changes in the local environment, and they also cannot dynamically adjust data fusion strategies, resulting in limited accuracy and decision-making capabilities in real-time risk assessment. In contrast, the core innovation of this application lies in introducing a quantitative indicator, the "perception reliability degradation index," and integrating it into the calculation of the comprehensive collision risk level. This mechanism enables proactive perception and understanding of complex operating situations, especially when the performance of primary sensing devices degrades. It allows for dynamic assessment of the reliability of auxiliary sensing signals, thereby synthesizing a high-confidence collision risk assessment in real time from incomplete, uncertain, multi-source heterogeneous data. For example, when vapor cloud obstruction causes camera and lidar failure, traditional systems may fail to provide effective warnings due to data loss. However, this application, through auxiliary sensing devices such as millimeter-wave radar, combined with the perception reliability degradation index, can still identify and assess core risk targets and generate corresponding safety operation guidance parameters, such as deceleration or avoidance, thus significantly improving the vehicle's safe operation capability in complex environments.

[0045] Furthermore, this application structures and packages key information such as perception-limited situations, auxiliary perception signals, perception reliability depreciation index, calculation process of comprehensive collision risk level, and safe operation guidance parameters into contextualized event packages. It also constructs a multi-dimensional index with perception-limited situation type, perception equipment status, and safe operation decisions as causal correlation dimensions, significantly improving the efficiency and accuracy of post-event review and tracing. While traditional systems also record multi-source heterogeneous data, they lack contextualized data association and causal chain construction mechanisms, making it difficult for analysts to efficiently reconstruct the truth of events from massive amounts of fragmented data. The technical solution of this application, however, can quickly locate specific types of perception-limited events through multi-dimensional indexing and trace their complete causal chain from perception to decision-making, thereby effectively identifying defects and making targeted improvements. This contextualized data organization and tracing capability is not possessed by existing technologies, providing a solid technical guarantee for the safe operation of special vehicles in complex dynamic environments.

[0046] In some embodiments, the step of performing a reliability quantification assessment of the auxiliary sensing signal to generate a sensing reliability loss index includes: The perception reliability loss index is generated by quantifying at least one of the following: signal-to-noise ratio of the auxiliary sensing signal, standard deviation of distance measurement, standard deviation of velocity measurement, and fluctuation amplitude of target features. The perception reliability loss index is used to calculate the comprehensive collision risk level.

[0047] The signal-to-noise ratio (SNR) can be understood as the ratio of effective signal to noise in the auxiliary sensing signal. A higher SNR indicates better signal quality and higher sensing reliability. The standard deviations of distance and speed measurements reflect the accuracy and stability of the auxiliary sensing device when measuring target distance and speed, respectively. Smaller standard deviations indicate more reliable measurement results. In practical applications, target feature fluctuation amplitude specifically refers to the range of fluctuations in the target's speed, physical contour size, and position relative to the vehicle, calculated within a preset time period by tracking continuous frames. A smaller fluctuation range indicates better target tracking stability and higher consistency of sensing data. By calculating at least one of these indicators, the quality and reliability of the auxiliary sensing signal can be quantitatively evaluated, generating a sensing reliability depreciation index. This index will serve as an important input parameter for subsequent comprehensive collision risk level calculations.

[0048] The aforementioned technical solutions enable a refined and quantitative assessment of the reliability of auxiliary sensing signals, resulting in a more objective and accurate perception reliability degradation index. This helps to more accurately reflect the true reliability of the data provided by auxiliary sensing devices in perception-limited scenarios, thus providing a more solid and reliable data foundation for subsequent comprehensive collision risk level calculations. Consequently, the vehicle intelligent monitoring and early warning system can more effectively cope with complex park environments, improving the accuracy of risk assessment and the safety of decision-making.

[0049] In some embodiments, the target feature fluctuation amplitude is determined by tracking the changes in the movement speed, physical contour size, and position of the detected target relative to the vehicle in consecutive frames, and calculating the fluctuation range of each change within a preset time period.

[0050] Specifically, during the continuous acquisition of auxiliary sensing signals by the auxiliary sensing device, the detected targets identified in the signals are continuously tracked. Tracking the detected targets identified in continuous frames involves continuously identifying and locating the same physical entity target in consecutive video frames or radar / LiDAR scan frames. Through this continuous tracking, the target's velocity, physical outline dimensions, and position information relative to the vehicle at different time points can be obtained. Velocity refers to the instantaneous velocity vector of the target relative to the vehicle; physical outline dimensions refer to the target's length, width, height, or its projected dimensions on the sensing plane; and position refers to the target's three-dimensional coordinates in the vehicle coordinate system. These variations reflect the dynamics of the target's motion and the uncertainty of the sensing data. The preset time can be set according to the actual application scenario and the real-time requirements for sensing reliability assessment; for example, it can be set to 0.5 seconds, 1 second, or longer. Within the preset time, data on these variations are collected, and their respective fluctuation ranges are calculated, for example, by calculating statistical measures such as the difference between the maximum and minimum values, standard deviation, or variance to characterize the fluctuation range.

[0051] Through the above technical solution, the calculation of the perception reliability degradation index no longer relies solely on the quality of the signal itself (such as signal-to-noise ratio), but further incorporates the stability assessment of the dynamic characteristics of the detected target. This method can more comprehensively and accurately assess the reliability of auxiliary perception signals, especially when the target's movement is complex or the environment is dynamically changing. It can promptly detect the instability and uncertainty of perception data, thereby making the calculation of the comprehensive collision risk level more accurate and providing a more reliable basis for vehicle safety operation decisions.

[0052] In some embodiments, the safe operating guidance parameters include a maximum permissible driving speed and a minimum safe braking distance, wherein the maximum permissible driving speed decreases as the overall collision risk level increases, and the minimum safe braking distance increases as the overall collision risk level increases.

[0053] Specifically, safe operating guidance parameters refer to specific operational suggestions or restrictions provided to the vehicle control system or driver in perception-limited situations to ensure safe vehicle operation. These parameters aim to adjust the vehicle's operating state within a safe range to avoid potential collision risks. Among them, the maximum permissible speed refers to the highest speed at which the vehicle is allowed to travel under the current comprehensive collision risk level, with the aim of reducing the energy and severity of a collision by limiting vehicle speed. The minimum safe braking distance refers to the minimum distance the vehicle needs to maintain from an obstacle or core risk target ahead in an emergency under the current comprehensive collision risk level, ensuring sufficient space for braking and avoiding a collision. As a preferred implementation, the maximum permissible speed is inversely related to the comprehensive collision risk level; that is, when the comprehensive collision risk level increases, the maximum permissible speed should be reduced accordingly to increase reaction time and reduce braking distance requirements. Simultaneously, the minimum safe braking distance is positively related to the comprehensive collision risk level; that is, when the comprehensive collision risk level increases, the minimum safe braking distance the vehicle needs to maintain should be increased accordingly to provide a greater safety margin.

[0054] Through the above technical solution, this application can provide more specific, quantifiable, and actionable vehicle safety operation guidance parameters. This clear parameter definition and dynamic adjustment mechanism allows the vehicle control system or driver to intuitively understand and follow safety recommendations, avoiding misjudgments or improper operation caused by ambiguous parameters. Especially in perception-limited scenarios, by precisely controlling the maximum permissible speed and minimum safe braking distance, the risk of collision can be effectively reduced, improving the vehicle's emergency response capability and overall safety, thereby significantly enhancing the practicality and reliability of the aforementioned intelligent vehicle monitoring and early warning method.

[0055] In some embodiments, the multidimensional index includes at least one field selected from scenario type, primary failure sensor, auxiliary effective sensor, and target type; wherein, the auxiliary effective sensor refers to an auxiliary sensing device that works normally and collects auxiliary sensing signals in a perception-limited scenario, and the target type refers to the entity type of the detected target, including forklifts, heavy trucks, workers, and stacks of goods.

[0056] The aforementioned multidimensional index includes at least one field: Context Type, Primary Failure Sensor, Auxiliary Effective Sensor, and Target Type. The auxiliary effective sensor refers to an auxiliary sensing device that functions normally and collects auxiliary sensing signals under perception-limited conditions. The target type refers to the entity type of the detected target, including forklifts, heavy trucks, workers, and stacks of goods. Specifically, the Context Type field records the specific category of the identified perception-limited situation, such as perception limitation caused by insufficient light, rainy / foggy weather, or physical obstruction. The Primary Failure Sensor field records which primary sensing devices' environmental perception performance drops to a preset performance threshold under perception-limited conditions, thus preventing them from achieving accurate environmental perception. The Auxiliary Effective Sensor field records which auxiliary sensing devices function normally and successfully collect auxiliary sensing signals under perception-limited conditions to compensate for the deficiencies of the primary sensing devices. The Target Type field records the specific entity type of the core risk target detected under perception-limited conditions, which may include forklifts, heavy trucks, workers, or stacks of goods. These fields are designed to provide more refined and comprehensive data dimensions for subsequent event review and tracing.

[0057] The multidimensional index constructed using the above technical solutions provides richer and more detailed event information, greatly improving the efficiency and accuracy of post-event review and tracing. Specifically, by recording the scenario type, different types of perception-limited events can be classified, statistically analyzed, and high-frequency risk scenarios can be identified; by recording the main failed sensors and auxiliary effective sensors, the performance and complementary effects of different sensing devices can be evaluated, providing data support for optimization; and by recording the target type, more targeted safety strategies can be formulated for different types of risk targets. Therefore, this application enables in-depth analysis and precise tracing of vehicle operation safety events in complex park environments, providing strong support for safety management and improvement.

[0058] In some embodiments, the preset high-risk entity behavior rules include a typical kinematic parameter range of high-risk vehicles in the park. The typical kinematic parameter range includes at least one of physical profile size range, speed range, acceleration range, and turning radius range. The matching degree is calculated by weighted sum of physical profile size matching degree, speed matching degree, acceleration matching degree, and trajectory matching degree.

[0059] Specifically, the pre-defined high-risk entity behavior rules aim to provide a benchmark for vehicles operating within the park, identifying entities that may pose a serious threat to vehicle safety. These rules are not general but focus on the typical kinematic parameters of high-risk vehicles. Typical kinematic parameters can be understood as key indicators describing the motion state and geometric characteristics of an object. Among them, the physical profile size range defines the external dimensional limits of high-risk vehicles; for example, large forklifts or heavy trucks typically have large physical profile dimensions. The speed range specifies the speed range these high-risk vehicles may reach in a specific area or situation. The acceleration range describes the rate of change of their speed, while the turning radius range reflects their maneuverability. At least one of these parameters is used to construct the high-risk entity behavior rules to comprehensively characterize the motion characteristics of potential threat targets.

[0060] Furthermore, the matching degree is an indicator that measures the degree to which the kinematic characteristics of the detected target conform to the pre-defined behavioral rules for high-risk entities. This matching degree is not a single-dimensional comparison, but rather a weighted sum of physical contour size matching, velocity matching, acceleration matching, and trajectory matching. Physical contour size matching assesses the degree to which the actual size of the detected target conforms to the size range specified in the rules; velocity matching assesses the degree to which the actual velocity of the detected target conforms to the velocity range specified in the rules; acceleration matching assesses the degree to which the actual acceleration of the detected target conforms to the acceleration range specified in the rules; and trajectory matching assesses the degree of similarity between the actual motion trajectory of the detected target and the possible trajectory of a high-risk entity. By assigning different weights to these different dimensions of matching degree and summing them, a comprehensive matching degree value can be obtained, thereby more accurately quantifying the degree to which the detected target conforms to the behavioral rules for high-risk entities.

[0061] Through the aforementioned technical solution, this application can significantly improve the accuracy and reliability of identifying core risk targets in perception-limited situations. Because the pre-defined high-risk entity behavior rules are refined into multiple typical kinematic parameters, and the matching degree is calculated through multi-dimensional weighting, entities with potentially high-risk behavioral characteristics can be identified more precisely. This avoids misjudgments caused by incomplete or inaccurate perception information, thereby ensuring that vehicle safety operation decisions are based on a more accurate assessment of risk targets. Ultimately, this precise identification capability helps reduce the accident risk of vehicles operating within the park and improves overall operational safety.

[0062] In some embodiments, the comprehensive collision risk level is calculated by fusing the perception reliability loss index, the basic collision risk value corresponding to the perception information, the environmental risk factor, and the vehicle state risk factor. The basic collision risk value is calculated based on the relative distance and relative speed between the detected target and the vehicle. The environmental risk factor is determined based on the degree of danger of the area where the vehicle is currently located. The area is divided into high-risk, medium-risk, and low-risk areas according to the degree of danger. High-risk areas include intersections and loading / unloading areas, medium-risk areas include main roads in the area, and low-risk areas include parking lots. The vehicle state risk factor is determined based on the vehicle's own operating state, which includes the vehicle's current speed, acceleration, and braking performance parameters. The braking performance parameters include braking response time and maximum braking deceleration. The comprehensive collision risk level is calculated by multiplying the perception reliability loss index, the basic collision risk value, the environmental risk factor, and the vehicle state risk factor by preset weighting coefficients and then summing them.

[0063] Specifically, the calculation of the comprehensive collision risk level is designed as a multi-factor fusion model. The perception reliability degradation index reflects the quality and reliability of the auxiliary perception signals; a higher value indicates greater uncertainty in the perception data and a higher risk. The basic collision risk value directly quantifies the immediate probability of a collision between the detected target and the vehicle, primarily calculated based on the relative distance and relative speed between them. For example, the closer the distance and the greater the relative speed, the higher the basic collision risk value.

[0064] Furthermore, environmental risk factors are incorporated to assess the inherent hazard level of the area where the vehicle is located. The area can be categorized into different hazard levels based on factors such as traffic flow, operational complexity, and visibility obstruction. For example, intersections and loading / unloading areas are typically defined as high-risk areas due to frequent vehicle and pedestrian activity and numerous potential conflict points; main roads within the park may be defined as medium-risk areas, where vehicle speeds may be higher, but visibility and planning are generally better; parking lots may be defined as low-risk areas, where vehicle density is high, but speeds are typically very low. Environmental risk factors can be obtained by querying a pre-defined risk map of the park area based on the vehicle's current location information.

[0065] In addition, the vehicle state risk factor is used to assess the impact of the vehicle's own operating state on collision risk. The vehicle's own operating state includes its current speed, acceleration, and braking performance parameters. Among these, braking performance parameters, such as braking response time (the time from when the driver issues a braking command to when the braking system begins to effectively decelerate) and maximum braking deceleration (the maximum deceleration the vehicle can achieve during emergency braking), directly affect the vehicle's ability to avoid hazards in emergency situations. For example, the higher the vehicle speed, the greater the acceleration, and the worse the braking performance (longer braking response time, smaller maximum braking deceleration), the higher the vehicle state risk factor.

[0066] Ultimately, the comprehensive collision risk level is calculated by summing the results after multiplying the perceived reliability loss index, the basic collision risk value, the environmental risk factor, and the vehicle state risk factor by preset weighting coefficients. These weighting coefficients can be optimized based on actual application scenarios, expert experience, or machine learning methods to reflect the relative importance of each factor's contribution to the overall collision risk.

[0067] Through the aforementioned technical solutions, this application significantly improves the accuracy and comprehensiveness of risk assessment for intelligent vehicle monitoring and early warning systems in complex park environments. By integrating the perception reliability degradation index, it effectively addresses the uncertainty of auxiliary perception data in perception-limited situations, avoiding risk underestimation due to data quality issues. Simultaneously, the introduction of environmental risk factors and vehicle state risk factors ensures that risk assessment is no longer limited to simple target distance and speed, but comprehensively considers the inherent hazards of the vehicle's environment and the vehicle's own dynamic risk avoidance capabilities, thereby more accurately identifying potential high-risk scenarios. This multi-dimensional and refined risk assessment method makes the generated comprehensive collision risk level more realistically reflect the actual threat, providing a more reliable and robust basis for generating vehicle safety operation guidance parameters and formulating safe operation decisions, effectively reducing the probability of collision accidents and improving the overall safety of vehicle operation within the park.

[0068] In some embodiments, the contextualized event package is stored in a serializable file format or a database that supports complex document structures. The serializable file format is JSON or Protocol Buffers, and the database that supports complex document structures is a NoSQL database.

[0069] Specifically, a contextualized event package is a collection of structured records and packages of key information such as perception-limited situations, auxiliary perception signals, perception reliability degradation index, calculation process of comprehensive collision risk level, and safe operation guidance parameters. To ensure that this complex and diverse data can be efficiently stored, transmitted, and parsed, this application proposes using serializable file formats or databases that support complex document structures for storage. Serializable file formats refer to file types that can convert data structures or object states into storable or transmittable formats. JSON (JavaScript Object Notation) is a lightweight data exchange format that is easy for humans to read and write, and also easy for machines to parse and generate, and is widely used in network data transmission and configuration file storage. Protocol Buffers, developed by Google, is a language- and platform-independent, scalable method for serializing structured data. Compared to JSON, it typically has advantages in data size and parsing speed, making it particularly suitable for performance-critical scenarios. Furthermore, databases that support complex document structures refer to database systems capable of flexibly storing unstructured or semi-structured data, such as NoSQL databases. NoSQL databases, such as MongoDB and Cassandra, can store data in various forms, including documents, key-value pairs, column families, or graphs. Their flexibility and scalability make them ideal for storing contextualized event packages, which contain information with multiple data types and complex nested structures, thus avoiding the schema rigidity problem that traditional relational databases may encounter when processing such data.

[0070] Through the above technical solutions, the storage method for contextualized event packages has been clarified and optimized. Employing serializable file formats or databases that support complex document structures not only ensures the integrity and consistency of complex data within the contextualized event packages but also significantly improves data storage efficiency and retrieval speed. This enables the rapid and accurate acquisition of all relevant environmental perception data, risk assessment results, and decision-making processes when post-event review and tracing of vehicle operation events are required. This effectively supports accident cause analysis, system performance evaluation, and future strategy improvements, significantly enhancing the practicality and reliability of the entire intelligent vehicle monitoring and early warning system.

[0071] In some embodiments, the step of continuously tracking the core risk target and making vehicle safety operation decisions includes: acquiring the instantaneous position, speed and trajectory of the core risk target in real time by fusing auxiliary sensing signals, vehicle self-positioning information and attitude information; Predict the subsequent movement trend of the core risk target within a preset time range, and formulate vehicle safety operation decisions based on the comprehensive collision risk level and the vehicle safety operation guidance parameters.

[0072] Specifically, fusing auxiliary perception signals, vehicle self-positioning information, and attitude information involves integrating data from different sensors and systems. Auxiliary perception signals, collected by radar, lidar, cameras, and other auxiliary sensing devices, provide information such as distance, speed, and angle of key risk targets. Vehicle self-positioning information, provided by GPS, INS, and other systems, determines the vehicle's precise position in the global coordinate system. Attitude information describes the vehicle's pitch, roll, and yaw angles, crucial for understanding its motion state. By synchronizing and spatially registering these heterogeneous data, a more comprehensive and accurate understanding of key risk targets and their relationship with the vehicle can be achieved.

[0073] Real-time acquisition of the instantaneous position, velocity, and trajectory of key risk targets means that these critical kinematic parameters can be continuously updated with extremely low latency. Instantaneous position refers to the spatial coordinates of the key risk target at a given moment; velocity includes its linear and angular velocities, reflecting its speed and direction of movement; and trajectory is a set of consecutive positions of the key risk target over a period of time, revealing its historical movement path. The real-time nature and accuracy of this information are the foundation for subsequent predictions and decision-making.

[0074] Predicting the subsequent movement trend of a core risk target within a preset time range involves using historical motion data and the current motion state to infer the target's position, speed, and trajectory over a future period using motion models (such as Kalman filtering, particle filtering, and deep learning models). The preset time range can be dynamically adjusted based on the actual application scenario and risk level. For example, in high-speed driving or high-risk areas, the prediction time range may be longer to allow the vehicle more reaction time.

[0075] Making vehicle safety operation decisions by combining the comprehensive collision risk level and vehicle safety operation guidance parameters involves comprehensively considering the predicted movement trend of core risk targets with the currently calculated comprehensive collision risk level and the generated safety operation guidance parameters. The comprehensive collision risk level reflects the probability and severity of a collision in the current situation; the vehicle safety operation guidance parameters provide specific safety operation suggestions, such as the maximum permissible driving speed and the minimum safe braking distance. Based on this information, the most appropriate safety operation decision can be intelligently selected, including but not limited to deceleration, avoidance, or emergency braking, to minimize potential risks.

[0076] Through the aforementioned technical solutions, this application significantly improves the accuracy and stability of tracking core risk targets, especially in scenarios where primary sensing equipment is limited. Multi-source information fusion ensures that even when some information is damaged or uncertain, it can be supplemented and corrected by other information sources, thereby obtaining more reliable target motion state data. Furthermore, the introduction of a motion trend prediction mechanism enables the vehicle to anticipate the future behavior of core risk targets, gaining valuable reaction time for safe operation decisions and allowing for the development of more forward-looking and effective deceleration, avoidance, or emergency braking strategies. This not only enhances the vehicle's operational safety in complex park environments but also reduces the risk of false alarms or missed alarms due to perception uncertainty, making safe operation decisions more intelligent and reliable.

[0077] This application also proposes a vehicle intelligent monitoring and early warning system, such as... Figure 2 As shown, a vehicle intelligent monitoring and early warning system 100 includes: The scenario judgment module 10 is used to continuously acquire the working status information of the vehicle's main sensing devices, determine whether the environmental perception performance of the main sensing devices has dropped to a preset performance threshold, and thus identify the perception-restricted scenario caused by sudden obstructions in the park. The perception-restricted scenario is a scenario in which the main sensing devices cannot achieve accurate environmental perception due to obstructions. The index calculation module 20 is used to collect auxiliary sensing signals through auxiliary sensing devices when the perception-limited situation is identified, and to perform a reliability quantification assessment of the auxiliary sensing signals to generate a perception reliability loss index. The core risk target identification module 30 is used to extract perception information related to vehicle safe operation and kinematic features of the detection target from the auxiliary perception signal, match the kinematic features with preset high-risk entity behavior rules, and identify the detection target with a matching degree higher than a preset matching degree threshold as the core risk target. The risk assessment and decision-making module 40 is used to calculate the comprehensive collision risk level based on the perceived reliability loss index, the perceived information and the vehicle's own operating status, generate vehicle safety operation guidance parameters based on the comprehensive collision risk level, continuously track the core risk targets, and formulate vehicle safety operation decisions, the safety operation decisions including at least one of deceleration, avoidance and emergency braking. The storage and retrieval module 50 is used to structurally record and package the perception-limited situation, the auxiliary perception signal, the perception reliability loss index, the calculation process of the comprehensive collision risk level, and the safety operation guidance parameters into a contextualized event package and store it. At the same time, it constructs a multi-dimensional index with perception-limited situation type, perception equipment status, and safety operation decision as causal correlation dimensions to realize post-event review and traceability.

[0078] Traditional vehicle monitoring and early warning systems often lack a clear module to intelligently identify and quantify situations where perception is limited due to sudden environmental changes within a park. They also cannot dynamically adjust data fusion strategies, resulting in limited accuracy and decision-making capabilities in real-time risk assessment. In contrast, this application introduces a scenario judgment module, an index calculation module, a core risk target identification module, a risk assessment and decision-making module, and a storage and retrieval module, forming a highly integrated and functionally defined system architecture. This modular design enables the system to proactively perceive and understand complex operating scenarios. Especially when the performance of the main sensing equipment degrades, the index calculation module can dynamically assess the reliability of auxiliary sensing signals, thereby synthesizing a high-confidence collision risk assessment in real time from incomplete, uncertain, multi-source heterogeneous data. For example, when vapor cloud obscures cameras and lidar, causing them to malfunction, traditional systems may be unable to provide effective warnings due to data loss. However, the system in this application can still identify and assess core risk targets by using auxiliary sensing devices such as millimeter-wave radar, combined with a perception reliability loss index generated by an index calculation module. The risk assessment and decision-making module then generates corresponding safety operation guidance parameters, such as deceleration or avoidance, thereby significantly improving the vehicle's safe operation capability in complex environments.

[0079] Furthermore, this application uses a storage and retrieval module to structure and package key information such as perception-limited situations, auxiliary perception signals, perception reliability depreciation index, the calculation process of comprehensive collision risk level, and safe operation guidance parameters into contextualized event packages. It also constructs a multi-dimensional index with perception-limited situation type, perception equipment status, and safe operation decisions as causal correlation dimensions, greatly improving the efficiency and accuracy of post-event review and tracing. While traditional systems also record multi-source heterogeneous data, they lack contextualized data association and causal chain construction mechanisms, making it difficult for analysts to efficiently reconstruct the truth of events from massive amounts of fragmented data. The system in this application, however, can quickly locate specific types of perception-limited events through the multi-dimensional index of the storage and retrieval module and trace its complete causal chain from perception to decision, thereby effectively identifying system defects and making targeted improvements. This contextualized data organization and tracing capability is not available in existing technologies, providing a solid technical guarantee for the safe operation of special vehicles in complex and dynamic environments.

[0080] The above description is merely an embodiment of this application and is not intended to limit the scope of protection 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 protection of this application.

Claims

1. A vehicle intelligent monitoring and early warning method, characterized in that, Includes the following steps: The system continuously acquires the working status information of the vehicle's main sensing devices, determines whether the environmental perception performance of the main sensing devices has dropped to a preset performance threshold, and thus identifies perception-restricted situations caused by sudden obstructions in the park. The perception-restricted situation is a situation where the main sensing devices cannot achieve accurate environmental perception due to obstructions. When the perception-limited situation is identified, auxiliary perception signals are collected through auxiliary perception devices, and the reliability of the auxiliary perception signals is quantitatively evaluated to generate a perception reliability loss index. The system extracts perception information related to vehicle safety operation and kinematic features of the detected targets from the auxiliary perception signals. It then matches the kinematic features with preset high-risk entity behavior rules and identifies detected targets with a matching degree higher than a preset matching degree threshold as core risk targets. The perception information related to vehicle safety operation includes the relative distance and relative speed between the detected targets and the vehicle, as well as the location of obstacles. Based on the perceived reliability loss index, the perceived information and the vehicle's own operating status, a comprehensive collision risk level is calculated. Based on the comprehensive collision risk level, vehicle safety operation guidance parameters are generated. The core risk targets are continuously tracked, and vehicle safety operation decisions are made. The safety operation decisions include at least one of deceleration, avoidance and emergency braking. The perception-limited situation, the auxiliary perception signal, the perception reliability loss index, the calculation process of the comprehensive collision risk level, and the safety operation guidance parameters are structured, recorded, packaged into a contextualized event package, and stored. At the same time, a multi-dimensional index is constructed with perception-limited situation type, perception equipment status, and safety operation decision as causal correlation dimensions to achieve post-event review and traceability.

2. The vehicle intelligent monitoring and early warning method according to claim 1, characterized in that, The step of performing a reliability quantification assessment of the auxiliary sensing signal to generate a sensing reliability loss index includes: The perception reliability loss index is generated by quantifying at least one of the following: signal-to-noise ratio of the auxiliary sensing signal, standard deviation of distance measurement, standard deviation of velocity measurement, and fluctuation amplitude of target features. The perception reliability loss index is used to calculate the comprehensive collision risk level.

3. The vehicle intelligent monitoring and early warning method according to claim 2, characterized in that, The target feature fluctuation amplitude is determined by tracking the changes in the target's speed, physical contour size, and position relative to the vehicle within consecutive frames, and calculating the fluctuation range of each change within a preset time period.

4. The vehicle intelligent monitoring and early warning method according to claim 1, characterized in that, The safety operation guidance parameters include the maximum permissible driving speed and the minimum safe braking distance. The maximum permissible driving speed decreases as the comprehensive collision risk level increases, while the minimum safe braking distance increases as the comprehensive collision risk level increases.

5. The vehicle intelligent monitoring and early warning method according to claim 1, characterized in that, The multidimensional index includes at least one field from the following: scenario type, primary failure sensor, auxiliary effective sensor, and target type; wherein, the auxiliary effective sensor refers to an auxiliary sensing device that works normally and collects auxiliary sensing signals in a perception-limited scenario, and the target type refers to the entity type of the detected target, including forklifts, heavy trucks, workers, and stacks of goods.

6. The vehicle intelligent monitoring and early warning method according to claim 1, characterized in that, The preset high-risk entity behavior rules include the typical kinematic parameter range of high-risk vehicles in the park. The typical kinematic parameter range includes at least one of the physical contour size range, speed range, acceleration range, and turning radius range. The matching degree is calculated by weighted sum of physical contour size matching degree, speed matching degree, acceleration matching degree, and trajectory matching degree.

7. The intelligent vehicle monitoring and early warning method according to claim 1, characterized in that, The comprehensive collision risk level is calculated by integrating the perception reliability loss index, the basic collision risk value corresponding to the perception information, the environmental risk factor, and the vehicle state risk factor. The basic collision risk value is calculated based on the relative distance and relative speed between the detected target and the vehicle; Environmental risk factors are determined based on the degree of danger of the area where the vehicle is currently located. The area is divided into high-risk, medium-risk, and low-risk zones. High-risk zones include intersections and loading / unloading areas, medium-risk zones include the main roads of the park, and low-risk zones include parking lots. Vehicle status risk factors are determined based on the vehicle's own operating status, which includes the vehicle's current speed, acceleration, and braking performance parameters. The braking performance parameters include braking response time and maximum braking deceleration. The comprehensive collision risk level is calculated by multiplying the perceived reliability loss index, the basic collision risk value, the environmental risk factor, and the vehicle state risk factor by preset weighting coefficients and then summing them.

8. The intelligent vehicle monitoring and early warning method according to claim 1, characterized in that, The contextualized event package is stored in a serializable file format or a database that supports complex document structures. The serializable file format is JSON or Protocol Buffers, and the database that supports complex document structures is a NoSQL database.

9. The vehicle intelligent monitoring and early warning method according to claim 1, characterized in that, The steps for continuously tracking the core risk targets and making vehicle safety operation decisions include: obtaining the instantaneous position, speed and trajectory of the core risk targets in real time by fusing auxiliary sensing signals, vehicle self-positioning information and attitude information; Predict the subsequent movement trend of the core risk target within a preset time range, and formulate vehicle safety operation decisions based on the comprehensive collision risk level and the vehicle safety operation guidance parameters.

10. A vehicle intelligent monitoring and early warning system, characterized in that, The system includes: The scenario judgment module is used to continuously acquire the working status information of the vehicle's main sensing devices, determine whether the environmental perception performance of the main sensing devices has dropped to a preset performance threshold, and thus identify the perception-restricted scenario caused by sudden obstructions in the park. The perception-restricted scenario is a scenario in which the main sensing devices cannot achieve accurate environmental perception due to obstructions. The index calculation module is used to collect auxiliary sensing signals through auxiliary sensing devices when the perception-limited situation is identified, and to perform a reliability quantification assessment of the auxiliary sensing signals to generate a perception reliability loss index. The core risk target identification module is used to extract perception information related to vehicle safe operation and kinematic features of detection targets from the auxiliary perception signals, match the kinematic features with preset high-risk entity behavior rules, and identify detection targets with a matching degree higher than a preset matching degree threshold as core risk targets. The risk assessment and decision-making module is used to calculate the comprehensive collision risk level based on the perceived reliability loss index, the perceived information and the vehicle's own operating status, generate vehicle safety operation guidance parameters based on the comprehensive collision risk level, continuously track the core risk targets, and formulate vehicle safety operation decisions, the safety operation decisions including at least one of deceleration, avoidance and emergency braking. The storage and retrieval module is used to structurally record and package the perception-limited situation, the auxiliary perception signal, the perception reliability loss index, the calculation process of the comprehensive collision risk level, and the safety operation guidance parameters into a contextualized event package and store it. At the same time, it constructs a multi-dimensional index with perception-limited situation type, perception equipment status, and safety operation decision as causal correlation dimensions to realize post-event review and traceability.

Images