A vehicle dynamic state evaluation method and system fusing multi-sensing information

Abstract

The application discloses a kind of vehicle dynamic state evaluation method and system of fusion multi-sensing information, it is related to vehicle state evaluation related technical field, the method includes: obtaining vehicle CAN data, intelligent driving system data and carrying out independent analysis and fusion analysis, based on the relationship of analysis data safety response decomposition is carried out, obtains safety influence factor;Based on safety influence factor time series pattern recognition is carried out, and time series trend relationship is established;State time series evaluation is carried out by time series trend relationship, and dynamic risk evaluation information is obtained;According to dynamic risk evaluation information, vehicle state is mapped, and vehicle dynamic evaluation result is generated.Solve the technical problems that the prior art exists, it is difficult to fuse vehicle multi-source information, static evaluation cannot capture the dynamic risk evolution trend of vehicle, leading to the accuracy, prospective deficiency of vehicle state evaluation, reaches the realization vehicle state Dynamic safety evaluation, improve the accuracy and prospective of state evaluation technical effect.

Description

Technical Field

[0001] This application relates to the technical field of vehicle condition assessment, specifically to a method and system for assessing vehicle dynamic condition by integrating multi-sensor information. Background Technology

[0002] Container transfer vehicles, mining dump trucks, yard cranes, and forklifts in closed or semi-closed environments such as ports and mines are rapidly developing towards automation and intelligence. These environments are characterized by complex conditions such as fixed obstacles, mobile equipment, and mixed pedestrian traffic, as well as harsh working conditions such as uneven roads, high dust levels, and varying visibility. They are also characterized by high work intensity and mainly low-to-medium speed reciprocating transportation. Traditional vehicle condition assessments often rely on vehicle controller area network bus data or limited dynamic models. When facing complex and ever-changing actual driving scenarios, especially dynamic working conditions that require the prediction of potential risks, there are limitations such as assessment lag and insufficient foresight. CAN data alone is insufficient to fully understand a vehicle's "intentions" and the context of its environment. For example, vehicle CAN data may show that the braking system is normal, but the intelligent driving system may perceive an emergency obstacle ahead and issue an emergency braking request, while the vehicle's dynamic response may be delayed or deviated. The mismatch between the system's intentions and the vehicle's response can cause significant dynamic risks, which can easily lead to collisions, rollovers, or work disruptions in low- and medium-speed operations in ports and mines. Furthermore, it is impossible to extract features that accurately reflect the vehicle's overall dynamic safety status, and it is difficult to capture the accumulation and evolution trend of vehicle safety risks. Slight oversteer may develop into severe sideslip or fishtailing within several consecutive control cycles, affecting the accuracy and reliability of vehicle status assessment.

[0003] Therefore, current technologies suffer from technical problems such as difficulty in integrating multi-source vehicle information and the inability of static assessments to capture the dynamic risk evolution trends of vehicles, resulting in insufficient accuracy and foresight in vehicle condition assessments. Summary of the Invention

[0004] This application provides a vehicle dynamic state assessment method and system that integrates multi-sensor information, solving the technical problems in the prior art where it is difficult to integrate multi-source vehicle information and static assessment cannot capture the dynamic risk evolution trend of the vehicle, resulting in insufficient accuracy and foresight in vehicle state assessment. It achieves the technical effect of realizing dynamic safety assessment of vehicle state and improving the accuracy and foresight of state assessment.

[0005] This application provides a method for assessing the dynamic state of a vehicle by integrating multi-sensor information. The method includes: acquiring vehicle CAN data and intelligent driving system data; independently parsing and fusing the vehicle CAN data and intelligent driving system data; performing safety response decomposition based on the parsed data relationships to obtain safety impact factors; performing time-series pattern recognition based on the safety impact factors to establish time-series trend relationships of the safety impact factors; evaluating the state time-series of the vehicle CAN data and intelligent driving system data through the time-series trend relationships to obtain dynamic risk assessment information; and mapping the vehicle state according to the dynamic risk assessment information to generate a vehicle dynamic assessment result.

[0006] In a possible implementation, the vehicle dynamic state assessment method integrating multi-sensor information further performs the following processing: extracting vehicle control sensor data based on the vehicle CAN data according to the time dimension and driving scenario state dimension; obtaining autonomous driving module sensor data, environmental perception data, path planning data, and vehicle state monitoring data based on the intelligent driving system data; performing risk state analysis on the vehicle control sensor data, autonomous driving module sensor data, environmental perception data, path planning data, and vehicle state monitoring data respectively to determine independently analyzed risk data; analyzing the collaborative risk interaction relationship among the vehicle control sensor data, autonomous driving module sensor data, environmental perception data, path planning data, and vehicle state monitoring data to establish fused analyzed risk data.

[0007] In a possible implementation, the vehicle dynamic state assessment method that integrates multi-sensor information further performs the following processing: based on the independently parsed risk data, establish an independently parsed mapping topology relationship according to the mapping relationship between risk data and sensor data; based on the fused parsed risk data, add the fused risk relationship to the independently parsed mapping topology relationship according to the mapping relationship between risk data and sensor data, and perform hierarchical expansion of the mapping topology relationship to obtain the parsed data relationship.

[0008] In a possible implementation, the vehicle dynamic state assessment method integrating multi-sensor information further performs the following processing: extracting risk data based on the parsed data relationships; obtaining multiple risk events by performing cluster analysis on the risk data, each risk event representing a safety threat or potential dangerous scenario; tracing the sensor data relationships based on the parsed data relationships, centered on the risk events, to obtain sensor data paths; performing sample fitting on the sensor data of each node based on the sensor data paths, establishing a functional relationship between the sensor data of each node and the risk events, performing safety impact factor decomposition, obtaining the minimum cut set of the impact factor combination, and determining the safety impact factors.

[0009] In a possible implementation, the vehicle dynamic state assessment method that integrates multi-sensor information further performs the following processing: introducing a generative adversarial network model, dynamically adjusting the mapping weights and topology between risk data and sensor data based on actual vehicle behavior feedback and environmental state changes, and optimizing the relationship between the parsed data.

[0010] In a possible implementation, the vehicle dynamic state assessment method integrating multi-sensor information further performs the following processing: constructing a generator network and a discriminator network respectively, and combining the generator network and the discriminator network into an adversarial training framework; training and converging the adversarial training framework using historical sample data to obtain an adversarial network; wherein, based on the data feedback generated by the adversarial network, weak links in the mapping topology are identified; wherein, the simulated data generated by the generator network is injected into the analytical data relationship for risk assessment; if the risk assessment confidence is lower than a set risk threshold, it is determined that the current mapping topology has a weak link in processing this type of data pattern; and based on the discrimination accuracy of the discriminator network, the contribution of each sensor data to risk identification is calculated; and the connection weights in the topology are dynamically adjusted based on the contribution to optimize the analytical data relationship.

[0011] In a possible implementation, the vehicle dynamic state assessment method that integrates multi-sensor information further performs the following processing: collecting sensor data sequences under normal driving conditions from historical data as positive samples, and collecting sensor data sequences before known risk events occur as negative samples; inputting vehicle state parameters as conditional information, along with the data sequences, into the network, and alternately training the generator network and the discriminator network, so that the generator network can generate simulated data that conforms to the real data distribution but contains potential risk characteristics; injecting the high-quality simulated data generated by the generator into the mapping topology training process to enhance the model's ability to identify edge cases, until the convergence target is reached, thus obtaining the adversarial network.

[0012] In a possible implementation, the vehicle dynamic state assessment method integrating multi-sensor information further performs the following processing: extracting multi-scenario safety events with independent parameters and fused parameters based on the safety impact factors; fitting the time-series response relationship of the safety impact factors to each scenario safety event to obtain a multi-scenario safety time-series model; collecting vehicle runtime sequence data based on the multi-scenario safety events to construct a positive example validation set and a negative example validation set; verifying the parameters of the multi-scenario safety time-series model using the positive example validation set and the negative example validation set; and using the verification feedback data to perform time-series iterative updates to the multi-scenario safety time-series model.

[0013] In a possible implementation, the vehicle dynamic state assessment method integrating multi-sensor information further performs the following processing: The vehicle CAN data and intelligent driving system data are segmented according to multiple time scales to form multi-scale data sequences. Each time scale data sequence is associated with the corresponding real-time total vehicle weight, historical health data of key components, and the current driving scenario to form an analysis unit with contextual information. The analysis unit is then input into the time-series trend prediction model corresponding to each time scale to obtain risk identification and prediction data for each time scale. The risk identification and prediction data for each time scale are then time-aligned and arranged. Based on the aligned data, risk response calculations are performed according to multiple scenario safety events to obtain the dynamic risk assessment information. The dynamic risk assessment information is a quantified dynamic risk value and includes structured dynamic risk assessment information containing the main risk sources, risk evolution trajectory, and suggested intervention measures.

[0014] This application also provides a vehicle dynamic state assessment system that integrates multi-sensor information. The system includes: a data parsing and decomposition module, used to acquire vehicle CAN data and intelligent driving system data, perform independent parsing and fusion parsing of the vehicle CAN data and intelligent driving system data, and perform safety response decomposition based on the parsed data relationships to obtain safety impact factors; a time-series pattern recognition module, used to perform time-series pattern recognition based on the safety impact factors and establish the time-series trend relationship of the safety impact factors; a state-time-series evaluation module, used to perform state-time-series evaluation of the vehicle CAN data and intelligent driving system data through the time-series trend relationship to obtain dynamic risk assessment information; and a vehicle state mapping module, used to map the vehicle state according to the dynamic risk assessment information to generate vehicle dynamic assessment results.

[0015] This application proposes a vehicle dynamic state assessment method and system that integrates multi-sensor information. The method acquires vehicle CAN data and intelligent driving system data, performs independent and fused analysis, decomposes safety responses based on the analyzed data relationships to obtain safety impact factors, identifies temporal patterns based on these factors to establish temporal trend relationships, evaluates the state temporal sequence through these trends to obtain dynamic risk assessment information, and maps the vehicle state based on the dynamic risk assessment information to generate a vehicle dynamic assessment result. This addresses the technical problems in existing technologies, such as the difficulty in integrating multi-source vehicle information and the inability of static assessments to capture the dynamic risk evolution trends of vehicles, leading to insufficient accuracy and foresight in vehicle state assessment. The method achieves dynamic safety assessment of vehicle state, improving the accuracy and foresight of state assessment. Attached Figure Description

[0016] To more clearly illustrate the technical solutions of the embodiments of this disclosure, the accompanying drawings of the embodiments of this disclosure will be briefly described below. Flowcharts are used in this application to illustrate the operations performed by the system according to the embodiments of this application. It should be understood that the preceding or following operations are not necessarily performed precisely in sequence. Instead, various steps can be processed in reverse order or simultaneously as needed. Furthermore, other operations can be added to these processes, or one or more steps can be removed from these processes.

[0017] Figure 1 This is a schematic flowchart of a vehicle dynamic state assessment method that integrates multi-sensor information, provided in an embodiment of this application.

[0018] Figure 2 This is a schematic diagram of a vehicle dynamic state assessment system that integrates multi-sensor information, provided as an embodiment of this application.

[0019] Figure labeling: Data parsing and decomposition module 10, time sequence pattern recognition module 20, state time sequence evaluation module 30, vehicle state mapping module 40. Detailed Implementation

[0020] The above description is merely an overview of the technical solution of this application. In order to better understand the technical means of this application and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of this application more obvious and understandable, specific embodiments of this application are given below.

[0021] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description of this application will be provided in conjunction with the accompanying drawings. The described embodiments should not be considered as limitations on this application. All other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0022] In the following description, references to "some embodiments" describe a subset of all possible embodiments. However, it is understood that "some embodiments" can be the same or different subsets of all possible embodiments and can be combined with each other without conflict. The terms "first" and "second" are used merely to distinguish similar objects and do not represent a specific ordering of objects. The terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or server that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or modules not explicitly listed or inherent to these processes, methods, products, or devices. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing embodiments of this application only.

[0023] This application provides a method for evaluating the dynamic state of a vehicle by fusing information from multiple sensors, such as... Figure 1 As shown, the method includes:

[0024] Step S100: Obtain vehicle CAN data and intelligent driving system data; perform independent and fusion analysis on the vehicle CAN data and intelligent driving system data; decompose safety response based on the analyzed data relationships to obtain safety impact factors.

[0025] Preferably, vehicle CAN data and intelligent driving system data are acquired. Vehicle CAN data refers to bus data from the vehicle's underlying controller area network, which provides basic information about the vehicle's underlying actuators and vehicle body status. It mainly reflects the vehicle's own status and control commands, such as vehicle speed, wheel speed, yaw rate, longitudinal / lateral acceleration, steering wheel angle, throttle / brake opening, braking pressure, and gear position. Intelligent driving system data mainly reflects the vehicle's perception of the external environment and its own driving decisions. It may include environmental perception data, such as distance to the vehicle in front, lane departure, and the position and speed of identified pedestrians / vehicles; path planning data, such as preset driving trajectory, target speed, and upcoming lane change or turning commands; and vehicle status monitoring data, such as system self-test status, module confidence, and actuator status.

[0026] Preferably, the vehicle CAN data and intelligent driving system data are analyzed independently to determine whether there are any abnormal or risky states within their respective data streams. Specifically, based on the vehicle dynamics model, the vehicle CAN data is analyzed independently to determine whether the vehicle's state is stable or tending to become unstable. For example, if the ratio of yaw rate to steering wheel angle is found to be abnormal, it may indicate that the vehicle is beginning to experience oversteer. Based on the driving strategy and environmental perception results, the intelligent driving system data is analyzed independently to determine whether the decision-making level is facing danger. For example, if the estimated collision time with the vehicle in front is less than 2 seconds, it indicates a risk of rear-end collision.

[0027] Preferably, vehicle CAN data and intelligent driving system data are fused and analyzed, that is, the results of independent analysis are correlated and cross-validated to analyze the synergistic or contradictory relationships between different data sources, thereby discovering more complex and hidden composite risks. For example, the intelligent driving system independently analyzes the risk of a rear-end collision with too low TTC and issues an emergency braking command, but the independent analysis of vehicle CAN data shows that the braking pressure has not increased significantly. Correlating these two independent risks identifies a higher level of risk, such as braking system response failure or actuator failure. Then, a mapping topology relationship is established between the various risk judgments determined by independent and fused analysis and the original sensor data to form an analytical data relationship. Then, a safety response decomposition is performed based on the analytical data relationship, that is, the most critical driving parameters are extracted by reverse tracing and quantification. Specifically, starting from a specific identified risk, all underlying data sources and paths contributing to the risk are traced back along the established mapping topology relationship. Among the multiple parameters on the tracing path, a set of core parameters with the smallest number but the greatest impact is identified through cluster analysis and sensitivity analysis as safety impact factors. For example, after decomposing the safety response of the braking system response failure, the final determined safety impact factors may include the intelligent driving system braking request command, the braking pressure value fed back by ESP, and the delay time between the two.

[0028] Furthermore, step S100 also includes step S110, extracting vehicle control sensor data based on the vehicle CAN data according to the time dimension and driving scenario state dimension; step S120, obtaining autonomous driving module sensor data, environmental perception data, path planning data, and vehicle status monitoring data based on the intelligent driving system data; step S130, performing risk state analysis on the vehicle control sensor data, autonomous driving module sensor data, environmental perception data, path planning data, and vehicle status monitoring data respectively, and determining independently analyzed risk data; step S140, analyzing the collaborative risk interaction relationship among the vehicle control sensor data, autonomous driving module sensor data, environmental perception data, path planning data, and vehicle status monitoring data, and establishing fused analyzed risk data.

[0029] Preferably, vehicle control sensor data is extracted from vehicle CAN data according to the time dimension, that is, data sequences within a specific time window are extracted to analyze trends, such as the acceleration change curve within the past 3 seconds, to determine whether the vehicle is continuously accelerating or decelerating. Vehicle control sensor data is also extracted according to the driving scenario state dimension, that is, relevant data is extracted based on the current scenario. For example, in a highway cruising scenario, the focus is on vehicle speed, following distance, and yaw rate; in a congested following scenario, the focus is on vehicle speed, acceleration, braking pressure, and distance to the vehicle in front. Autonomous driving module sensor data, environmental perception data, path planning data, and vehicle status monitoring data are obtained from the intelligent driving system data. Autonomous driving module sensor data refers to the raw or pre-processed data from the intelligent driving system's own sensors, such as target lists from cameras, radar, and LiDAR. Environmental perception data refers to the results of fusion and understanding of sensor data, such as lane line equations, drivable areas, and traffic sign recognition results. Path planning data is the system's decision intent, such as the target path trajectory, target vehicle speed, and planned lane change or braking commands. Vehicle status monitoring data refers to the intelligent driving system's assessment of its own health status, such as sensor confidence, system module operating status, and positioning accuracy.

[0030] Preferably, risk status analysis is performed on vehicle control sensor data, autonomous driving module sensor data, environmental perception data, path planning data, and vehicle status monitoring data to determine whether they are in an abnormal or risky state. Specifically, the vehicle control sensor data is analyzed based on the vehicle dynamics model to output dynamic risk, actuator risk, or behavioral risk; the autonomous driving module sensor data and environmental perception data are analyzed based on perception algorithm confidence and physical laws to output perception risk, environmental risk, or collision risk; and the path planning data is analyzed based on safety rules and comfort indicators to output planning risk or decision risk.

[0031] Preferably, vehicle control sensor data, autonomous driving module sensor data, environmental perception data, path planning data, and vehicle status monitoring data are correlated and cross-validated. This involves analyzing the collaborative risk interactions between them to identify deeper-level composite risks, including consistency and inconsistencies in time, logic, and physical aspects of different data streams. For example, if path planning data issues a strong braking command but vehicle control sensor data shows a very small actual deceleration, the fused analysis risk data may indicate actuator failure, such as abnormal braking system response. If environmental perception data identifies a curve ahead but vehicle control sensor data shows excessive speed and insufficient steering wheel angle, the fused analysis risk data may indicate trajectory tracking risk, such as the current speed causing lane departure during a curve. If independent analysis of risk data reveals perception risks such as blurred lane lines and health risks such as decreased camera confidence, the fused analysis risk may indicate reduced vehicle lateral positioning reliability, requiring a downgrade of intelligent driving functions. If environmental perception data shows sensors detecting severe bumps and vehicle control sensor data shows abnormal suspension height sensor data, the fused analysis risk data may indicate environmental interference risk, with perception anomalies possibly caused by adverse road conditions.

[0032] Furthermore, step S100 also includes step S150, which establishes an independently parsed mapping topology relationship based on the independently parsed risk data and according to the mapping relationship between risk data and sensor data; step S160, which adds the fused risk relationship to the independently parsed mapping topology relationship based on the fused parsed risk data and according to the mapping relationship between risk data and sensor data, and performs hierarchical expansion of the mapping topology relationship to obtain the parsed data relationship.

[0033] Preferably, each independently parsed risk data is used as the endpoint node in the topology. The logic that generated the risk is used to trace back to identify the input sensor data that directly led to the risk determination. These sensor data are then used as starting nodes and connected to the risk node via directed edges to construct a local topology, thereby obtaining the mapping topology relationship of the independently parsed data. Then, each fused parsed risk data is used as a new endpoint node, added to the mapping topology relationship of the independently parsed data according to the mapping relationship between risk data and sensor data. Next, the mapping topology relationship is hierarchically expanded, i.e., lower-level nodes are connected to the newly added fused risk nodes using directed edges, thus expanding the original local topology into a hierarchical relationship network. Finally, the parsed data relationship is determined, i.e., a multi-layered directed mapping topology network, with the bottom layer being sensors and data sources, the middle layer being independent risks, and the upper layer being cross-validated fused risks.

[0034] Furthermore, step S160 also includes introducing a generative adversarial network model, dynamically adjusting the mapping weights and topology between risk data and sensor data based on the actual vehicle behavior feedback and changes in environmental state, and optimizing the relationship between the parsed data.

[0035] Preferably, a generative adversarial network (GAN) model is introduced to optimize the analytical data relationships, enabling the risk identification model to dynamically adjust based on the vehicle's actual performance and real-world environmental changes. Specifically, the GAN model consists of a generator and a discriminator. The generator is a risk scenario simulator that uses random noise and the current vehicle / environment state as input, learns the distribution of real data, and generates realistic simulated sensor data sequences that may contain potentially hazardous features. The discriminator is a risk identification capability evaluator that uses real historical data or the generator's sensor data sequences as input and attempts to determine whether the data is real or fabricated by the generator. Based on the vehicle's actual behavior feedback and changes in environmental state, the mapping between risk data and sensor data is dynamically adjusted. The algorithm considers sensor weights and topology. Vehicle behavior feedback refers to the vehicle's final actual state, such as risk confirmation, false alarms, and missed alarms. Environmental state changes refer to alterations in external conditions, such as from sunny to rainy, from highway to congested city roads, or from daytime to nighttime. The algorithm optimizes the analytical data relationships through an adversarial game between the generator and discriminator. If the generator can consistently fool the discriminator, it indicates that the generator's data represents a blind spot in the current analytical data relationships. These data are then used as negative samples to backpropagate errors and adjust the connection weights in the topology. Simultaneously, the discriminator's discrimination process is analyzed, calculating the contribution of each sensor's data in correctly identifying true and false data. Sensor data with low contribution has its connection weight reduced, while data with high contribution has its weight increased.

[0036] Furthermore, step S160 also includes step S161, constructing a generator network and a discriminator network respectively, and combining the generator network and the discriminator network into an adversarial training framework; step S162, using historical sample data to train and converge the adversarial training framework to obtain an adversarial network, wherein, based on the data feedback generated by the adversarial network, weak links in the mapping topology are identified, wherein, the simulated data generated by the generator network is injected into the parsed data relationship for risk assessment, if the risk assessment confidence is lower than a set risk threshold, it is determined that the current mapping topology has a weak link in processing this type of data pattern, and the contribution of each sensor data to risk identification is calculated according to the discrimination accuracy of the discriminator network, and the connection weights in the topology are dynamically adjusted based on the contribution to optimize the parsed data relationship.

[0037] Step S162 further includes: collecting sensor data sequences under normal driving conditions from historical data as positive samples, and collecting sensor data sequences before known risk events occur as negative samples; inputting vehicle state parameters as conditional information along with the data sequences into the network, and alternately training the generator network and the discriminator network so that the generator network can generate simulated data that conforms to the real data distribution but contains potential risk characteristics; injecting the high-quality simulated data generated by the generator into the mapping topology training process to enhance the model's ability to identify edge cases until the convergence target is reached, thereby obtaining the adversarial network.

[0038] Preferably, a generator network is constructed, which takes random noise vectors and conditional information, such as vehicle state parameters like current vehicle speed, road type, and weather, as input. It learns the distribution of real historical data to simulate various potential risk scenarios and outputs realistic multi-sensor data sequences that look real but may contain rare, dangerous, or "edge case" features that existing models fail to identify well. A discriminator network is constructed, which takes data sequences that may come from real historical data or be fabricated by the generator as input. It distinguishes between real and generated data as accurately as possible and outputs a probability value indicating whether the input data is "real" or "generated by the generator." Then, the generator network and the discriminator network are combined into an adversarial training framework, where the generator network and the discriminator network compete against each other, forcing both to continuously improve.

[0039] Preferably, historical sample data is acquired, with sensor data sequences under normal driving conditions in the historical data used as positive samples and sensor data sequences before known risk events in the historical data used as negative samples. The positive samples and corresponding vehicle state conditions are then input into the adversarial training framework. By alternately training the generator network and the discriminator network, convergence is achieved to reach equilibrium, enabling the generator network to generate simulated data that conforms to the real data distribution but contains potential risk characteristics. The high-quality simulated data generated by the generator is then injected into the mapping topology training process to enhance the model's ability to identify edge cases until the convergence target is reached, thus obtaining the adversarial network.

[0040] Preferably, the weak links in the mapping topology are identified based on data feedback generated by the adversarial network. This involves injecting simulated data generated by the generator network into the analytical data relationship for risk assessment. If the confidence level of the risk assessment is lower than a set risk threshold, it indicates that the risk model can be easily fooled, proving that the current mapping topology has a weakness in processing similar data patterns. For example, it may not correctly associate relevant sensor signals or assign too low a weight, leading to an underestimation or omission of risk. The risk threshold is set based on historical positive samples. All historical positive sample data are input into the current analytical data relationship model for risk assessment. The risk assessment score of each positive sample is recorded and statistically analyzed, such as calculating its mean and standard deviation to obtain the risk distribution, and then the risk threshold is set accordingly. The risk level is set near the upper limit of the normal safe data risk distribution. Then, based on the discrimination accuracy of the discriminator network (by calculating the gradient of different input features in the discriminator network or using an attention mechanism), the contribution of each sensor data channel to the discriminator's correct judgment is quantified. The connection weights in the topology are dynamically adjusted based on the contribution. For sensor data with high contribution, the connection weights between it and related risk nodes are increased in the mapping topology; for sensor data with low contribution, its connection weights are decreased to reduce its influence and noise interference. For weak links, simulated data from successfully deceiving the model is used as new training samples to fine-tune the mapping topology model, enabling it to correctly identify ignored risk patterns, ultimately optimizing the analytical data relationships.

[0041] Furthermore, step S100 also includes step S170, extracting risk data based on the parsed data relationship, obtaining multiple risk events by performing cluster analysis on the risk data, each risk event representing a security threat or potential dangerous scenario; step S180, tracing the sensor data relationship based on the parsed data relationship with the risk event as the center to obtain the sensor data path; step S190, performing sample fitting on the sensor data of each node based on the sensor data path, establishing the functional relationship between the sensor data of each node and the risk event, performing security impact factor decomposition, obtaining the minimum cut set of the impact factor combination, and determining the security impact factor.

[0042] Preferably, all identified independent and fused analytical risk data are extracted from the analytical data relationships. K-means, DBSCAN, and other clustering analysis methods are used to perform clustering analysis on the risk data, classifying similar or frequently occurring risks into higher-level, representative, scenario-based risk events, thus obtaining multiple risk events. Each risk event represents a safety threat or potential dangerous scenario, such as rear-end collision risk events, lane departure risk events, and loss of control risk events. Then, a risk event is selected as the center for sensor data relationship tracing. That is, starting from the risk event node in the analytical data relationships, all data paths that lead to the risk triggered by the event are traced backward to determine the complete causal chain leading to the risk event. Finally, the sensor data path is obtained to show the step-by-step transmission status of the risk event, starting from the lowest level of raw sensor data readings. Then, based on the sensor data path, sample fitting is performed on the sensor data of each node. This includes collecting a large number of historical data samples, analyzing the relationship between the data distribution of the bottom-level sensor data nodes and the probability of the final risk event, verifying the mathematical relationship between the input and output of the intermediate calculation nodes, and quantifying the contribution of each link in the path to the final risk. A functional relationship between the sensor data of each node and the risk event is established. For example, through data fitting, it is found that the probability of a rear-end collision risk event is inversely proportional to the distance to the vehicle in front and positively correlated with the vehicle speed. Then, safety impact factor decomposition is performed. Based on the established functional relationship, all sensor data paths are analyzed to identify all abnormal combinations of sensor data that may lead to risk events. Sensitivity analysis is performed on these combinations to identify the combination with the fewest occurrences that will inevitably lead to risk from all abnormal combinations of sensor data. This combination is the minimum cut set of the impact factor combination. Finally, the basic sensor data contained in the minimum cut set is determined as the safety impact factor of the risk event.

[0043] Step S200: Based on the safety impact factors, perform time-series pattern recognition and establish the time-series trend relationship of the safety impact factors.

[0044] Preferably, historical data streams corresponding to safety impact factors are obtained and temporal pattern recognition is performed to identify dynamic pattern types, such as long-term upward or downward trends in safety impact factors, patterns that recur periodically / seasonally in specific scenarios, abrupt changes / drastic changes in safety impact factor values, and the impact of safety impact factors from one moment to the next. Using a long short-term memory network, a temporal trend relationship of safety impact factors is established based on the pattern recognition results, i.e., a multi-scenario safety temporal model, which is used to describe the dynamic relationship between past and present safety impact factors and can predict the trajectory of changes in safety impact factors in the near future. At the same time, the dynamic state of the vehicle is associated with the safety impact factors, and the probability of state transition is predicted.

[0045] Step S200 further includes step S210, extracting multi-scenario safety events with independent parameters and fused parameters based on the safety impact factors, fitting the time-series response relationship of safety impact factors for each scenario safety event, and obtaining a multi-scenario safety time-series model; step S220, collecting vehicle running time-series data based on the multi-scenario safety events, and constructing a positive example verification set and a negative example verification set; step S230, verifying the parameters of the multi-scenario safety time-series model through the positive example verification set and the negative example verification set, and using the verification feedback data to perform time-series iterative updates to the multi-scenario safety time-series model.

[0046] Preferably, the multi-scenario safety events are based on extracting independent parameters from safety impact factors, i.e., extreme or dangerous trends based on a single safety impact factor, such as a continuous and rapid decrease in the distance to the vehicle in front; and multi-scenario safety events are based on extracting fused parameters from safety impact factors, i.e., complex pattern scenarios based on multiple safety impact factors, such as increased lateral displacement and insufficient yaw rate in a curve. Then, for each scenario safety event, a large amount of historical data is collected and the time-series response relationship of safety impact factors is fitted, i.e., the changes of safety impact factors in a period of time before and after the event occur are analyzed, and a safety time-series model is established for each scenario to describe the time-series evolution law of safety impact factors in that scenario, thus obtaining a multi-scenario safety time-series model, where each safety time-series model corresponds to a specific scenario.

[0047] Preferably, vehicle runtime sequence data is collected based on multiple safety events. This involves continuously collecting time-series data including vehicle status, environmental information, and final results in real roads or simulation tests. A positive example validation set is constructed using the data sequence that ultimately leads to the safety event. For example, a positive example for validating the lane departure model is a real data sequence from when the vehicle starts crossing the line to when it completely deviates from the lane. A negative example validation set is constructed using data sequences that show similar risk signs but are successfully avoided by the driver's correct intervention. For example, a negative example for validating the lane departure model is a data sequence where the vehicle is about to cross the line but is corrected by steering wheel to return to the center of the lane. Then, the parameters of the multi-scenario safety time-series model are validated. The data from the positive and negative example validation sets are input into the corresponding multi-scenario safety time-series model to evaluate the predictive ability for positive examples and the ability to distinguish negative examples. Validation feedback data is obtained, such as prediction errors and false alarms / false negatives. The multi-scenario safety time-series model is then iteratively updated based on the validation feedback data, including retraining the safety time-series model to adapt it to more diverse driving styles and road conditions and output more realistic scenarios.

[0048] Step S300: The vehicle CAN data and intelligent driving system data are evaluated for state timing based on the time-series trend relationship to obtain dynamic risk assessment information.

[0049] Step S300 further includes step S310, which involves segmenting the vehicle CAN data and intelligent driving system data according to multiple time scales to form a multi-scale data sequence, and associating the data sequence at each time scale with the corresponding real-time total weight of the vehicle, historical health data of key components, and the current driving scenario to form an analysis unit with contextual information; step S320, which involves inputting the analysis unit into the time-series trend prediction model corresponding to each time scale to obtain risk identification prediction data for each time scale; step S330, which involves aligning the risk identification prediction data for each time scale in a time sequence, and calculating risk response based on the aligned data according to multiple scenario safety events to obtain the dynamic risk assessment information. The dynamic risk assessment information is a quantified dynamic risk value and includes structured dynamic risk assessment information containing the main risk sources, risk evolution trajectory, and suggested intervention measures.

[0050] Preferably, multiple time scales are configured, including short time scales such as a 100-millisecond window, medium time scales such as a 1-second window, and long time scales such as a 5-10 second window. The real-time incoming CAN data and intelligent driving system data are segmented according to different time windows to form a multi-scale data sequence. The short time scale is used to capture instantaneous and urgent risks, such as instantaneous tire lock-up, ESP triggering, and emergency braking; the medium time scale is used to identify trend risks, such as a continuous decrease in the distance to the vehicle in front and a steady increase in lateral deviation; the long time scale is used to assess risks in the macro situation, such as continuous acceleration demand caused by changes in the curvature of the road ahead and micro-swaying of the vehicle body caused by long-term fatigue driving. Then, the data sequence at each time scale is correlated with the corresponding real-time total vehicle weight, historical health data of key components, and the current driving scenario. The real-time total vehicle weight affects the vehicle's inertia, braking distance, and handling characteristics, with large and heavy vehicles having a high risk level. Historical health data of key components, such as brake pad wear and tire pressure trends, affect the reliability and performance ceiling of the system. The current driving scenario, such as congestion, rain, snow, and curves, determines the threshold and priority of risk assessment. This forms an analysis unit with contextual information, ensuring that the data analysis closely reflects actual working conditions.

[0051] Preferably, the analysis unit is input into the multi-scenario safety time series model corresponding to the time scale to obtain risk identification and prediction data for each time scale. The short-scale model may predict that there is a risk of brake pressure overflow within the next 200ms, the medium-scale model may predict that the probability of collision with the vehicle in front will increase to 80% within the next 2 seconds, and the long-scale model may predict that the risk of overheating will increase within the next 10 seconds due to continuous curves and continuous high load operation of the vehicle.

[0052] Preferably, the risk identification and prediction data at each time scale are time-aligned and synchronized on a unified time axis to avoid contradictions or overlaps in prediction results at different scales. Then, based on the aligned data, risk response calculations are performed according to multiple safety events to generate dynamic risk assessment information. The dynamic risk assessment information is a structured dynamic risk assessment information that includes quantified dynamic risk values ​​and contains the main risk sources, risk evolution trajectories, and suggested intervention measures. The quantified dynamic risk value is an overall score that integrates risks at all time scales and in all scenarios, used to quickly perceive the risk level. The main risk sources are used to clearly identify the most important risks, such as the main risk - rear-end collision - contribution rate 70%, and the secondary risk - oversteering - contribution rate 20%. The risk evolution trajectory describes the development of the risk and its most likely future direction. The suggested intervention measures are based on the source and evolution trajectory of the risk, proposing specific and prioritized response suggestions, such as the primary measure of immediately implementing automatic emergency braking and the secondary measure of reminding the driver to hold the steering wheel firmly.

[0053] Step S400: Map the vehicle status according to the dynamic risk assessment information to generate a vehicle dynamic assessment result.

[0054] Preferably, mapping vehicle status refers to making a final judgment based on a preset rule base and dynamic risk assessment information. The preset rule base is a risk-status mapping lookup table. For example, if the dynamic risk value is >90 and the main risk source is a rear-end collision, and the risk evolution trajectory deteriorates rapidly, it is mapped to an emergency state; if the dynamic risk value is between 70 and 90 and the main risk source is lane departure, it is mapped to a high-risk state; if the dynamic risk value is between 40 and 70 and the risk evolution trajectory is slowly increasing, it is mapped to a medium-risk state; if the dynamic risk value is <40, it is mapped to a low-risk / normal state. Finally, a vehicle dynamic assessment result is generated and output, including the overall status level and specific control strategy suggestions, to ensure dynamic safety assessment of vehicle status, improve the accuracy and foresight of status assessment, and enhance the safety and experience of human-machine co-driving.

[0055] In the above text, refer to Figure 1 A method for assessing vehicle dynamic state by fusing multi-sensor information according to an embodiment of the present invention is described in detail. Next, reference will be made to... Figure 2 A vehicle dynamic state assessment system that integrates multi-sensor information is described according to an embodiment of the present invention.

[0056] According to an embodiment of the present invention, a vehicle dynamic state assessment system integrating multi-sensor information is used to solve the technical problems in the prior art, namely, the difficulty in integrating multi-source vehicle information and the inability of static assessment to capture the dynamic risk evolution trend of the vehicle, resulting in insufficient accuracy and foresight in vehicle state assessment. This system achieves the technical effect of realizing dynamic safety assessment of vehicle state and improving the accuracy and foresight of state assessment. Figure 2 As shown, a vehicle dynamic state assessment system integrating multi-sensor information includes: a data parsing and decomposition module 10, a temporal pattern recognition module 20, a state temporal evaluation module 30, and a vehicle state mapping module 40.

[0057] The data parsing and decomposition module 10 is used to acquire vehicle CAN data and intelligent driving system data, perform independent and fused parsing of the vehicle CAN data and intelligent driving system data, and perform safety response decomposition based on the parsed data relationships to obtain safety impact factors; the time sequence pattern recognition module 20 is used to perform time sequence pattern recognition based on the safety impact factors and establish the time sequence trend relationship of the safety impact factors; the state time sequence evaluation module 30 is used to perform state time sequence evaluation of the vehicle CAN data and intelligent driving system data through the time sequence trend relationship to obtain dynamic risk assessment information; the vehicle state mapping module 40 is used to map the vehicle state according to the dynamic risk assessment information to generate vehicle dynamic evaluation results.

[0058] The specific configuration of the data parsing and decomposition module 10 will be described in detail below. The data parsing and decomposition module 10 further includes: extracting vehicle control sensor data based on the vehicle CAN data according to the time dimension and driving scenario state dimension; obtaining autonomous driving module sensor data, environmental perception data, path planning data, and vehicle status monitoring data based on the intelligent driving system data; performing risk state analysis on the vehicle control sensor data, autonomous driving module sensor data, environmental perception data, path planning data, and vehicle status monitoring data respectively to determine independently analyzed risk data; and analyzing the collaborative risk interaction relationship among the vehicle control sensor data, autonomous driving module sensor data, environmental perception data, path planning data, and vehicle status monitoring data to establish fused analyzed risk data.

[0059] The specific configuration of the data parsing and decomposition module 10 will be described in detail below. The data parsing and decomposition module 10 further includes: establishing an independent parsing mapping topology based on the independently parsed risk data and according to the mapping relationship between risk data and sensor data; adding the fused risk relationship to the independently parsed mapping topology based on the fused parsed risk data and according to the mapping relationship between risk data and sensor data; and performing hierarchical expansion of the mapping topology to obtain the parsed data relationship.

[0060] The specific configuration of the data parsing and decomposition module 10 will be described in detail below. The data parsing and decomposition module 10 further includes: extracting risk data based on the parsed data relationships; obtaining multiple risk events by performing cluster analysis on the risk data, each risk event representing a security threat or potential dangerous scenario; tracing sensor data relationships based on the parsed data relationships, centered on the risk events, to obtain sensor data paths; performing sample fitting on the sensor data of each node based on the sensor data paths, establishing a functional relationship between the sensor data of each node and the risk events, performing security impact factor decomposition, obtaining the minimum cut set of the impact factor combination, and determining the security impact factors.

[0061] The specific configuration of the data parsing and decomposition module 10 will be described in detail below. The data parsing and decomposition module 10 further includes: introducing a generative adversarial network model to dynamically adjust the mapping weights and topology between risk data and sensor data based on actual vehicle behavior feedback and environmental state changes, thereby optimizing the relationships between the parsed data.

[0062] The specific configuration of the data parsing and decomposition module 10 will be described in detail below. The data parsing and decomposition module 10 further includes: constructing a generator network and a discriminator network respectively, and combining the generator network and the discriminator network into an adversarial training framework; training and converging the adversarial training framework using historical sample data to obtain an adversarial network; wherein, based on the data feedback generated by the adversarial network, weak links in the mapping topology are identified; wherein, simulated data generated by the generator network is injected into the parsed data relationship for risk assessment; if the risk assessment confidence is lower than a set risk threshold, it is determined that the current mapping topology has a weak link in processing this type of data pattern; and based on the discrimination accuracy of the discriminator network, the contribution of each sensor data to risk identification is calculated; and the connection weights in the topology are dynamically adjusted based on the contribution to optimize the parsed data relationship.

[0063] The specific configuration of the data parsing and decomposition module 10 will be described in detail below. The data parsing and decomposition module 10 further includes: collecting sensor data sequences under normal driving conditions from historical data as positive samples, and collecting sensor data sequences before known risk events occur as negative samples; inputting vehicle state parameters as conditional information, along with the data sequences, into the network to alternately train the generator network and the discriminator network, enabling the generator network to generate simulated data that conforms to the real data distribution but contains potential risk characteristics; injecting the high-quality simulated data generated by the generator into the mapping topology training process to enhance the model's ability to identify edge cases, until the convergence target is reached, thus obtaining the adversarial network.

[0064] The specific configuration of the temporal pattern recognition module 20 will be described in detail below. The temporal pattern recognition module 20 further includes: extracting multi-scenario safety events based on the safety impact factors using independent and fused parameters; fitting the temporal response relationship of the safety impact factors to each scenario's safety events to obtain a multi-scenario safety temporal model; collecting vehicle runtime timing data based on the multi-scenario safety events to construct a positive example validation set and a negative example validation set; verifying the parameters of the multi-scenario safety temporal model using the positive and negative example validation sets; and using the verification feedback data to perform temporal iterative updates to the multi-scenario safety temporal model.

[0065] The specific configuration of the state timing evaluation module 30 will be described in detail below. The state timing evaluation module 30 further includes: segmenting the vehicle CAN data and intelligent driving system data according to multiple time scales to form multi-scale data sequences; associating the data sequence at each time scale with the corresponding real-time total vehicle weight, historical health data of key components, and the current driving scenario to form an analysis unit with contextual information; inputting the analysis unit into the timing trend prediction model corresponding to each time scale to obtain risk identification prediction data for each time scale; aligning and arranging the risk identification prediction data for each time scale in a time sequence; and calculating risk response based on the aligned data according to multi-scenario safety events to obtain the dynamic risk evaluation information. The dynamic risk evaluation information is a quantified dynamic risk value and includes structured dynamic risk evaluation information containing the main risk sources, risk evolution trajectory, and suggested intervention measures.

[0066] The vehicle dynamic state assessment system that integrates multi-sensor information provided in this embodiment of the invention can execute the vehicle dynamic state assessment method that integrates multi-sensor information provided in any embodiment of the invention, and has the corresponding functional modules and beneficial effects of the method.

[0067] Although this application makes various references to certain modules in the system according to the embodiments of this application, any number of different modules can be used and run on user terminals and / or servers. The various units and modules included are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be achieved; in addition, the specific names of each functional unit are only for easy distinction between each other and are not used to limit the scope of protection of this invention.

[0068] The specific embodiments described above do not constitute a limitation on the scope of protection of this application. Those skilled in the art should understand that various modifications, combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A method for assessing the dynamic state of a vehicle by integrating information from multiple sensors, characterized in that, include: Acquire vehicle CAN data and intelligent driving system data, perform independent and fusion analysis on the vehicle CAN data and intelligent driving system data, decompose safety response based on the analysis data relationship, and obtain safety impact factors; Based on the aforementioned safety impact factors, time-series pattern recognition is performed to establish the time-series trend relationship of the safety impact factors; The vehicle CAN data and intelligent driving system data are evaluated for state timing based on the time-series trend relationship to obtain dynamic risk assessment information. The vehicle status is mapped based on the dynamic risk assessment information to generate a vehicle dynamic assessment result; Among them, security response decomposition based on analytical data relationships yields security impact factors, including: Based on the parsed data relationships, risk data is extracted, and multiple risk events are obtained by clustering the risk data. Each risk event represents a security threat or potential dangerous scenario. Centered on risk events, sensor data relationships are traced based on the analyzed data relationships to obtain sensor data paths; Based on the aforementioned sensing data path, sample fitting is performed on the sensing data of each node to establish a functional relationship between the sensing data of each node and risk events. Safety impact factors are decomposed to obtain the minimum cut set of the combination of impact factors and to determine the safety impact factors. Specifically, based on the safety impact factors, time-series pattern recognition is performed to establish the time-series trend relationship of the safety impact factors, including: Based on the security impact factors, multi-scenario security events are extracted using independent parameters and fused parameters. The time-series response relationship of security impact factors is fitted to the security events in each scenario to obtain a multi-scenario security time-series model. Based on the aforementioned multi-scenario safety events, vehicle runtime sequence data is collected to construct a positive example verification set and a negative example verification set. The parameters of the multi-scenario security time series model are verified using the positive example verification set and the negative example verification set, and the time series update of the multi-scenario security time series model is performed using the verification feedback data. Specifically, dynamic risk assessment information is obtained by evaluating the state timing of the vehicle CAN data and intelligent driving system data based on the aforementioned time-series trend relationship, including: The vehicle CAN data and intelligent driving system data are segmented according to multiple time scales to form a multi-scale data sequence. The data sequence at each time scale is then associated with the corresponding real-time total weight of the vehicle, historical health data of key components, and the current driving scenario to form an analysis unit with contextual information. The analysis unit is input into the time series trend prediction model corresponding to the time scale to obtain risk identification and prediction data for each time scale. The risk identification and prediction data at each time scale are time-aligned and arranged. Based on the aligned data, risk response calculations are performed according to multiple security events to obtain the dynamic risk assessment information. The dynamic risk assessment information is a structured dynamic risk assessment information that is a quantified dynamic risk value and includes the main risk sources, risk evolution trajectory, and suggested intervention measures.

2. The vehicle dynamic state assessment method integrating multi-sensor information according to claim 1, characterized in that, Acquire vehicle CAN data and intelligent driving system data, and perform independent and fused analysis on the vehicle CAN data and intelligent driving system data, including: Based on the vehicle CAN data, vehicle control sensor data is extracted according to the time dimension and the driving scenario state dimension. Based on the data from the intelligent driving system, obtain autonomous driving module sensing data, environmental perception data, path planning data, and vehicle status monitoring data; Risk status analysis is performed on the vehicle control sensor data, autonomous driving module sensor data, environmental perception data, path planning data, and vehicle status monitoring data respectively to determine the risk data to be analyzed independently. The collaborative risk interaction relationships among the vehicle control sensor data, autonomous driving module sensor data, environmental perception data, path planning data, and vehicle status monitoring data are analyzed to establish a fusion and analysis risk data.

3. The vehicle dynamic state assessment method integrating multi-sensor information according to claim 2, characterized in that, Also includes: Based on the independently parsed risk data, a mapping topology relationship for independent parsing is established according to the mapping relationship between risk data and sensor data; Based on the fused and parsed risk data, according to the mapping relationship between risk data and sensor data, the fused risk relationship is added to the independently parsed mapping topology relationship, and the mapping topology relationship is hierarchically expanded to obtain the parsed data relationship.

4. The vehicle dynamic state assessment method integrating multi-sensor information according to claim 3, characterized in that, Obtaining the parsed data relationship also includes: A generative adversarial network model is introduced to dynamically adjust the mapping weights and topology between risk data and sensor data based on actual vehicle behavior feedback and changes in environmental state, thereby optimizing the relationship between the parsed data.

5. The vehicle dynamic state assessment method integrating multi-sensor information according to claim 4, characterized in that, Optimizing the parsed data relationships includes: A generator network and a discriminator network are constructed separately, and the generator network and the discriminator network are combined into an adversarial training framework; The adversarial training framework is trained and converged using historical sample data to obtain an adversarial network. Based on the data feedback generated by the adversarial network, weak links in the mapping topology are identified. Specifically, simulated data generated by the generator network is injected into the parsed data relationship for risk assessment. If the risk assessment confidence is lower than a set risk threshold, it is determined that the current mapping topology has a weak link in processing this type of data pattern. Based on the discrimination accuracy of the discriminator network, the contribution of each sensor data to risk identification is calculated. The connection weights in the topology are dynamically adjusted based on the contribution to optimize the parsed data relationship.

6. The vehicle dynamic state assessment method integrating multi-sensor information according to claim 5, characterized in that, The adversarial training framework is trained and converged using historical sample data to obtain an adversarial network, including: Sensor data sequences under normal driving conditions are collected from historical data as positive samples, and sensor data sequences before known risk events occur are collected as negative samples. Vehicle state parameters are used as conditional information and input into the network along with the data sequence. The generator network and discriminator network are trained alternately so that the generator network can generate simulated data that conforms to the real data distribution but contains potential risk characteristics. The high-quality simulated data generated by the generator is injected into the mapping topology training process to enhance the model's ability to identify edge cases until the convergence target is reached, thus obtaining the adversarial network.

7. A vehicle dynamic state assessment system integrating multi-sensor information, characterized in that, The system is used to implement the vehicle dynamic state assessment method integrating multi-sensor information as described in any one of claims 1 to 6, the system comprising: The data parsing and decomposition module is used to acquire vehicle CAN data and intelligent driving system data, perform independent parsing and fusion parsing of the vehicle CAN data and intelligent driving system data, and perform safety response decomposition based on the parsed data relationships to obtain safety impact factors. The time-series pattern recognition module is used to perform time-series pattern recognition based on the security impact factors and establish the time-series trend relationship of the security impact factors. The state timing evaluation module is used to perform state timing evaluation on the vehicle CAN data and intelligent driving system data through the timing trend relationship to obtain dynamic risk evaluation information. The vehicle status mapping module is used to map the vehicle status based on the dynamic risk assessment information and generate a vehicle dynamic assessment result.

Images