Fault Decision-Making Method and Device Based on Rear Vehicle Information

By combining fault diagnosis results with information from vehicles behind, the fault response strategy is dynamically adjusted, which solves the problems of rigid response and insufficient risk assessment in high-level autonomous vehicles during fault degradation, thereby improving safety and traffic efficiency.

CN123104699APending Publication Date: 2026-11-17CHERY AUTOMOBILE CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611593978.2
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-10-14
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

Existing advanced autonomous vehicles fail to comprehensively consider the distance and relative speed of vehicles approaching from behind during fault degradation, resulting in rigid responses, insufficient risk assessment, and reduced traffic efficiency.

Method used

By combining fault diagnosis results with information about vehicles behind, the system determines whether to immediately pull over or choose an alternative strategy based on the rear-end collision risk value, and dynamically adjusts the fault response strategy to adapt to the current traffic situation.

Benefits of technology

It improves the accuracy of identifying the degree of danger in fault conditions, avoids secondary risks caused by blindly stopping, enhances the safety and adaptability of fault downgrade handling, and improves traffic continuity and decision-making rationality.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123104699A_ABST
    Figure CN123104699A_ABST
Patent Text Reader

Abstract

The fault decision-making method and apparatus based on rear vehicle information provided in this application include: when a vehicle fault is detected, acquiring the vehicle's fault diagnosis result and rear vehicle information, wherein the fault diagnosis result includes: fault type, fault level, and the vehicle's current controllable state; the rear vehicle information includes: the distance and relative speed between the nearest rear vehicle and the vehicle; calculating a rear-end collision risk value for the vehicle to immediately pull over and stop based on the rear vehicle information; performing a pullover action if the rear-end collision risk value is less than a preset threshold; and determining an alternative strategy based on the fault diagnosis result and executing the alternative strategy if the rear-end collision risk value is greater than or equal to the preset threshold.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This application relates to the field of autonomous vehicle control, and in particular to a fault decision-making method and apparatus based on information about vehicles behind. Background Technology

[0002] In scenarios such as highways and urban expressways, highly automated vehicles typically employ a combination of fault diagnosis and minimum risk maneuver (MRM) for safety degradation control.

[0003] However, existing solutions mostly rely on static mapping of fault types to preset actions, without taking into account the distance and relative speed of vehicles approaching from behind. Furthermore, the strategies are mostly limited to parking-related actions, which can easily lead to rigid responses, insufficient risk assessment, and reduced traffic efficiency when facing high-speed approaching traffic.

[0004] Therefore, it is urgent to solve the problem of how to comprehensively reflect the traffic situation behind the fault during the fault degradation process and achieve safe and appropriate strategy selection. Summary of the Invention

[0005] This application provides a fault decision-making method and apparatus based on rear vehicle information to solve the aforementioned technical problems. When a vehicle malfunctions, this solution combines fault diagnosis results with rear vehicle information to comprehensively assess the traffic situation behind the vehicle. Based on this assessment, it decides whether to immediately pull over or adopt alternative strategies. This allows fault degradation control to move beyond a static response based solely on the fault type, and instead incorporates a strategy selection that better reflects the actual road risks, taking into account the vehicle's current controllable state and the situation of vehicles approaching from behind.

[0006] In a first aspect, this application provides a fault decision-making method based on rear vehicle information, including:

[0007] When a vehicle malfunction is detected, the fault diagnosis results and information about vehicles behind are obtained. The fault diagnosis results include: fault type, fault level, and current controllable status of the vehicle. The information about vehicles behind includes: distance and relative speed between the nearest vehicles.

[0008] Calculate the rear-end collision risk value of the vehicle if it must immediately pull over and stop based on information about vehicles behind it.

[0009] If the rear-end collision risk value is less than the preset threshold, the vehicle will pull over to the side of the road.

[0010] If the rear-end collision risk value is greater than or equal to a preset threshold, an alternative strategy is determined based on the fault diagnosis results and the alternative strategy is executed.

[0011] In one possible embodiment, before obtaining the vehicle's fault diagnosis results and information about vehicles behind, the method further includes:

[0012] Traffic flow features are collected through multi-source heterogeneous sensing devices, and the traffic flow features are time-aligned through a high-precision timestamp synchronization and alignment mechanism.

[0013] The aligned traffic flow features are filtered to remove abnormal transition points and false alarm targets.

[0014] The filtered traffic flow features are then spatially transformed and aligned with the time series.

[0015] In one possible embodiment, obtaining the vehicle's fault diagnosis results and information about vehicles behind it includes:

[0016] Calculate the real-time confidence weights of multiple multi-source heterogeneous sensing devices based on the vehicle's current driving conditions;

[0017] The aligned traffic flow features are linearly weighted and fused based on real-time confidence weights to obtain information about vehicles behind.

[0018] In one possible embodiment, obtaining the vehicle's fault diagnosis results and information about vehicles behind it includes:

[0019] Periodically acquire heart rate data from multiple electronic control units of the vehicle, as well as fault codes output by the unified diagnostic service;

[0020] Fault diagnosis is performed on multiple electronic control units based on heart rate data, and fault data is determined based on fault codes;

[0021] Fault diagnosis results are generated based on the fault determination results and fault data.

[0022] In one possible embodiment, calculating the rear-end collision risk value for an immediate pullover based on information about vehicles behind includes:

[0023] Invoke the preset rear-end collision risk quantification model;

[0024] The rear-end collision risk quantification model uses mathematical mapping calculations to map distance and relative speed to generate rear-end collision risk values.

[0025] In one possible embodiment, after calculating the rear-end collision risk value for the vehicle to immediately pull over based on information about vehicles behind, the method further includes:

[0026] Real-time reading of vehicle tire pressure and longitudinal acceleration; calculation of current road friction coefficient.

[0027] The actual achievable deceleration boundary is calculated based on the real-time load of the vehicle and the health status of the braking system.

[0028] The rear-end collision risk value is adaptively adjusted based on the current road friction coefficient and deceleration boundary.

[0029] In one possible embodiment, a rear-end collision risk value is generated by mathematically mapping distance and relative speed using a rear-end collision risk quantification model, including:

[0030] Using the formula R=α*v 2 / d calculates the rear-end collision risk value R, where α is the calibration coefficient obtained by fitting real vehicle collision test data, v is the relative speed, and d is the distance.

[0031] In one possible embodiment, after calculating the rear-end collision risk value for the vehicle to immediately pull over based on information about vehicles behind, the method further includes:

[0032] Lane environment features are extracted based on real-time lane geometry width and traffic density of adjacent lanes;

[0033] The basic safety threshold is flexibly fine-tuned based on the characteristics of the lane environment to obtain the preset threshold.

[0034] In one possible embodiment, obtaining the vehicle's fault diagnosis results and information about vehicles behind it includes:

[0035] If the fault diagnosis result indicates a rearward sensor failure, the information of vehicles behind is predicted by the side and rearward sensors, where the rearward sensor is used to directly acquire the information of vehicles behind.

[0036] In one possible embodiment, after predicting rear vehicle information using side and rear sensors, the method further includes:

[0037] If the prediction of rear vehicle information by the side and rear sensors fails, the rear-end collision risk value is set to a first threshold, wherein the first threshold is greater than a preset threshold.

[0038] In one possible embodiment, determining an alternative strategy based on the fault diagnosis results includes:

[0039] Based on the fault diagnosis results and the information of vehicles behind, multiple preset strategies are pruned for feasibility, resulting in multiple feasible preset strategies.

[0040] Collision risk values ​​are predicted for multiple feasible preset strategies, resulting in multiple corresponding collision risk values.

[0041] An alternative strategy is determined from multiple feasible preset strategies based on multiple collision risk values.

[0042] In one possible embodiment, multiple preset strategies are pruned based on fault diagnosis results and information about vehicles behind, including:

[0043] If the fault diagnosis result indicates that the vehicle is not allowed to continue driving, remove the options from the preset strategies of slowing down to the service area and maintaining the current lane to the service area. The preset strategies include: slowing down and stopping in the current lane, slowing down to the service area, and maintaining the current lane to the service area.

[0044] In one possible embodiment, multiple preset strategies include: decelerating and stopping in the current lane, gradually moving to the service area, and maintaining the current lane while continuing towards the service area; predicting collision risk values ​​for multiple feasible preset strategies to obtain multiple corresponding collision risk values, including:

[0045] The minimum probability of being rear-ended is calculated based on the information of vehicles behind in the current lane, and the collision risk value of decelerating and stopping in the current lane is calculated based on the minimum value. The probability of being rear-ended is related to the deceleration speed.

[0046] The collision risk value for slowing down to the service area is calculated based on the first unit time driving risk, the first estimated driving time, and the additional driving risk. The first unit time driving risk is calculated based on the information of vehicles behind and the slowing speed.

[0047] The collision risk value for maintaining the current lane and heading towards the service area is calculated based on the second unit time driving risk, the second estimated driving time, and the additional risk of fault deterioration. The second unit time driving risk is calculated based on the information of vehicles behind and the reduced driving speed.

[0048] In one possible embodiment, an alternative strategy is determined from multiple feasible preset strategies based on multiple collision risk values, including:

[0049] If a single minimum value exists among multiple collision risk values, the feasible preset strategy corresponding to the minimum value is determined as the alternative strategy;

[0050] When there are N minimum values ​​among multiple collision risk values, the target feasible pre-defined strategy with the highest conservatism is determined from the feasible pre-defined strategies corresponding to the N minimum values ​​according to the conservatism ranking of multiple feasible pre-defined strategies, and the target feasible pre-defined strategy is determined as the alternative strategy.

[0051] Secondly, embodiments of this application provide a fault decision-making device based on rear vehicle information, including:

[0052] The acquisition module is used to acquire the vehicle's fault diagnosis results and information about vehicles behind it when a vehicle fault is detected. The fault diagnosis results include: fault type, fault level, and the current controllable status of the vehicle. The information about vehicles behind it includes: the distance and relative speed between the nearest vehicles behind it.

[0053] The calculation module is used to calculate the rear-end collision risk value of the vehicle if it immediately pulls over to the side of the road based on the information of the vehicles behind it;

[0054] The first execution module is used to execute the action of pulling over to the side of the road when the rear-end collision risk value is less than a preset threshold.

[0055] The second execution module is used to determine an alternative strategy based on the fault diagnosis results and execute the alternative strategy when the rear-end collision risk value is greater than or equal to a preset threshold.

[0056] Thirdly, embodiments of this application provide an electronic device, including: a memory and a processor;

[0057] The memory stores instructions that the computer executes;

[0058] The processor executes computer execution instructions stored in memory, causing the processor to perform the methods described above.

[0059] Fourthly, embodiments of this application provide a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, are used to implement the methods provided above.

[0060] Fifthly, embodiments of this application provide a computer program product, including a computer program that, when executed by a processor, implements the method described above.

[0061] This application, by acquiring fault diagnosis results and information about vehicles behind when a vehicle malfunctions, and assessing the rear-end collision risk of immediately pulling over based on the rear-end vehicle information, incorporates traffic situation factors such as the distance and relative speed of vehicles behind into the fault decision-making process, thereby improving the accuracy of identifying the current level of danger. By executing a pullover action when the rear-end collision risk is below a preset threshold, and determining an alternative strategy based on the fault diagnosis results when the rear-end collision risk reaches the preset threshold, it can avoid the secondary risks caused by blindly stopping in high-speed approach traffic scenarios, while improving the safety and adaptability of fault degradation handling, thereby improving the continuity of traffic and the rationality of overall decision-making when the vehicle is in a fault state. Attached Figure Description

[0062] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0063] Figure 1 This application provides a schematic diagram of an application data processing system architecture.

[0064] Figure 2 A flowchart illustrating the fault decision-making method based on rear vehicle information provided in this application embodiment. Figure 1 ;

[0065] Figure 3A flowchart illustrating the fault decision-making method based on rear vehicle information provided in this application embodiment. Figure 2 ;

[0066] Figure 4 A flowchart illustrating the fault decision-making method based on rear vehicle information provided in this application embodiment. Figure 3 ;

[0067] Figure 5 A flowchart illustrating the fault decision-making method based on rear vehicle information provided in this application embodiment. Figure 4 ;

[0068] Figure 6 A flowchart illustrating the fault decision-making method based on rear vehicle information provided in this application embodiment. Figure 5 ;

[0069] Figure 7 A flowchart illustrating the fault decision-making method based on rear vehicle information provided in this application embodiment. Figure 6 ;

[0070] Figure 8 A flowchart illustrating the fault decision-making method based on rear vehicle information provided in this application embodiment. Figure 7 ;

[0071] Figure 9 A flowchart illustrating the fault decision-making method based on rear vehicle information provided in this application embodiment. Figure 8 ;

[0072] Figure 10 A flowchart illustrating the fault decision-making method based on rear vehicle information provided in this application embodiment. Figure 9 ;

[0073] Figure 11 A flowchart illustrating the fault decision-making method based on rear vehicle information provided in this application embodiment. Figure 10 ;

[0074] Figure 12 A flowchart of an autonomous driving fault degradation decision-making method based on rear vehicle risk assessment provided in an embodiment of this application;

[0075] Figure 13 A schematic diagram of the fault decision-making device based on rear vehicle information provided in an embodiment of this application;

[0076] Figure 14 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application.

[0077] The accompanying drawings illustrate specific embodiments of this application, which will be described in more detail below. These drawings and descriptions are not intended to limit the scope of the concept in any way, but rather to illustrate the concept of this application to those skilled in the art through reference to particular embodiments. Detailed Implementation

[0078] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application as detailed in the appended claims.

[0079] Fault safety control for advanced autonomous vehicles falls under the category of autonomous driving and vehicle safety degradation, and is typically applied in dynamic, traffic-intensive scenarios such as highways and urban expressways. The system generally consists of an onboard fault diagnosis module, a rear-view perception unit, a decision control unit, and a drive-by-wire chassis actuator. Upon detecting an anomaly in the vehicle control system, it must quickly complete the strategy switching and command issuance related to minimum-risk maneuvers.

[0080] In existing technologies, fault response often employs a one-to-one correspondence between fault type and preset actions. Upon detecting an anomaly, the system directly executes actions such as pulling over, slowing down, or entering a designated area. The judgment is primarily based on the vehicle's internal fault state, failing to incorporate the distance and relative speed of the nearest following vehicle into the same decision-making process. When faced with a high-speed approaching vehicle, this static mapping mechanism struggles to promptly reflect the risk of rear-end collisions during lane changes or deceleration. Strategy selection is also largely focused on stopping actions, lacking the ability to switch response paths based on real-time traffic conditions. When traffic flow is simultaneously high-density both ahead and behind, mechanically executing preset actions can easily amplify the risk of secondary collisions and disrupt road traffic flow.

[0081] Based on the aforementioned technical issues, after a fault is triggered, the vehicle fault diagnosis results and information about vehicles behind are simultaneously incorporated into the same decision input. A rear-end collision risk assessment logic is established based on the distance and relative speed of the nearest vehicle behind, and the applicable boundaries of pulling over and alternative strategies are distinguished accordingly. For high-risk situations where a vehicle is approaching from behind, the decision-making process shifts to selecting an alternative handling path based on the fault level and the vehicle's current controllable state. This is then combined with the onboard perception and control links to complete the command output, transforming the fault response from a fixed action mapping to a branch decision based on real-time traffic conditions. This reduces the problems of misjudgment and blind handling in high-level autonomous vehicles under fault conditions.

[0082] The technical solution of this application and how the technical solution of this application solves the above-mentioned technical problems are described in detail below with specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments. The embodiments of this application will now be described with reference to the accompanying drawings.

[0083] Figure 1 This is a schematic diagram of an application data processing system architecture provided in an embodiment of this application. The application data processing system is a computer device. Figure 1 As shown, the above architecture includes at least one of a data acquisition device 101, a processing device 102, and a display device 103.

[0084] It is understood that the structures illustrated in the embodiments of this application do not constitute a specific limitation on the architecture of the application data processing system. In other feasible embodiments of this application, the above architecture may include more or fewer components than illustrated, or combine some components, or split some components, or arrange different components, which can be determined according to the actual application scenario and is not limited here. Figure 1 The components shown can be implemented in hardware, software, or a combination of both.

[0085] In the specific implementation process, the data acquisition device 101 may include an input / output interface or a communication interface, and the data acquisition device 101 can be connected to the processing device through the input / output interface or the communication interface.

[0086] When a vehicle malfunction is detected, the processing device 102 can acquire the vehicle's fault diagnosis results and information about vehicles behind it. The fault diagnosis results include the fault type, fault level, and the vehicle's current controllable state. The information about vehicles behind it includes the distance and relative speed between the nearest vehicles. Based on the information about vehicles behind it, the device calculates the rear-end collision risk value for the vehicle to immediately pull over. If the rear-end collision risk value is less than a preset threshold, the device performs the action of pulling over. If the rear-end collision risk value is greater than or equal to the preset threshold, the device determines an alternative strategy based on the fault diagnosis results and executes the alternative strategy.

[0087] The display device 103 can also be a touch screen or the screen of a terminal device, used to receive user commands while displaying the above-mentioned content, so as to realize interaction with the user.

[0088] It should be understood that the aforementioned processing device can be implemented by a processor reading instructions from memory and executing those instructions, or it can be implemented by a chip circuit.

[0089] Furthermore, the network architecture and business scenarios described in the embodiments of this application are for the purpose of more clearly illustrating the technical solutions of the embodiments of this application, and do not constitute a limitation on the technical solutions provided in the embodiments of this application. As those skilled in the art will know, with the evolution of network architecture and the emergence of new business scenarios, the technical solutions provided in the embodiments of this application are also applicable to similar technical problems.

[0090] Figure 2 A flowchart illustrating the fault decision-making method based on rear vehicle information provided in this application embodiment. Figure 1 ,like Figure 2 As shown, the fault decision-making method based on rear vehicle information provided in this embodiment includes:

[0091] S201. When a vehicle malfunction is detected, obtain the vehicle malfunction diagnosis results and information about vehicles behind. The malfunction diagnosis results include the malfunction type, malfunction level, and current controllable status of the vehicle. The information about vehicles behind includes the distance and relative speed between the nearest vehicles and the vehicles behind.

[0092] S202. Calculate the rear-end collision risk value of the vehicle if it must immediately pull over to the side of the road based on the information of the vehicles behind.

[0093] S203. If the rear-end collision risk value is less than the preset threshold, execute the action of pulling over to the side of the road.

[0094] S204. If the rear-end collision risk value is greater than or equal to the preset threshold, determine the alternative strategy based on the fault diagnosis results and execute the alternative strategy.

[0095] This application provides a fault decision-making method based on rear vehicle information. It integrates fault diagnosis results with the distance and relative speed of the nearest rear vehicle into the same decision input. A rear-end collision risk value is used to determine the applicable boundary for immediate pullover. When the risk reaches a threshold, an alternative strategy is selected based on the fault type, fault level, and the vehicle's current controllable state. This creates a branched decision-making process related to rear vehicle information, thereby constraining the risk to approaching vehicles after a fault occurs and ensuring that vehicle actions match the currently available controllable capabilities. It should be understood that the above example is merely illustrative and not limiting.

[0096] In one possible implementation, Figure 3 A flowchart illustrating the fault decision-making method based on rear vehicle information provided in this application embodiment. Figure 2 ,like Figure 3 As shown, before obtaining the vehicle's fault diagnosis results and information about vehicles behind, the method also includes:

[0097] S301 collects traffic flow features through multi-source heterogeneous sensing devices and aligns the traffic flow features in time through a high-precision timestamp synchronization and alignment mechanism;

[0098] S302, filter the aligned traffic flow features to remove abnormal jump points and false alarm targets;

[0099] S303 performs spatial coordinate transformation and time series alignment on the filtered traffic flow features.

[0100] Among them, multi-source heterogeneous sensing devices are used to collect traffic flow observations from different sources and of different types. These observations may include lane occupancy status, target location information, target motion status, and vehicle distribution information within the road neighborhood. A high-precision timestamp synchronization and alignment mechanism is used to attach a unified time reference to the data output by each sensing device, ensuring that the observation results within the same time window are consistent in time sequence, so that the features are comparable during subsequent fusion processing. Spatial coordinate transformation is used to uniformly map the features output by each sensing device in its own coordinate system to the vehicle coordinate system, road coordinate system, or a preset world coordinate system. Temporal series alignment is used to organize the filtered features into continuous time-series data to meet the requirements of spatiotemporal consistency when extracting subsequent vehicle information.

[0101] In practical implementation, traffic flow characteristics can be collaboratively acquired by millimeter-wave radar, lidar, cameras, and vehicle-mounted positioning units. The radar outputs target distance and radial velocity, the camera outputs lane boundaries and target categories, the lidar outputs spatial point cloud contours, and the positioning unit outputs vehicle pose information. The raw data from each sensing device is timestamped and then enters the synchronization module. The synchronization module aligns the data based on a unified clock or high-precision timing source, limiting the deviation to a preset millisecond range. Filtering can employ sliding window suppression, threshold removal, or continuous frame consistency verification to eliminate short-term increases in distance jitter, sudden speed changes, and false detections in single frames, while retaining valid traffic targets that meet the condition of continuous occurrence. Spatial coordinate transformation unifies the coordinates of different devices to the same reference system through an extrinsic parameter matrix. Then, the coordinates are rearranged according to the time sequence and interpolated to fill in missing frames, forming a continuous and stable traffic flow characteristic sequence.

[0102] This processing method ensures that the data entering subsequent fault diagnosis correlation analysis has a unified time scale, a unified spatial benchmark, and high target stability, thereby providing reliable input for the generation of rear vehicle information and improving the consistency and accuracy of the data used for subsequent rear-end collision risk calculation. By eliminating abnormal transition points and false alarm targets, the interference of transient noise on the identification of rear vehicles can be reduced, lowering the probability of mistakenly including invalid targets in the decision-making process.

[0103] In an optional embodiment, Figure 4A flowchart illustrating the fault decision-making method based on rear vehicle information provided in this application embodiment. Figure 3 ,like Figure 4 As shown, the system obtains the vehicle's fault diagnosis results and information about vehicles behind it, including:

[0104] S401 calculates the real-time confidence weights of multiple multi-source heterogeneous sensing devices based on the vehicle's current driving conditions.

[0105] S402, based on the real-time confidence weight, performs linear weighted fusion of the aligned traffic flow features to obtain the information of vehicles behind.

[0106] The real-time confidence weight is used to characterize the reliability of multiple multi-source heterogeneous sensing devices under the current driving conditions, and is obtained by mapping the current driving conditions of the vehicle with the output stability of each sensing device. Rear vehicle information represents the information related to the nearest rear vehicle obtained after confidence-weighted fusion, and is used for subsequent risk calculation. Aligned traffic flow features refer to the output results of each sensing device after timestamp synchronization, spatial coordinate transformation, and time series alignment, ensuring that the target positions, speeds, and relative motion states from different sources are in the same reference frame.

[0107] In this embodiment, the vehicle's current driving condition can be characterized by vehicle speed, acceleration, yaw rate, steering angle, road curvature, lane change status, environmental conditions such as rain and fog, and the density of traffic flow in front and behind. The controller evaluates the reliability of each multi-source heterogeneous sensing device based on these parameters. For devices such as millimeter-wave radar, cameras, lidar, and ultrasonic sensors, the system extracts target detection rate, occlusion degree, echo stability, image clarity, and point cloud density, respectively, and maps them to corresponding real-time confidence weights. In practical applications, other models of the device can also be selected, and this application does not limit this.

[0108] Linear weighted fusion, based on the same time reference, multiplies the rear target feature vectors output by each sensing device by their corresponding weights and then sums them to form a rear vehicle information vector. This vector may include fields such as the distance to the nearest rear vehicle, relative speed, lane affiliation, and motion trend, and is output to the fault risk calculation module. During operation, when the reliability of a device decreases due to curves, rain, fog, or obstructed conditions, its corresponding weight is automatically reduced, and the fusion result retains more observation information from stable devices.

[0109] Through the above methods, the system can dynamically adjust the contribution of multi-source perception based on the current driving conditions and perform weighted fusion of the aligned traffic flow features, thereby making the information on vehicles behind more consistent with the real-time traffic situation. The resulting information on vehicles behind can be directly used to calculate the rear-end collision risk value and provide a more reliable input for subsequent selection of pulling over or alternative strategies.

[0110] In one possible embodiment, Figure 5 A flowchart illustrating the fault decision-making method based on rear vehicle information provided in this application embodiment. Figure 4 ,like Figure 5 As shown, the system obtains the vehicle's fault diagnosis results and information about vehicles behind it, including:

[0111] S501 periodically acquires heartbeat data from multiple electronic control units of the vehicle, as well as fault codes output by the unified diagnostic service.

[0112] S502 performs fault determination on multiple electronic control units based on heartbeat data and determines fault data based on fault codes;

[0113] S503 generates fault diagnosis results based on fault determination results and fault data.

[0114] The electronic control unit (ECU) is responsible for the control and diagnostic functions of various vehicle subsystems. The "heartbeat data" consists of periodically reported status information from each ECU, reflecting its online, communication, or operational status. The unified diagnostic service aggregates and outputs internal vehicle diagnostic information. Its output fault codes characterize the types of anomalies in the vehicle or ECU. Fault data consists of identifiable diagnostic content generated after parsing the fault codes. The fault diagnosis result is a unified diagnostic conclusion derived from summarizing the judgments of multiple ECUs and the fault data.

[0115] In its implementation, the vehicle controller reads heartbeat data from multiple electronic control units (ECUs) according to a preset sampling period and writes the most recent reporting time, consecutive reporting count, and status identifier of each ECU into the diagnostic cache. When an ECU experiences a heartbeat interruption, reporting timeout, or abnormal status value change, the controller determines that a fault exists and records the fault determination result. Simultaneously, the unified diagnostic service outputs the corresponding fault code via the on-board diagnostic bus. The controller parses the fault code, establishes a mapping relationship between the fault code and fault data, and converts the fault code into fault data corresponding to the abnormal component, abnormal type, and severity. Subsequently, the controller correlates and summarizes the fault determination results from multiple ECUs with the parsed fault data to generate a fault diagnosis result characterizing the vehicle's current overall fault state. This result is then output to the subsequent fault decision module, providing input for risk assessment involving subsequent vehicle information.

[0116] This diagnostic process continuously assesses the online and operational status of the electronic control unit using heartbeat data, and combines this with fault codes output by a unified diagnostic service to form a more complete fault characterization. This allows vehicle internal anomalies to be summarized into fault diagnosis results in a standardized format. When this result, along with information from other vehicles, is used as input for subsequent decision-making, fault responses can be based on a more accurate understanding of the vehicle's own condition, thus providing a consistent diagnostic basis for selecting pullover or alternative strategies.

[0117] In one possible implementation, Figure 6 A flowchart illustrating the fault decision-making method based on rear vehicle information provided in this application embodiment. Figure 5 ,like Figure 6 As shown, the rear-end collision risk value for a vehicle to immediately pull over is calculated based on information about vehicles behind it, including:

[0118] S601, invoke the preset rear-end collision risk quantification model;

[0119] S602 uses a rear-end collision risk quantification model to perform mathematical mapping calculations on distance and relative speed to generate rear-end collision risk values.

[0120] The rear-end collision risk quantification model is a mathematical model used to calculate the rear-end collision risk value of a vehicle immediately pulling over to the side of the road based on information about vehicles behind it. It uses distance and relative speed as input variables, mapping the longitudinal distance and speed difference between the nearest rear vehicle and the vehicle itself to comparable risk values. Distance characterizes the longitudinal distance between the nearest rear vehicle and the vehicle itself, while relative speed characterizes the speed difference between the nearest rear vehicle and the vehicle itself; both together constitute the basic input of the rear-end collision risk quantification model.

[0121] In practical implementation, after receiving information about vehicles behind, the vehicle controller extracts distance and relative speed parameters from the information and inputs them into a preset rear-end collision risk quantification model. This preset model can be pre-configured during vehicle design, testing, or calibration. Its form can be a linear weighted function, a nonlinear function, or a piecewise mapping function, as long as it reflects the relationship between decreasing distance and increasing relative speed leading to an increased rear-end collision risk. After calling the model, the controller performs mathematical mapping operations on the distance and relative speed according to the weighting coefficients, threshold ranges, or functional expressions set in the model, outputting a rear-end collision risk value that characterizes the safety of the vehicle immediately pulling over to the side of the road.

[0122] In one possible implementation, the model can combine the normalized distance value with the normalized relative speed value to calculate a risk value, and generate a risk value through a preset mapping relationship. When the distance to the following vehicle is small and the relative speed is high, the risk value is calculated as a larger value; when the distance to the following vehicle is large and the relative speed is low, the risk value is calculated as a smaller value. In practical applications, this rear-end collision risk quantification model can also choose other mapping forms, which are not limited in this application.

[0123] In this way, the system can transform the information of vehicles behind into a unified risk value, so that the feasibility of immediately pulling over can be directly represented by the quantitative result. In the subsequent comparison with the threshold, it provides a basis for switching between parking actions and alternative strategies, thereby making the decision input in the fault state clearer and the response result easier for the controller to process.

[0124] In one possible embodiment, Figure 7 A flowchart illustrating the fault decision-making method based on rear vehicle information provided in this application embodiment. Figure 6 ,like Figure 7 As shown, after calculating the rear-end collision risk value of the vehicle immediately pulling over based on the information of vehicles behind, the method also includes:

[0125] S701 reads the vehicle's tire pressure and longitudinal acceleration in real time and calculates the current road friction coefficient;

[0126] S702, calculates the actual achievable deceleration boundary based on the real-time load of the vehicle and the health status of the braking system;

[0127] S703 adaptively corrects the rear-end collision risk value based on the current road friction coefficient and deceleration boundary.

[0128] Tire pressure is a vehicle state parameter used in the calculation of the current road friction coefficient. It is read from the output values ​​of the tire pressure sensors on each vehicle. These sensors can be integrated into the tire valve or the inner side of the wheel hub and upload data at millisecond intervals. Longitudinal acceleration is a vehicle dynamic parameter used in the calculation of the current road friction coefficient. It is typically output by an inertial measurement unit (IMU) and is installed along the vehicle's longitudinal axis. IMUs can work with a filtering unit to suppress transient disturbances caused by vehicle pitch. The current road friction coefficient characterizes the vehicle's current road surface adhesion conditions. It is specifically calculated from tire pressure and longitudinal acceleration using a mapping model. This mapping model can be calculated using a calibration table or an estimation based on regression coefficients. In practical applications, other models of this component can also be selected, and this application does not limit the choice.

[0129] The real-time vehicle load can be jointly determined by suspension height sensors, axle load sensors, or the vehicle attitude estimation module, and together with the vehicle's curb weight, passenger weight, and cargo weight, forms the current load input. The braking system health status is comprehensively determined by brake pressure response, brake pad wear, brake hydraulic circuit integrity, and actuator fault codes. The actual achievable deceleration boundary is obtained by constraining the above load and health status. Its value corresponds to the maximum deceleration limit that the vehicle can stably output under the current operating conditions, usually expressed in m / s², and is used as the braking capacity boundary in risk correction. After the current road friction coefficient and deceleration boundary are input into the correction module, the correction module dynamically weights or recalibrates the original rear-end collision risk value, so that the correction result can reflect the coupling changes of the current road surface adhesion and braking capacity.

[0130] During operation, the system first obtains an initial rear-end collision risk value based on the distance and relative speed of vehicles behind. Then, it combines tire pressure, longitudinal acceleration, real-time vehicle load, and the health status of the braking system to form adhesion conditions and braking boundary parameters. Subsequently, the initial risk value is adaptively corrected, and a corrected rear-end collision risk value is output. This corrected result is used for subsequent pullover decisions, ensuring that the risk assessment is consistent with current road conditions and the vehicle's braking capability, thereby improving the accuracy and relevance of risk judgment under fault conditions.

[0131] In one possible implementation, a rear-end collision risk value is generated by mathematically mapping distance and relative speed using a rear-end collision risk quantification model, including: using the formula R=α*v 2 / d calculates the rear-end collision risk value R, where α is the calibration coefficient obtained by fitting real vehicle collision test data, v is the relative speed, and d is the distance.

[0132] The rear-end collision risk value characterizes the degree of rear-end collision risk posed by a vehicle approaching from behind when the vehicle immediately pulls over to the side of the road. The calibration coefficient is used for engineering calibration of the risk calculation results, and its value is obtained by regression fitting of real vehicle test data under different vehicle speeds, different distances, and different collision conditions. Relative speed corresponds to the speed difference between the nearest vehicle behind and the vehicle itself, and distance corresponds to the longitudinal distance between the nearest vehicle behind and the vehicle itself; both are directly extracted from the rear vehicle information.

[0133] In its implementation, after receiving information about vehicles behind, the controller unifies the units and aligns the relative speed *v* and distance *d* with the time frame, and then substitutes them into a pre-defined rear-end collision risk quantification model to complete the calculation. This model uses R=α*v 2The formula is in the form of / d, where relative speed is included in the calculation as a squared term to reflect the amplification effect of the approaching vehicle on the risk value, and distance is used as the denominator to reflect the mitigating effect of changes in vehicle distance on the risk value. The calibration coefficient α is determined by fitting real vehicle crash test data before model deployment and stored in the parameter table of the on-board control unit, and is directly called during actual operation.

[0134] During data processing, information about vehicles behind can be output by the onboard perception fusion module. The control unit filters this information and selects the closest target vehicle with the most relevant motion state as the calculation object, and inputs the corresponding v and d into the risk model. The calculated rear-end collision risk value is output to the fault decision module in continuous numerical form as one of the criteria for determining whether to execute a pull-over maneuver. If different vehicle models, different sensor installation locations, or different road adhesion conditions require re-adaptation of model parameters, α can be refitted and corrected using real vehicle collision test data. In practical applications, other calibration results can also be selected for this parameter, which is not limited in this application.

[0135] In this way, distance and relative speed are uniformly mapped to a single rear-end collision risk value. The risk quantification result can directly participate in the decision-making in the fault scenario, so that the approaching situation of the following vehicle enters the control link in a quantitative way. This makes the basis for switching between parking on the side of the road and alternative strategies clearer, and after the model parameters are calibrated by real vehicle testing, the calculation results have a higher consistency with the actual risk level.

[0136] Optionally, Figure 8 A flowchart illustrating the fault decision-making method based on rear vehicle information provided in this application embodiment. Figure 7 ,like Figure 8 As shown, after calculating the rear-end collision risk value of the vehicle immediately pulling over based on the information of vehicles behind, the method also includes:

[0137] S801 extracts lane environment features based on real-time lane geometry width and adjacent lane traffic density;

[0138] S802, based on the characteristics of the lane environment, flexibly adjusts the basic safety threshold to obtain the preset threshold.

[0139] Lane environment features characterize the lateral available space of the current lane and the traffic occupancy status of adjacent lanes. Real-time lane geometric width refers to the effective drivable width of the lane the vehicle is currently in at the time of perception, while traffic density refers to the degree of vehicle concentration per unit length or unit area of ​​adjacent lanes. A basic safety threshold serves as a comparison boundary for rear-end collision risk values. Its value can be determined by pre-calibrated risk tolerance parameters and adjusted based on the current road environment to form a preset threshold for decision-making.

[0140] In practical implementation, lane environment features are jointly calculated by the onboard perception unit based on lane line recognition results, road boundary information, and target vehicle lateral positioning information. Lane geometric width can be converted to metric width by combining the pixel width between the left and right lane lines with a calibration scale. Adjacent lane traffic density can be calculated based on the number of target vehicles detected in adjacent lanes, the distance between target vehicles, and the target vehicle coverage rate. This feature vector may include width and density components and can be further mapped to a continuous environment coefficient to characterize whether the current road space is suitable for immediate pull-over. In practical applications, this lane environment feature can also be replaced by other available road geometric information or traffic statistical parameters; this application does not limit this.

[0141] A basic safety threshold can be set as the initial boundary for judging rear-end collision risk values ​​and stored in the vehicle controller. After acquiring lane environment characteristics, the vehicle controller flexibly fine-tunes the basic safety threshold based on the relationship that a smaller lane width and higher adjacent lane density result in a lower safety margin, forming a preset threshold that matches the current road conditions. This flexible fine-tuning can be achieved through threshold increment / decrement calculations, where the increment / decrement is obtained by multiplying the environmental coefficient by the basic safety threshold. This increment / decrement can be limited to preset upper and lower limits to ensure that the threshold adjustment range is controlled. In practical applications, this fine-tuning model can also be implemented using lookup tables, piecewise functions, or linear mapping methods; this application does not limit the specific implementation.

[0142] This allows the comparison boundary for rear-end collision risk values ​​to change synchronously with road space conditions and the degree of surrounding traffic congestion, thus ensuring that the criteria for determining whether to pull over are consistent with the current road environment. Consequently, when lanes are narrow or adjacent lanes are congested, the preset threshold can be adjusted to a level that better meets safety requirements, thereby making the decision-making results under fault conditions more aligned with the real-time traffic situation.

[0143] Based on the aforementioned embodiments, further, obtaining vehicle fault diagnosis results and rear vehicle information includes: predicting rear vehicle information through a side-rear sensor when the fault diagnosis result indicates a rearward sensor malfunction, wherein the rearward sensor is used to directly obtain rear vehicle information.

[0144] The rear-view sensor is a sensing unit deployed in the central area of ​​the rear of the vehicle. Its output represents the distance, orientation, and relative speed of targets directly behind the vehicle. When functioning correctly, the system directly reads its output to generate rear vehicle information. The side-rear sensors are sensing units deployed on the rear fenders, rearview mirror bases, or the rear sides of the vehicle. Their field of view covers the adjacent rear area and the rear blind spot. When the rear-view sensor fails, they are used to indirectly estimate rear targets. Rear vehicle information represents the distance and relative speed between the nearest vehicles behind, serving as input data for subsequent risk assessment.

[0145] In its implementation, the fault diagnosis module first determines whether the rear sensor is faulty based on the sensor's self-test status, communication status, and output data consistency. When the determination indicates a rear sensor fault, the controller switches to the backup sensing channel, fuses the target angle, lateral displacement changes, and temporal trajectory collected by the side and rear sensors, and combines this with the vehicle's own speed and yaw state to predict the current position and movement trend of the rear vehicle, thus outputting rear vehicle information. This prediction can be achieved using target trajectory extrapolation or multi-frame correlation estimation, and the prediction results are converted into distance and relative speed parameters to maintain the continuity of rear situational awareness. If the rear sensor recovers, the system resumes using its direct output to update the rear vehicle information. In practical applications, other models of this component can also be selected, and this application does not limit this choice.

[0146] This solution enables the use of the side and rear sensors to predict information about vehicles behind when the rear sensor fails, thus linking the fault diagnosis results with the acquisition of the rear situation. This ensures that information about the nearest vehicle behind can still be obtained when the main perception channel fails, and provides continuous input for subsequent rear-end collision risk calculation and fault decision-making, thereby improving the completeness of perception and the continuity of control decisions under fault conditions.

[0147] Based on the aforementioned embodiments, after predicting rear vehicle information using the side and rear sensors, the method further includes: if the prediction of rear vehicle information using the side and rear sensors fails, setting the rear-end collision risk value to a first threshold, wherein the first threshold is greater than a preset threshold.

[0148] Among them, the side and rear sensors are used to replace the perception of the distance, relative speed and target presence of the nearest vehicle behind when the rear sensors fail or are blocked; the rear-end collision risk value is used to characterize the risk level of the vehicle immediately pulling over to the side of the road; the preset threshold is used as the risk judgment benchmark under the normal prediction path; and the first threshold is used as a conservative alternative setting in the scenario of prediction failure.

[0149] In practical implementation, the side and rear sensors can include one or more of the following: millimeter-wave radar, ultrasonic radar, or side and rear cameras deployed on the rear fenders or sides of the rear bumper. When the controller receives a prediction failure flag, it no longer continues to calculate based on the missing rear vehicle information, but directly sets the rear-end collision risk value to a first threshold pre-written into memory and inputs this value into the subsequent decision-making module. The first threshold can be pre-configured by the calibration program to a fixed value higher than a preset threshold. For example, if the preset threshold is 0.6, the first threshold can be set to a value in the range of 0.8 to 0.95, so that the risk judgment remains more conservative when the fault perception fails. In practical applications, this threshold configuration can also be adjusted according to the vehicle model, sensor deployment method, and control strategy, which is not limited in this application.

[0150] When the side and rear sensors cannot generate effective information about vehicles behind, this processing method replaces the dynamic calculation result with a fixed higher risk value, so that the subsequent control unit can enter the conservative decision branch without waiting for supplementary perception, and complete the suppression or alternative strategy switching of the immediate pull-over action accordingly.

[0151] By adopting this approach, the continuity of risk assessment input can be maintained in scenarios where back-end information prediction fails, enabling the control system to participate in decision-making in a high-risk state when a failure is perceived, thereby improving the consistency and conservatism of fault degradation control.

[0152] Optional, Figure 9 A flowchart illustrating the fault decision-making method based on rear vehicle information provided in this application embodiment. Figure 8 ,like Figure 9 As shown, alternative strategies are determined based on the fault diagnosis results, including:

[0153] S901, based on the fault diagnosis results and the information of vehicles behind, performs feasibility pruning on multiple preset strategies to obtain multiple feasible preset strategies.

[0154] S902, predict the collision risk value of multiple feasible preset strategies to obtain multiple collision risk values ​​that correspond one-to-one;

[0155] S903 determines an alternative strategy from multiple feasible preset strategies based on multiple collision risk values.

[0156] The system employs several mechanisms: Preset strategies, which compare multiple candidate degradation schemes during the alternative strategy selection phase, include several pre-defined action paths. These paths are subsequently pruned for feasibility, predicted for collision risk, and optimized based on fault diagnosis results and information about following vehicles. Feasible preset strategies represent the set of candidate strategies that remain executable after feasibility pruning. These strategies consist of those that satisfy the current fault state and the constraints of the following traffic conditions, and are further used in collision risk value prediction. Collision risk values ​​characterize the collision hazard level of each feasible preset strategy under the current conditions. These values ​​are obtained by predicting each feasible preset strategy and are used to determine alternative strategies from among the multiple feasible preset strategies. Alternative strategies are the final degradation scheme selected and executed from among the multiple feasible preset strategies. These strategies are determined based on multiple collision risk values ​​and serve as alternative control paths for the pull-over action under fault conditions.

[0157] In this scheme, after receiving the vehicle fault diagnosis results, the system uses the fault type, fault level, and the distance and relative speed between the vehicle's current controllable state and the nearest vehicle behind as screening conditions to perform constraint matching on multiple preset strategies one by one. For preset strategies that are incompatible with the current fault state or cannot be safely executed under the current traffic flow situation behind, the system removes them from the candidate set to form multiple feasible preset strategies. Preset strategies may include control paths such as deceleration and lane keeping, restricted lane changing, graded speed reduction to leave the driving lane, and temporarily maintaining the current trajectory. Different preset strategies differ in terms of the braking capacity, lateral controllability, and target lane safety margin required for execution. The system determines whether the vehicle has the corresponding execution conditions based on the fault diagnosis results and, in conjunction with information on vehicles behind, determines whether a rear-end collision is possible.

[0158] After obtaining multiple feasible preset strategies, the system predicts the collision risk value for each strategy. The collision risk value prediction is calculated based on changes in the vehicle's longitudinal speed, lateral trajectory deviation, relative distance evolution to the nearest following vehicle, and relative speed changes during strategy execution. The system outputs risk characterization results corresponding to each feasible preset strategy. The system can use a pre-calibrated risk mapping model to jointly evaluate the expected trajectory when executing a strategy with the following vehicle's motion state, quantifying the collision hazard level of different strategies under the current traffic situation. If a feasible preset strategy, during execution, results in an excessively short approach time for the following vehicle, braking demands exceeding the vehicle's current controllable state, or a high probability of conflict during lateral entry, the corresponding collision risk value will be predicted as high.

[0159] The system then ranks multiple feasible preset strategies according to the magnitude of their collision risk values, and selects the strategy with the lowest collision risk value that meets the vehicle fault constraints as the alternative strategy. This alternative strategy, after being output, can be directly sent to the decision control unit, where the drive-by-wire chassis actuators execute the corresponding longitudinal or lateral control commands, thereby replacing the originally unsuitable pull-to-the-side parking maneuver. By incorporating fault diagnosis results and rear vehicle information into the feasibility pruning and risk prediction link, the system can select a control path more suitable for the current traffic situation from multiple candidate degradation solutions under fault conditions, and reduce the probability of conflict when executing a single static action.

[0160] After adopting the above method, the selection of alternative strategies no longer depends on the mapping from a single fault type to a fixed action. Instead, unexecutable strategies are first eliminated, and then the collision risk of the remaining strategies is evaluated one by one. Finally, a low-risk control path is output, so that the fault response is consistent with the real-time traffic situation behind, thereby improving the adaptability and control stability of strategy switching under fault conditions.

[0161] In one possible implementation, multiple preset strategies are pruned based on the fault diagnosis results and information about vehicles behind, including: when the fault diagnosis results indicate that the vehicle is not allowed to continue driving, the preset strategies of slowing down to the service area and maintaining the current lane to drive to the service area are removed. The preset strategies include: slowing down and stopping in the current lane, slowing down to the service area, and maintaining the current lane to drive to the service area.

[0162] The "preset strategy" refers to a predefined set of actions used for selecting alternative strategies. This set includes at least three actions: decelerating and stopping in the current lane, proceeding slowly to the service area, and maintaining the current lane while moving towards the service area. "Decelerating and stopping in the current lane" refers to a stopping action where the vehicle decelerates and eventually comes to a stop within the current lane. "Proceeding slowly to the service area" refers to a flexible driving strategy where the vehicle maintains a low speed and moves towards the service area. "Maintaining the current lane while moving towards the service area" refers to a flexible driving strategy where the vehicle continues to move towards the service area along the current lane without changing lanes. The fault diagnosis results characterize the vehicle's current controllable state and its permission to continue driving, while the information on vehicles behind characterizes the traffic situation behind.

[0163] In its implementation, after receiving the fault diagnosis result from the fault diagnosis module, the controller first determines whether the vehicle is in a state where continued driving is not permitted. If the result indicates that continued driving is not permitted, the controller removes strategies related to continuing driving, such as slowing down to the service area and maintaining the current lane towards the service area, from the candidate strategy set, retaining only the current lane deceleration and stopping strategy for further selection. Removal can be achieved by setting the corresponding strategy as unselectable in the strategy table or by deleting the corresponding strategy from the strategy queue, thus ensuring that subsequent collision risk prediction and control command generation only executes the remaining strategies. This process is linked to information about vehicles behind, allowing for direct contraction of the strategy space when the vehicle cannot continue driving, reducing the calculation and output of inapplicable actions.

[0164] Through the aforementioned feasibility pruning, when the fault level is high and the vehicle is not allowed to continue driving, the system retains only parking strategies that match the current state. This ensures that the set of alternative strategies remains consistent with the vehicle's controllable boundaries and reduces the reliance on ineffective driving actions in subsequent strategy evaluation phases. Consequently, the post-fault decision output focuses more on executable actions, improves the certainty of alternative strategy selection, and allows subsequent risk comparisons based on information about vehicles behind to be built upon a valid candidate set.

[0165] Optional, Figure 10 A flowchart illustrating the fault decision-making method based on rear vehicle information provided in this application embodiment. Figure 9 ,like Figure 10As shown, multiple preset strategies include: slowing down and stopping in the current lane, gradually moving to the service area, and maintaining the current lane while moving towards the service area; collision risk values ​​are predicted for multiple feasible preset strategies to obtain multiple corresponding collision risk values, including:

[0166] S1001, calculate the minimum probability of being rear-ended based on the information of vehicles behind in the current lane, and calculate the collision risk value of decelerating and stopping in the current lane based on the minimum value. The probability of being rear-ended is related to the deceleration speed.

[0167] S1002, calculate the collision risk value of slowing down to the service area based on the first unit time driving risk, the first estimated driving time and the additional driving risk, wherein the first unit time driving risk is calculated based on the information of the vehicles behind and the slow speed;

[0168] S1003, calculates the collision risk value for maintaining the current lane and heading towards the service area based on the second unit time driving risk, the second estimated driving time, and the additional risk of fault deterioration. The second unit time driving risk is calculated based on the information of vehicles behind and the reduced driving speed.

[0169] The following parameters are used: Rear vehicle information in the current lane is used to characterize the distance, relative speed, and motion trend of target vehicles behind in the current lane, serving as the basic input for calculating the probability of being rear-ended. The probability of being rear-ended quantifies the likelihood of a rear-end collision when the vehicle behind is decelerating and coming to a stop in the current lane; its value varies with the deceleration rate. The first unit-time travel risk characterizes the collision risk per unit time during the slow-moving journey to the service area; the first estimated travel time characterizes the estimated time required to slow down from the current location to the service area; and the additional travel risk supplements the additional risks introduced during slow-moving due to traffic disturbances, road geometry changes, or short-term control deviations. The second unit-time travel risk characterizes the collision risk per unit time while maintaining the current lane towards the service area; the second estimated travel time characterizes the estimated time required to reach the service area using a degraded driving method; and the fault deterioration risk supplements the additional risks introduced when the fault continues to evolve.

[0170] In its implementation, after receiving the fault diagnosis results and information about vehicles behind, the controller first establishes a rear-end collision probability mapping relationship based on the distance, relative speed, and longitudinal state of the nearest vehicle behind in the current lane. It then calculates the minimum probability of being rear-ended by combining this with the currently achievable stopping deceleration. This minimum value is then substituted into the risk function to obtain the collision risk value for decelerating and stopping in the current lane. For the strategy of slowing down to the service area, the controller calculates the unit-time travel risk based on information about vehicles behind and a preset slowing speed. It then estimates the first estimated travel time based on the remaining path length to the service area, the slowing speed, and road constraints. Finally, it synthesizes the unit-time travel risk with the estimated travel time and additional travel risks to obtain the corresponding collision risk value. For the strategy of maintaining the current lane and driving towards the service area, the controller calculates the second unit-time travel risk based on information about vehicles behind and a preset degraded travel speed. It then estimates the second estimated travel time based on the distance traveled continuously from the current lane to the service area, the degraded speed, and the fault's continued evolution conditions. Finally, it synthesizes this second estimated travel time with the additional fault deterioration risk to obtain the corresponding collision risk value.

[0171] During operation, the system maps three feasible preset strategies to independent risk quantification results, and compares the collision risks of different strategies on the same scale, thus providing a unified basis for subsequent alternative strategy selection. The risk of decelerating and stopping in the current lane reflects the rear-end collision sensitivity during deceleration, the risk of slowing down to the service area reflects the cumulative collision risk during continuous driving, and the risk of maintaining the current lane and driving to the service area further adds the additional risk of the fault continuing to worsen.

[0172] With this approach, the selection of alternative strategies is no longer based solely on static motion mapping. Instead, it incorporates the situation of vehicles behind, their speed, estimated duration, and factors that exacerbate the malfunction into the collision risk assessment. This makes the risk values ​​of different preset strategies comparable and makes the decision results more closely reflect the real-time traffic situation and the controllable state of the vehicles.

[0173] Optional, Figure 11 A flowchart illustrating the fault decision-making method based on rear vehicle information provided in this application embodiment. Figure 10 ,like Figure 11 As shown, alternative strategies are determined from multiple feasible preset strategies based on multiple collision risk values, including:

[0174] S1101, In the case where there is a single minimum value among multiple collision risk values, the feasible preset strategy corresponding to the minimum value is determined as the alternative strategy;

[0175] S1102, when there are N minimum values ​​among multiple collision risk values, the target feasible preset strategy with the highest conservatism is determined from the feasible preset strategies corresponding to the N minimum values ​​according to the conservatism ranking of multiple feasible preset strategies, and the target feasible preset strategy is determined as the alternative strategy.

[0176] Among them, multiple collision risk values ​​are used to characterize the collision hazard level corresponding to multiple feasible preset strategies. Feasible preset strategies are candidate disposal schemes that still meet the current vehicle fault state, power state and road constraints after feasibility pruning. Conservatism ranking is used to rank the aggressiveness of the candidate schemes, their continuous driving intention and their degree of intrusion into surrounding traffic. Alternative strategies are the final output and the degraded schemes used to control the vehicle's execution.

[0177] In practical implementation, after filtering multiple feasible preset strategies, the controller calls the collision risk assessment module to quantify each strategy, obtaining a risk value set corresponding to each strategy. When there is only one minimum risk value in the set, the controller directly selects the feasible preset strategy corresponding to that risk value as the alternative strategy and writes it into the decision output cache. When there are multiple identical minimum risk values ​​in the set, the controller further calls the conservatism determination module to sort these parallel candidate strategies. The sorting can be based on features such as whether the strategy includes continuous driving, whether lane changing is required, whether a larger braking deceleration is required, and whether it is closer to a complete stop, generating a conservatism level. The controller selects the strategy with the highest conservatism level from the strategies corresponding to the parallel minimum values ​​as the target feasible preset strategy, determines it as the alternative strategy, and outputs it to the chassis control unit. This process can be completed in real time by the on-board decision controller, which can be composed of an embedded processor and memory. In practical applications, other models of this component can also be selected, and this application does not limit this.

[0178] In this implementation, the strategy with the lowest risk value is prioritized, representing the lowest overall collision probability. If a tie for the lowest risk value occurs, a conservative ranking is used to avoid selecting a more aggressive action under equal risk conditions. This ensures that the final output alternative strategy balances risk minimization and robustness. Since this alternative strategy is determined jointly by risk value comparison and conservative selection, it maintains decision consistency in degraded control after vehicle failure and makes the control output more consistent with the current traffic situation and vehicle controllability.

[0179] In one possible implementation, after flexibly fine-tuning the basic safety threshold based on lane environment characteristics to obtain a preset safety threshold, the method further includes:

[0180] Set a hysteresis threshold band with upper and lower limits. The hysteresis threshold band includes an upper threshold for entry and a lower threshold for fallback. The upper threshold for entry is greater than or equal to a preset safety threshold, and the lower threshold for fallback is less than or equal to the preset safety threshold.

[0181] If the risk of a rear-end collision exceeds the threshold for intervention, the alternative strategy branch will be forcibly switched.

[0182] The pull-to-the-side parking maneuver will only be permitted to resume when the rear-end collision risk value continues to fall below the lower threshold.

[0183] During the decision-making process, maintain the decision state machine, record the changing trends of the preceding strategy branches and the tail-end risk value. When the tail-end risk value fluctuates within the critical range and meets the anti-oscillation judgment condition, maintain the currently executed strategy branch to avoid repeated logic switching.

[0184] In one possible implementation, after determining and implementing alternative decision-making strategies, the method further includes:

[0185] During the implementation of alternative decision-making strategies, the risk value of rear-end collisions is recalculated frequently in real time.

[0186] Monitor the current traffic flow status and vehicle malfunction status to determine whether there is at least one of the following situations: abnormal acceleration and approach of vehicles from behind, substantial deterioration of vehicle malfunction level, or sudden obstacle ahead of the path corresponding to the alternative strategy.

[0187] When at least one of the following conditions occurs, a secondary emergency interception mechanism is triggered, forcibly interrupting the currently executing alternative decision-making strategy and switching to a higher level of conservative control action, thus completing the dynamic degradation of the strategy.

[0188] In one possible embodiment, Figure 12 A flowchart of the autonomous driving fault degradation decision-making method based on rear vehicle risk assessment provided in this application embodiment is shown below. Figure 12 As shown, the solution includes the following process steps:

[0189] Step 1: Obtain fault information and data from vehicles behind. Receive fault diagnosis results from the autonomous driving system, including fault type, fault level, and the current controllable state of the system. Simultaneously, obtain the distance to the nearest vehicle behind (in meters) and the relative speed between the nearest vehicle and the vehicle behind (in meters per second) via rear-facing millimeter-wave radar or camera; a positive value indicates proximity.

[0190] Step 2: Calculate the rear-end collision risk value for immediately pulling over. Define the rear-end collision risk function as R = f(d, v). The physical meaning of this function is that the smaller the distance to the following vehicle and the greater the relative speed, the higher the probability of being rear-ended when the vehicle is decelerating and changing lanes. The specific calculation formula is R = α * v. 2 / d, where α is a calibration coefficient obtained by fitting real vehicle crash test data. The mathematical form of this formula makes the risk value increase with the square of the relative speed and inversely with the distance, which is consistent with the physical laws of rear-end collisions, that is, the effect of speed difference is greater than the effect of distance.

[0191] In its implementation, after receiving information from vehicles behind, the controller unifies the units and aligns the time for the relative speed *v* and distance *d*, then substitutes them into a pre-defined rear-end collision risk quantification model for calculation. This model uses the form R = α·v² / d, where the relative speed is included as a square term to reflect the amplification effect of the approaching vehicle's approach on the risk value, and the distance is used as the denominator to reflect the suppression effect of changes in vehicle distance on the risk value. The calibration coefficient α is determined by fitting real-vehicle collision test data before model deployment and stored in the parameter table of the onboard control unit. To adapt to the sampling dimensions of different sensors, the system performs engineering dimension mapping on *v* and *d* before substituting them into the formula for calculation, ensuring that the output risk value directly falls within the decision comparison range. If different vehicle models, sensor installation locations, or road adhesion conditions require re-adaptation of model parameters, real-vehicle test data can be used to refit and correct α. In practical applications, other calibration results can also be selected for this parameter; this application does not limit this selection.

[0192] Step 3: Determine whether to pull over based on the rear-end collision risk value. If R is less than the threshold T1, it indicates no threat from behind, and the vehicle pulls over. If R is greater than or equal to the threshold T1, it indicates a rear-end collision risk, and the vehicle abandons the immediate pull-over and proceeds to the alternative strategy selection stage. The threshold T1 is obtained through simulation calibration, and the calibration method is as follows:

[0193] This embodiment describes how to calibrate the rear-end collision risk threshold T1 through simulation.

[0194] The calibration aims to determine a value T1 such that when R is greater than or equal to T1, the accident rate of pulling over to the side of the road is significantly higher than that of using an alternative strategy.

[0195] The simulation setup includes: using a co-simulation platform, the scenario being a straight section of a highway, and implementing two strategies after the vehicle malfunctions: pulling over to the side of the road and slowly driving to a service area. Variables include the distance to following vehicles (range 30 to 150 meters) and relative speed (range 0 to 15 meters per second). The evaluation metric is the collision rate.

[0196] The calibration method is as follows: traverse each interval where v squared divided by d, record the collision rate under the two strategies, and select the point where the collision rate of the two strategies is equal as the threshold T1.

[0197] Calibration results: When v² divided by d is less than 2.0, the collision rate for pulling over is 3%, and the collision rate for slow-moving traffic is 8%, with the recommended action being to pull over. When v² divided by d is between 2.0 and 5.0, the collision rate for pulling over is 15%, and the collision rate for slow-moving traffic is 10%, with the recommended action being to slow down. When v² divided by d is greater than 5.0, the collision rate for pulling over is 35%, and the collision rate for slow-moving traffic is 18%, with the recommended action being to slow down.

[0198] Based on the above data, the threshold T1 is set to 2.0. When v² divided by d is less than 2.0, the collision rate for parking at the side of the road is 3%, lower than the 8% for slow traffic. When v² divided by d is greater than or equal to 2.0, the collision rate for slow traffic is lower than that for parking at the side of the road.

[0199] The coefficient α is determined as follows: when v squared divided by d equals 2.0, let R equal T1 equal 2.0, that is, α multiplied by 2.0 equals 2.0, solving for α yields 1.0. However, in actual systems, α needs to be adjusted according to factors such as vehicle braking performance and response time. In Example 1, α is set to 100 to account for the engineering value after unit conversion. The unit of v is meters per second, and the unit of d is meters. In this case, v squared divided by d is usually a small value, and the coefficient needs to be increased to make the risk value fall within a range that is easy to compare.

[0200] Step 4: Choosing an Alternative Strategy. When pulling over is deemed high-risk, choose from the following alternative strategies: **Slowing Down and Stopping in the Current Lane:** Suitable for situations where following vehicles have reaction time. Action description: Activate hazard lights and smoothly decelerate to zero within the current lane. **Slowly Moving to a Service Area:** Suitable for situations where the distance to a service area is less than five kilometers and the malfunction is not fatal. Action description: Reduce speed to a lower value and maintain the current lane while moving towards the service area. **Degrading Lane and Maintaining Control to Continue Driving:** Suitable for situations where the malfunction does not affect basic longitudinal and lateral control. Action description: Maintain lane centering, increase following distance, and warn following vehicles.

[0201] The choice of an alternative strategy is based on a comparison of the expected risks of each strategy. For example, if the service area is two kilometers away, the risk of driving slowly (continuous exposure while driving) is lower than the risk of being rear-ended while stationary in the driving lane, in which case driving slowly is the better option.

[0202] This application addresses two types of problems. The first is when a large truck is rapidly approaching from behind, the system determines that changing lanes or pulling over will result in a rear-end collision and chooses a slow-moving or warning option instead of blindly pulling over. The second is when there is no threat from behind and there is an empty lane on the right, the system normally pulls over, with decision-making logic consistent with that of a human driver.

[0203] The decision-making logic is interpretable. The reasons why the system chooses a certain strategy can be traced, such as the following vehicle being 50 meters away, the relative speed being 1.4 meters per second, and the rear-end collision risk value of 3.92 exceeding the threshold of 2.0, thus abandoning the option of pulling over. This characteristic is beneficial for system validation and regulatory certification.

[0204] This method is superimposed on the existing redundant control architecture as a decision module, without requiring modification to the underlying architecture.

[0205] The implementation process of the above method will be further explained below with reference to several specific embodiments.

[0206] Example 1:

[0207] This embodiment describes a scenario on a highway where a truck is rapidly approaching from behind, and the system abandons the option of pulling over and instead chooses to proceed at a slower speed.

[0208] Scenario parameters include: Vehicle speed 120 km / h, equivalent to 33.3 m / s. Current lane is the leftmost fast lane. Fault type: Temperature sensor malfunction, a non-fatal fault; main control system is available. The nearest vehicle behind is a truck, 50 meters away, traveling at 125 km / h, equivalent to 34.7 m / s. The relative speed of the vehicle behind is a positive 1.4 m / s, indicating it is approaching. The right lane is congested with traffic, with no safe lane-changing space.

[0209] Step 1: Fault Trigger. The fault diagnosis module outputs an abnormal temperature sensor signal, with a fault level of Level 2 out of 5, where Level 5 is the most severe. The system still retains redundant control capabilities.

[0210] Step 2: Obtain vehicle data from behind. The rearward millimeter-wave radar detection results show a target vehicle 50 meters directly behind, with a relative speed of ±1.4 meters per second.

[0211] Step 3: Calculate the rear-end collision risk value. The formula R equals α multiplied by the square of v divided by d. α is set to 100, and this coefficient is obtained by fitting real-vehicle collision data. Substituting the values: R equals 100 multiplied by the square of 1.4 divided by 50, which equals 100 multiplied by 1.96 divided by 50, equaling 3.92. Let the threshold T1 equal 2.0. The calculated R equals 3.92, which is greater than 2.0, thus indicating a high risk.

[0212] Step 4: Triggering the alternative strategy. The system determines that pulling over at this point requires slowing down before changing lanes, a process that takes approximately 8 to 10 seconds. During this time, a truck approaches from behind at a relative speed of 1.4 meters per second. By the time the vehicle begins to change lanes, the distance to the truck behind has shortened to approximately 36 meters, posing an extremely high risk of a rear-end collision. The system selects to reduce the cruise speed to 90 kilometers per hour, activate the hazard lights, maintain the current lane, and drive towards the service area 3 kilometers away.

[0213] Effect: The vehicle gradually slows down, giving the truck behind it more reaction time and braking distance, thus avoiding accidents caused by sudden stops or forced lane changes.

[0214] Example 2:

[0215] This embodiment describes a scenario where there is no threat from behind, and the system normally executes the pull-over maneuver.

[0216] Scenario parameters include: Vehicle speed 80 km / h, equivalent to 22.2 meters per second. Current lane is the middle lane. Fault type: Forward radar failure, requiring a stop. Nearest vehicle behind is a sedan, 120 meters away, speeding 80 km / h. Relative speed of vehicles behind is zero. Right lane is empty.

[0217] Steps 1 to 2: Fault triggering, acquiring rear data. Distance d equals 120 meters, relative velocity v equals zero.

[0218] Step 3: Calculate the rear-end collision risk value. R equals 100 multiplied by 0 squared divided by 120, which equals 0. R equals 0, which is less than the threshold T1 equals 2.0, and is therefore judged as low risk.

[0219] Step 4: Perform a pullover. The system determines there is no dynamic approaching threat from behind and directly generates a pullover control command. Upon receiving the command, the drive-by-wire chassis actuator activates the steering signal, smoothly changes lanes and enters the emergency lane, then controls the braking system to smoothly decelerate the vehicle to a stop.

[0220] Result: When there is no threat from behind, pulling over to the side of the road is safe and effective. This example illustrates that this approach does not prohibit pulling over to the side of the road in all cases, but rather uses this alternative only when there is a high risk.

[0221] Figure 13 This is a schematic diagram of the fault decision-making device based on rear vehicle information provided in an embodiment of this application. The device in this embodiment can be in software and / or hardware form. For example... Figure 13 As shown in the embodiment of this application, the fault decision-making device 1300 based on rear vehicle information includes: an acquisition module 1301, a calculation module 1302, a first execution module 1303, and a second execution module 1304.

[0222] The acquisition module 1301 is used to acquire the vehicle's fault diagnosis results and rear vehicle information when a vehicle fault is detected. The fault diagnosis results include: fault type, fault level, and the vehicle's current controllable status. The rear vehicle information includes: the distance and relative speed between the nearest rear vehicle and the vehicle.

[0223] Calculation module 1302 is used to calculate the rear-end collision risk value of the vehicle if it immediately pulls over to the side of the road based on the information of the vehicles behind.

[0224] The first execution module 1303 is used to execute a pull-to-the-side parking action when the rear-end collision risk value is less than a preset threshold.

[0225] The second execution module 1304 is used to determine an alternative strategy based on the fault diagnosis results and execute the alternative strategy when the rear-end collision risk value is greater than or equal to a preset threshold.

[0226] This device simultaneously acquires the fault diagnosis results and the distance and relative speed of the nearest vehicle behind, and calculates the rear-end collision risk value for immediate pullover. This allows fault response to move beyond static mapping based solely on internal fault states, incorporating the traffic situation behind into the same decision-making process. When the rear-end collision risk value is below a threshold, a pullover is executed, enabling rapid, minimum-risk maneuvering when the risk is manageable. When the risk reaches or exceeds the threshold, an alternative strategy is selected based on the fault type, fault level, and current controllable state, thus avoiding blindly stopping when approaching a vehicle at high speed. Therefore, this device can more rationally branch out responses after a fault occurs, reducing the risk of secondary collisions and improving the fault safety of advanced autonomous vehicles.

[0227] In one possible implementation, the acquisition module 1301 is further configured to:

[0228] Traffic flow features are collected through multi-source heterogeneous sensing devices, and the traffic flow features are time-aligned through a high-precision timestamp synchronization and alignment mechanism.

[0229] The aligned traffic flow features are filtered to remove abnormal transition points and false alarm targets.

[0230] The filtered traffic flow features are then spatially transformed and aligned with the time series.

[0231] In one possible implementation, the acquisition module 1301 is further configured to:

[0232] Calculate the real-time confidence weights of multiple multi-source heterogeneous sensing devices based on the vehicle's current driving conditions;

[0233] The aligned traffic flow features are linearly weighted and fused based on real-time confidence weights to obtain information about vehicles behind.

[0234] In one possible implementation, the acquisition module 1301 is further configured to:

[0235] Periodically acquire heart rate data from multiple electronic control units of the vehicle, as well as fault codes output by the unified diagnostic service;

[0236] Fault diagnosis is performed on multiple electronic control units based on heart rate data, and fault data is determined based on fault codes;

[0237] Fault diagnosis results are generated based on the fault determination results and fault data.

[0238] In one possible implementation, the computing module 1302 is further configured to:

[0239] Invoke the preset rear-end collision risk quantification model;

[0240] The rear-end collision risk quantification model uses mathematical mapping calculations to map distance and relative speed to generate rear-end collision risk values.

[0241] In one possible implementation, the computing module 1302 is further configured to:

[0242] Real-time reading of vehicle tire pressure and longitudinal acceleration; calculation of current road friction coefficient.

[0243] The actual achievable deceleration boundary is calculated based on the real-time load of the vehicle and the health status of the braking system.

[0244] The rear-end collision risk value is adaptively adjusted based on the current road friction coefficient and deceleration boundary.

[0245] In one possible implementation, the computing module 1302 is further configured to:

[0246] Using the formula R=α*v 2 / d calculates the rear-end collision risk value R, where α is the calibration coefficient obtained by fitting real vehicle collision test data, v is the relative speed, and d is the distance.

[0247] In one possible implementation, the computing module 1302 is further configured to:

[0248] Lane environment features are extracted based on real-time lane geometry width and traffic density of adjacent lanes;

[0249] The basic safety threshold is flexibly fine-tuned based on the characteristics of the lane environment to obtain the preset threshold.

[0250] In one possible implementation, the acquisition module 1301 is further configured to:

[0251] If the fault diagnosis result indicates a rearward sensor failure, the information of vehicles behind is predicted by the side and rearward sensors, where the rearward sensor is used to directly acquire the information of vehicles behind.

[0252] In one possible implementation, the acquisition module 1301 is further configured to:

[0253] If the prediction of rear vehicle information by the side and rear sensors fails, the rear-end collision risk value is set to a first threshold, wherein the first threshold is greater than a preset threshold.

[0254] In one possible implementation, the second execution module 1304 is further configured to:

[0255] Based on the fault diagnosis results and the information of vehicles behind, multiple preset strategies are pruned for feasibility, resulting in multiple feasible preset strategies.

[0256] Collision risk values ​​are predicted for multiple feasible preset strategies, resulting in multiple corresponding collision risk values.

[0257] An alternative strategy is determined from multiple feasible preset strategies based on multiple collision risk values.

[0258] In one possible implementation, the second execution module 1304 is further configured to:

[0259] If the fault diagnosis result indicates that the vehicle is not allowed to continue driving, remove the options from the preset strategies of slowing down to the service area and maintaining the current lane to the service area. The preset strategies include: slowing down and stopping in the current lane, slowing down to the service area, and maintaining the current lane to the service area.

[0260] In one possible implementation, the second execution module 1304 is further configured to:

[0261] The minimum probability of being rear-ended is calculated based on the information of vehicles behind in the current lane, and the collision risk value of decelerating and stopping in the current lane is calculated based on the minimum value. The probability of being rear-ended is related to the deceleration speed.

[0262] The collision risk value for slowing down to the service area is calculated based on the first unit time driving risk, the first estimated driving time, and the additional driving risk. The first unit time driving risk is calculated based on the information of vehicles behind and the slowing speed.

[0263] The collision risk value for maintaining the current lane and heading towards the service area is calculated based on the second unit time driving risk, the second estimated driving time, and the additional risk of fault deterioration. The second unit time driving risk is calculated based on the information of vehicles behind and the reduced driving speed.

[0264] The fault decision-making device based on rear vehicle information provided in this embodiment can execute the method provided in the above method embodiment. Its implementation principle and technical effect are similar, and will not be described in detail here.

[0265] Figure 14 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Figure 14 As shown, the electronic device 1400 provided in this embodiment includes at least one processor 1401 and a memory 1402. Optionally, the device 1400 further includes a communication component 1403. The processor 1401, memory 1402, and communication component 1403 are connected via a bus.

[0266] In a specific implementation, at least one processor 1401 executes computer execution instructions stored in memory 1402, causing at least one processor 1401 to perform the above-described method.

[0267] The specific implementation process of processor 1401 can be found in the above method embodiment, and its implementation principle and technical effect are similar, so it will not be repeated here.

[0268] In the above embodiments, it should be understood that the processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), etc. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the method disclosed in this invention can be directly implemented by a hardware processor, or implemented by a combination of hardware and software modules within the processor.

[0269] The memory may include high-speed memory (Random Access Memory, RAM) and may also include non-volatile memory (NVM), such as at least one disk storage device.

[0270] The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. Buses can be categorized as address buses, data buses, control buses, etc. For ease of illustration, the buses shown in the accompanying drawings are not limited to a single bus or a single type of bus.

[0271] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the above-described method.

[0272] This application also provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the above-described method.

[0273] The aforementioned readable storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. The readable storage medium can be any available medium accessible to a general-purpose or special-purpose computer.

[0274] An exemplary readable storage medium is coupled to a processor, enabling the processor to read information from and write information to the readable storage medium. Of course, the readable storage medium can also be a component of the processor. The processor and the readable storage medium can reside in application-specific integrated circuits (ASICs). Alternatively, the processor and the readable storage medium can exist as discrete components in the device.

[0275] The division of units is merely a logical functional division; in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be indirect coupling or communication connection through some interfaces, devices, or units, and may be electrical, mechanical, or other forms.

[0276] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0277] In addition, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0278] If a function is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0279] Those skilled in the art will understand that all or part of the steps of the above-described method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When executed, the program performs the steps of the above-described method embodiments; and the aforementioned storage medium includes various media capable of storing program code, such as ROM, RAM, magnetic disks, or optical disks.

[0280] Finally, it should be noted that other embodiments of the invention will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This invention is intended to cover any variations, uses, or adaptations of the invention that follow the general principles of the invention and include common knowledge or customary techniques in the art not disclosed herein, and is not limited to the precise structures described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of the invention is limited only by the appended claims.

Claims

1. A fault decision-making method based on rear vehicle information, characterized in that, include: When a vehicle malfunction is detected, the vehicle's malfunction diagnosis results and information about vehicles behind it are obtained. The malfunction diagnosis results include: malfunction type, malfunction level, and the vehicle's current controllable status. The information about vehicles behind it includes: the distance and relative speed between the nearest vehicle behind it and the vehicle. Calculate the rear-end collision risk value of the vehicle immediately pulling over based on the information of the vehicles behind; If the rear-end collision risk value is less than a preset threshold, a pull-over parking maneuver will be performed. If the rear-end collision risk value is greater than or equal to the preset threshold, an alternative strategy is determined based on the fault diagnosis result and the alternative strategy is executed.

2. The method according to claim 1, characterized in that, Before obtaining the vehicle's fault diagnosis results and information about vehicles behind, the method further includes: Traffic flow features are collected using multi-source heterogeneous sensing devices, and the traffic flow features are time-aligned using a high-precision timestamp synchronization and alignment mechanism. The aligned traffic flow features are filtered to remove abnormal transition points and false alarm targets. The filtered traffic flow features are then spatially transformed and aligned with the time series.

3. The method according to claim 2, characterized in that, The acquisition of vehicle fault diagnosis results and rear vehicle information includes: Calculate the real-time confidence weights of multiple multi-source heterogeneous sensing devices based on the current driving conditions of the vehicle; The aligned traffic flow features are linearly weighted and fused based on the real-time confidence weights to obtain the information of the vehicles behind.

4. The method according to claim 1, characterized in that, The acquisition of vehicle fault diagnosis results and rear vehicle information includes: Periodically acquire heart rate data from multiple electronic control units of the vehicle, as well as fault codes output by the unified diagnostic service; The multiple electronic control units are used to determine faults based on the heartbeat data, and fault data is determined based on the fault codes. The fault diagnosis result is generated based on the fault determination result and the fault data.

5. The method according to claim 1, characterized in that, The step of calculating the rear-end collision risk value of the vehicle immediately pulling over based on the information of the vehicles behind includes: Invoke the preset rear-end collision risk quantification model; The rear-end collision risk value is generated by performing mathematical mapping calculations on the distance and the relative speed using the rear-end collision risk quantification model.

6. The method according to claim 5, characterized in that, After calculating the rear-end collision risk value of the vehicle immediately pulling over based on the information of the vehicles behind, the method further includes: The tire pressure and longitudinal acceleration of the vehicle are read in real time to calculate the current road friction coefficient; The actual achievable deceleration boundary is calculated based on the real-time load of the vehicle and the health status of the braking system. The rear-end collision risk value is adaptively corrected based on the current road friction coefficient and the deceleration boundary.

7. The method according to claim 5, characterized in that, The step of generating the rear-end collision risk value by performing a mathematical mapping operation on the distance and the relative speed using the rear-end collision risk quantification model includes: Using the formula R=α*v 2 / d calculates the rear-end collision risk value R, where α is a calibration coefficient obtained by fitting real vehicle collision test data, v is the relative speed, and d is the distance.

8. The method according to claim 1, characterized in that, After calculating the rear-end collision risk value of the vehicle immediately pulling over based on the information of the vehicles behind, the method further includes: Lane environment features are extracted based on real-time lane geometry width and traffic density of adjacent lanes; The preset threshold is obtained by flexibly fine-tuning the basic safety threshold based on the lane environment characteristics.

9. The method according to claim 1, characterized in that, Obtain vehicle fault diagnosis results and information about vehicles behind, including: If the fault diagnosis result indicates a rearward sensor malfunction, the rearward vehicle information is predicted using a side-rear sensor, wherein the rearward sensor is used to directly acquire the rearward vehicle information.

10. The method according to claim 9, characterized in that, After predicting the information of the vehicles behind using the side and rear sensors, the method further includes: If the prediction of the rear vehicle information by the side and rear sensors fails, the rear-end collision risk value is set to a first threshold, wherein the first threshold is greater than the preset threshold.

11. The method according to claim 1, characterized in that, Based on the fault diagnosis results, alternative strategies are determined, including: Based on the fault diagnosis results and the information of the vehicles behind, multiple preset strategies are pruned to obtain multiple feasible preset strategies. Collision risk values ​​are predicted for the multiple feasible preset strategies to obtain multiple corresponding collision risk values; The alternative strategy is determined from the plurality of feasible preset strategies based on the plurality of collision risk values.

12. The method according to claim 11, characterized in that, Based on the fault diagnosis results and the information of the vehicles behind, a feasible pruning of multiple preset strategies is performed, including: If the fault diagnosis result indicates that the vehicle is not allowed to continue driving, the following two preset strategies are eliminated: slow down to the service area and keep in the current lane to drive to the service area. The preset strategies include: slowing down and stopping in the current lane, slowing down to the service area, and keeping in the current lane to drive to the service area.

13. The method according to claim 11, characterized in that, The multiple preset strategies include: slowing down and stopping in the current lane, gradually moving to the service area, and maintaining the current lane while continuing towards the service area; collision risk value prediction is performed on the multiple feasible preset strategies to obtain multiple corresponding collision risk values, including: The minimum probability of being rear-ended is calculated based on the information of vehicles behind in the current lane, and the collision risk value of decelerating and stopping in the current lane is calculated based on the minimum value, wherein the probability of being rear-ended is related to the stopping deceleration rate; The collision risk value for slowing down to the service area is calculated based on the first unit time driving risk, the first estimated driving time, and the additional driving risk, wherein the first unit time driving risk is calculated based on the information of the vehicles behind and the slowing speed. The collision risk value for maintaining the current lane and heading towards the service area is calculated based on the second unit time driving risk, the second estimated driving time, and the additional risk of fault deterioration. The second unit time driving risk is calculated based on the information of the vehicles behind and the reduced driving speed.

14. The method according to claim 11, characterized in that, The alternative strategy is determined from the plurality of feasible preset strategies based on the plurality of collision risk values, including: If a single minimum value exists among the multiple collision risk values, the feasible preset strategy corresponding to the minimum value is determined as the alternative strategy; If there are N minimum values ​​among the multiple collision risk values, the target feasible preset strategy with the highest conservatism is determined from the feasible preset strategies corresponding to the N minimum values ​​according to the conservatism ranking of the multiple feasible preset strategies, and the target feasible preset strategy is determined as the alternative strategy.

15. A fault decision-making device based on rear vehicle information, characterized in that, include: The acquisition module is used to acquire the vehicle's fault diagnosis results and rear vehicle information when a vehicle fault is detected. The fault diagnosis results include: fault type, fault level, and the vehicle's current controllable status. The rear vehicle information includes: the distance and relative speed between the nearest rear vehicle and the vehicle. The calculation module is used to calculate the rear-end collision risk value of the vehicle if it immediately pulls over to the side of the road based on the information of the vehicles behind it. The first execution module is used to execute a pull-to-the-side parking action when the rear-end collision risk value is less than a preset threshold. The second execution module is used to determine an alternative strategy based on the fault diagnosis result and execute the alternative strategy when the rear-end collision risk value is greater than or equal to the preset threshold.

16. An electronic device, characterized in that, include: Memory, processor; The memory stores computer-executed instructions; The processor executes computer execution instructions stored in the memory, causing the processor to perform the method as described in any one of claims 1 to 14.

17. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-executable instructions, which, when executed by a processor, are used to implement the method as described in any one of claims 1 to 14.

18. A computer program product, characterized in that, Includes a computer program that, when executed by a processor, implements the method of any one of claims 1 to 14.