Scene adaptive safety evaluation scheduling method based on multi-source road conditions and related equipment

Abstract

The application discloses a scene adaptive safety evaluation scheduling method based on multi-source road conditions and related equipment, and relates to the technical field of automotive functional safety. The method comprises the following steps: acquiring multi-source road condition perception data; performing standardized preprocessing on the multi-source road condition perception data to obtain standardized input data; generating a scene enabling signal according to the comparison result of the road identification longitudinal distance data and the preset distance threshold; performing logical AND operation on the scene enabling signal and the periodic timing pulse generated based on the hardware counter to generate a task trigger signal; in response to the task trigger signal, starting the intersection safety evaluation model or the multi-lane safety evaluation model for safety evaluation operation to obtain a risk level evaluation result; and performing differentiated data collection and reporting according to the risk level evaluation result. The application can reduce algorithm redundancy, improve scene recognition reliability, ensure scheduling timing stability and reduce invalid data reporting.

Description

Technical Field

[0001] This application relates to the field of automotive electronics technology, specifically to a scenario-adaptive safety assessment and scheduling method and related equipment based on multi-source road conditions. Background Technology

[0002] With the rapid development of intelligent driving technology, vehicle collision risk safety assessment has become one of the core technologies for ensuring driving safety. Current mainstream vehicle collision risk safety assessment solutions fall into two categories: one uses a fixed single-scenario assessment algorithm for all operating conditions, which forces the amplification of safety judgment thresholds to be compatible with all road conditions, resulting in a large number of invalid warning reports and wasting vehicle-to-cloud communication bandwidth; the other uses two assessment models for intersections and multi-lane roads, operating in parallel throughout the entire timeframe. Continuous synchronous calculations consume significant amounts of microcontroller unit computing and memory resources, requiring the use of high-specification automotive-grade chips to ensure real-time performance at fixed intervals, significantly increasing hardware bill of materials costs.

[0003] Furthermore, traditional scenario switching logic relies on pre-stored high-precision vehicle map coordinates to mark intersection areas. In scenarios such as outdated maps, temporary road construction and detours, or tunnels without map signals, it cannot accurately distinguish intersections or multi-vehicle road conditions, leading to mismatched assessment algorithms and missed or incorrect collision risk assessments, failing to meet the accuracy indicators for functional safety risk assessment. Simultaneously, the original controller LAN signal formats for vehicle motion parameters, obstacle coordinates, lane line pixels, and road sign perception are chaotic, with null values ​​and out-of-range abnormal data not being uniformly removed. These are directly sent to the assessment module, occupying computation cycles and easily causing algorithm timeouts. Conventional solutions rely on embedded operating system task management for periodic scheduling. Operating system task preemption and clock drift can cause the safety assessment task start time to deviate, making it impossible to strictly lock a fixed execution cycle. Moreover, there is no hardware logic linkage enable mechanism, making it impossible to dynamically put idle assessment modules into hibernation based on road conditions. Summary of the Invention

[0004] This application provides a scenario-adaptive safety assessment and scheduling method and related equipment based on multi-source road conditions, which can reduce computing power redundancy, improve the reliability of scenario recognition, ensure the stability of scheduling timing, and reduce invalid data reporting.

[0005] On the one hand, embodiments of this application provide a scenario-adaptive safety assessment and scheduling method based on multi-source road conditions, the method comprising the following steps: S100. Acquire multi-source traffic condition perception data, wherein the multi-source traffic condition perception data includes longitudinal distance data of road signs; S200. Perform standardized preprocessing on the multi-source road condition perception data to obtain standardized input data; S300. Based on the comparison result between the longitudinal distance data of the road sign and the preset distance threshold, a scene enable signal is generated. The scene enable signal is used to identify intersection scenes or multi-lane scenes. S400: Perform a logical AND operation between the scene enable signal and the periodic timing pulse generated based on the hardware counter to generate a task trigger signal; S500. In response to the task trigger signal, select one of the intersection safety assessment model or the multi-lane safety assessment model to perform a safety assessment calculation on the standardized input data and obtain the risk level assessment result. S600. Based on the risk level assessment results, perform differentiated data collection and reporting.

[0006] For example, in some embodiments, the multi-source road condition perception data further includes vehicle motion parameters and obstacle motion parameters; S200 includes: S210. Perform a unified angle conversion on the heading angle in the vehicle motion parameters, and perform difference correction when the heading angle jumps across the boundary. S220. Traverse the obstacle motion parameters, identify and mark obstacles containing invalid data, assign default values ​​to the obstacles containing invalid data and skip subsequent calculations; S230, pre-initializes the obstacle distance matrix and collision time array, providing a uniformly formatted input data source for the intersection safety assessment model and the multi-lane safety assessment model.

[0007] For example, in some embodiments, the road sign longitudinal distance data includes the longitudinal distance of traffic lights, the longitudinal distance of stop lines, and the longitudinal distance of pedestrian crossings; S300 includes: S310. Compare the longitudinal distance of the traffic lights with a first preset distance threshold; S320. Compare the longitudinal distance of the stop line with the second preset distance threshold; S330. Compare the longitudinal distance of the pedestrian crossing with a third preset distance threshold; S340. When any one of the following conditions is met: the longitudinal distance of the traffic light falls within the first preset distance threshold range, the longitudinal distance of the stop line falls within the second preset distance threshold range, or the longitudinal distance of the pedestrian crossing falls within the third preset distance threshold range, a scene enable signal for identifying the intersection scene is generated. S350. When the longitudinal distance of the traffic lights does not fall within the first preset distance threshold range, the longitudinal distance of the stop line does not fall within the second preset distance threshold range, and the longitudinal distance of the pedestrian crossing does not fall within the third preset distance threshold range, a scene enable signal for identifying a multi-lane scenario is generated.

[0008] Exemplarily, in some embodiments, S400 includes: S410. The periodic timing pulse is generated by driving the counter with a hardware crystal oscillator. S420. The scene enable signal is converted to Boolean and used as the first input of the logic AND gate, and the periodic timing pulse is used as the second input of the logic AND gate. S430. Perform a logical AND operation through the logical AND gate, and output the task trigger signal when the scene enable signal is valid and the periodic timing pulse arrives. S440. Output the task trigger signal to the input port of the safety assessment model for the corresponding scenario to trigger the intersection safety assessment model or the multi-lane safety assessment model to start calculation.

[0009] For example, in some embodiments, in S500, the step of selectively activating the intersection safety assessment model to perform a safety assessment calculation on the standardized input data to obtain a risk level assessment result includes: S510. Based on the comparison between the yaw rate and the preset angular rate threshold, choose either the turning integral displacement formula or the linear uniform acceleration formula to predict the trajectory coordinates of the vehicle and the obstacle in the future preset time period. S520. Calculate the distance between the vehicle and the obstacle at each time step based on the trajectory coordinates, and generate a distance matrix; S530. Perform dual threshold screening on the distance matrix, the first physical collision threshold, and the second danger threshold to determine the effective collision target; S540. Based on the minimum collision time of the effective collision target, the normalized risk coefficient is calculated by substituting it into the piecewise exponential function to obtain the risk level assessment result.

[0010] For example, in some embodiments, in S500, the step of selectively activating a multi-lane safety assessment model to perform a safety assessment calculation on the standardized input data to obtain a risk level assessment result further includes: S550: Based on the lane markings of the vehicle and the total number of lanes on the road, switch the interactive multi-model filtering sub-model and refresh the lane ownership probability, and identify high-probability vehicles in the same lane as the collision calculation targets. S560: Fits lane curves based on discrete lane pixel coordinates, and calculates the actual arc length of the curve through integral operation to replace the straight distance; S570: Compensate for acceleration signals based on historical vehicle speed difference results, and correct collision time by superimposing vehicle half-length and safety distance; S580. Calculate the normalized risk coefficient based on the corrected collision time to obtain the risk level assessment result.

[0011] Exemplarily, in some embodiments, S600 includes: S610. When the risk level assessment result is high-risk, capture and upload all data within the first preset time window. S620. When the risk level assessment result is medium risk, the data of the second preset time window is extracted and uploaded, and the duration of the second preset time window is greater than that of the first preset time window. S630. When the risk level assessment result is low risk, the risk log is stored locally without triggering real-time upload. S640. At the rising edge of the risk level transition, package the rule identifier, timestamp, and risk label and upload them.

[0012] On the other hand, embodiments of this application also provide a scenario-adaptive safety assessment and scheduling device based on multi-source road conditions, the device comprising: The data acquisition module is used to acquire multi-source traffic condition perception data, which includes longitudinal distance data of road signs. The preprocessing module is used to perform standardized preprocessing on the multi-source road condition sensing data to obtain standardized input data; The scene recognition module is used to generate a scene enabling signal based on the comparison result between the longitudinal distance data of the road sign and the preset distance threshold. The scene enabling signal is used to identify intersection scenes or multi-lane scenes. The scheduling module is used to perform a logical AND operation between the scene enable signal and the periodic timing pulse generated based on the hardware counter to generate a task trigger signal; The assessment module is used to respond to the task trigger signal by selectively activating either the intersection safety assessment model or the multi-lane safety assessment model to perform safety assessment calculations on the standardized input data and obtain the risk level assessment result. The reporting module is used to perform differentiated data collection and reporting based on the risk level assessment results.

[0013] In another aspect, embodiments of this application also provide an electronic device, including a processor and a memory, wherein the memory is used to store computer programs or instructions, and the processor executes the computer programs or instructions to implement the aforementioned scenario adaptive safety assessment and scheduling method based on multi-source road conditions.

[0014] In another aspect, embodiments of this application also provide a computer-readable storage medium storing a computer program or instructions, which are executed by a processor to implement the aforementioned scenario-adaptive safety assessment and scheduling method based on multi-source road conditions.

[0015] The embodiments of this application include at least the following beneficial effects: This application's solution generates scene enable signals based on multi-source road sign perception data and combines hardware counters and logical AND operations to achieve selective scheduling of periodic timing pulses. This solves the technical problems of redundant parallel computing power of dual models, reliance on high-precision maps for scene recognition, and unstable scheduling timing in the prior art. It achieves the technical effects of reducing the computing power load of the microcontroller unit, improving the reliability of scene recognition in all road conditions, and ensuring the stability of fixed periodic timing. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 A schematic diagram illustrating an application scenario of a scene-adaptive safety assessment and scheduling method based on multi-source road conditions provided in this application embodiment; Figure 2 A structural block diagram of a scene adaptive safety assessment and scheduling device based on multi-source road conditions provided in an embodiment of this application; Figure 3 This is a structural block diagram of an electronic device provided in an embodiment of this application. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0019] It is understood that the terms “first,” “second,” etc., used in this application may be used to describe various concepts herein, but unless otherwise stated, these concepts are not limited by these terms. These terms are used only to distinguish one concept from another.

[0020] The terms “at least one,” “multiple,” “each,” “any,” etc., used in this application, include “at least one”, “multiple”, “each”, “any”, etc., “at least one” includes one, two, or more, “multiple” includes two or more, “each” refers to each of the corresponding multiples, and “any” refers to any one of the multiples.

[0021] With the rapid development of intelligent driving technology, vehicle collision risk safety assessment has become one of the core technologies for ensuring driving safety. Among related technologies, existing mainstream vehicle collision risk safety assessment solutions fall into two categories: fixed single-scenario assessment algorithms under all operating conditions; and parallel computation of two assessment models for intersections and multi-lane scenarios at all times. However, these technologies have the following inherent drawbacks: redundant computing power due to dual-model parallelism, resulting in high hardware costs; reliance on offline high-precision maps for scene recognition, leading to poor robustness; lack of standardized data preprocessing links, resulting in dirty data interfering with computation timing; and unstable timing of traditional operating system task scheduling architectures, making adaptation to automotive-grade safety difficult.

[0022] In view of this, this application provides a scene adaptive safety assessment and scheduling method and related equipment based on multi-source road conditions. This application generates scene enable signals based on multi-source road sign perception data and combines hardware counters and logical AND operations to achieve one-way scheduling driven by periodic time pulses. This solves the technical problems of redundant computing power of dual-model parallel processing, scene recognition relying on high-precision maps, and unstable scheduling timing in the prior art. It achieves the technical effects of reducing the computing power load of microcontroller units, improving the reliability of all-road-condition scene recognition, and ensuring the stability of fixed-period timing.

[0023] System architecture and scenario description used in the embodiments of this application: The system in this embodiment is integrated and deployed within the vehicle-side safety monitoring component. It leverages multi-source road condition data from vehicle-mounted sensors to achieve an automatic switching safety assessment model for dual scenarios at intersections and multi-lane roads, meeting the requirements of ISO26262 functional safety and ISO21448 expected functional safety control. The overall system is divided into five modules: an input preprocessing module, a multi-source road condition scene recognition module, a counter and logic gate-based scheduler module, a dual-scenario safety assessment calculation module, and a hierarchical data reporting module. The entire chain is based on Simulink graphical modeling, without Stateflow components. The scheduling logic relies on hardware counters and digital logic gates, and the global baseline timing is uniformly constrained to a fixed period by the hardware counter.

[0024] Please refer to Figure 1 , Figure 1 This illustration shows a flowchart of a scene-adaptive safety assessment and scheduling method based on multi-source road conditions provided in an embodiment of this application. Figure 1 As shown, the scene adaptive safety assessment and scheduling method based on multi-source road conditions according to one embodiment of this application includes, but is not limited to, the following steps: Step S100: Obtain multi-source traffic condition perception data, which includes longitudinal distance data of road signs.

[0025] Step S200: Perform standardized preprocessing on the multi-source road condition perception data to obtain standardized input data.

[0026] Step S300: Based on the comparison result between the longitudinal distance data of the road sign and the preset distance threshold, a scene enable signal is generated. The scene enable signal is used to identify intersection scenes or multi-lane scenes.

[0027] Step S400: Perform a logical AND operation between the scene enable signal and the periodic timing pulse generated based on the hardware counter to generate a task trigger signal.

[0028] Step S500: In response to the task trigger signal, select one of the intersection safety assessment model or the multi-lane safety assessment model to perform safety assessment calculations on the standardized input data and obtain the risk level assessment result.

[0029] Step S600: Based on the risk level assessment results, perform differentiated data collection and reporting.

[0030] In this application embodiment, a scenario-adaptive safety assessment and scheduling method based on multi-source road conditions is provided. This method aims to improve the problems of dual-model parallel computing power redundancy, scenario recognition relying on high-precision maps, unstable scheduling timing, and excessive invalid data reporting in the vehicle safety assessment system of related technologies. By constructing an adaptive assessment framework based on multi-source road condition perception data and hardware logic scheduling, the collaborative optimization of dual-scenario selective operation, precise timing control, and hierarchical data reporting is achieved.

[0031] Specifically, this method first acquires multi-source road condition perception data, including vehicle motion parameters, obstacle motion parameters, lane line pixel coordinates, and longitudinal distance data of road signs. The multi-source road condition perception data is then standardized and preprocessed to obtain standardized input data. Based on this, a scene enabling signal is generated according to the comparison result of the longitudinal distance data of road signs with a preset distance threshold to identify intersection scenarios or multi-lane scenarios. Subsequently, the scene enabling signal is logically ANDed with a periodic timing pulse generated based on a hardware counter to generate a task trigger signal. The task trigger signal is only output when the periodic timing pulse arrives and the corresponding scene enabling signal is valid, thus triggering the start-up operation of the safety assessment model for the corresponding scenario. If the training process reaches preset conditions, the current model is determined as the first-round fine-tuned large language model. This method combines scene recognition results with hardware timing pulses to achieve selective scheduling, adaptively balancing computational power allocation and evaluation accuracy. While ensuring the stability of the fixed-period timing, it effectively reduces the computational load on the microcontroller unit, significantly improving the reliability of all-road-condition scene recognition.

[0032] The following describes and explains each step of the scenario adaptive safety assessment and scheduling method based on multi-source road conditions in the embodiments of this application.

[0033] In step S100, multi-source road condition perception data is acquired.

[0034] In this embodiment, acquiring multi-source road condition perception data is the starting point for executing subsequent safety assessment and scheduling tasks. The multi-source road condition perception data includes vehicle motion parameters, obstacle motion parameters, lane line pixel coordinates, and road sign longitudinal distance data. The vehicle motion parameters include vehicle position, speed, acceleration, heading angle, and yaw rate. The obstacle motion parameters include fifty-channel obstacle coordinates, vehicle speed, heading, and historical motion data. The lane line pixel coordinates are discrete pixel coordinates of the lane lines. The road sign longitudinal distance data includes the longitudinal distances of traffic lights, stop lines, and pedestrian crossings. This data originates from multiple raw signals from the vehicle controller's local area network bus, providing a data foundation for subsequent standardized preprocessing, scene recognition, and safety assessment calculations.

[0035] In step S200, the multi-source road condition perception data is preprocessed to obtain standardized input data.

[0036] In this embodiment, after receiving multiple raw signals from the vehicle controller local area network bus, the input preprocessing module performs unified standardization processing internally. This standardization preprocessing includes dimensional conversion, invalid data cleaning, and pre-initialization of the operand array.

[0037] Specifically, the unified dimensional conversion includes converting vehicle speed from kilometers per hour to meters per second, converting heading angle from angle to radians, and automatically performing a correction of the difference between positive and negative pi when the heading angle jumps across the zero-degree or 360-degree boundary, thus eliminating the yaw rate calculation error caused by the angle jump.

[0038] Invalid data cleaning involves traversing all fifty sets of obstacle parameters, identifying obstacles whose coordinates and speeds contain non-numeric invalid data, marking them as invalid obstacles, assigning a default collision time of ten seconds, and skipping subsequent complex calculations to save computing power.

[0039] The pre-initialization of the operation array includes the pre-initialization of the obstacle distance matrix and the collision time array. The default value of the collision time array is set to six seconds to represent no collision risk, providing a unified input data source for the two evaluation models of intersections and multi-lane roads.

[0040] The two-scenario algorithms share a single preprocessing link, reducing the overhead of redundant data processing and avoiding abnormal algorithm operation caused by abnormal dirty data.

[0041] In step S300, a scene enable signal is generated based on the comparison result between the longitudinal distance data of the road sign and the preset distance threshold. The scene enable signal is used to identify intersection scenes or multi-lane scenes.

[0042] In this embodiment, the multi-source road condition scene recognition module has preset calibration thresholds: the effective judgment range for traffic lights is 0 to 30 meters, the effective range for stop lines is 0 to 20 meters, and the effective range for pedestrian crossings is 0 to 20 meters. The sensing distances of the three types of road signs are collected in real time. If the measured distance of any target falls within the corresponding threshold range, an intersection scene enable signal is output; otherwise, a multi-lane scene enable signal is output, and the Boolean signal is sent to the backend Scheduler module.

[0043] Specifically, the longitudinal distance data of road signs includes the longitudinal distance of traffic lights, the longitudinal distance of stop lines, and the longitudinal distance of pedestrian crossings.

[0044] In one embodiment, generating a scene enable signal based on the comparison result of the road sign longitudinal distance data and a preset distance threshold includes: The longitudinal distance of the traffic light is compared with a first preset distance threshold; the longitudinal distance of the stop line is compared with a second preset distance threshold; the longitudinal distance of the pedestrian crossing is compared with a third preset distance threshold. When any one of the following conditions is met: the longitudinal distance of the traffic light falls within the first preset distance threshold range, the longitudinal distance of the stop line falls within the second preset distance threshold range, or the longitudinal distance of the pedestrian crossing falls within the third preset distance threshold range, a scene enable signal is generated to identify the intersection scene. When the longitudinal distance of the traffic lights does not fall within the first preset distance threshold range, the longitudinal distance of the stop line does not fall within the second preset distance threshold range, and the longitudinal distance of the pedestrian crossing does not fall within the third preset distance threshold range, a scene enable signal is generated to identify the multi-lane scenario.

[0045] The system is driven by real-time in-vehicle perception, which does not rely on offline high-precision maps at all, and scene recognition does not fail even in extreme road conditions.

[0046] In step S400, the scene enable signal is logically ANDed with the periodic timing pulse generated based on the hardware counter to generate a task trigger signal.

[0047] In this embodiment, the Scheduler module is the core of the invention's scheduling mechanism, consisting of a hardware counter, a Boolean conversion unit, and a logic AND gate. In one embodiment, performing a logical AND operation between the scene enable signal and a periodic timing pulse generated based on the hardware counter to generate a task trigger signal includes: The periodic timing pulse is generated by driving a counter with a hardware crystal oscillator, and the period of the periodic timing pulse is one hundred milliseconds. The scenario enable signal is converted to Boolean and used as the first input of the AND gate, and the periodic timing pulse is used as the second input of the AND gate. The logical AND gate performs a logical AND operation, and outputs the task trigger signal when the scene enable signal is valid and the periodic timing pulse arrives. The task trigger signal is output to the input port of the safety assessment model for the corresponding scenario to trigger the intersection safety assessment model or the multi-lane safety assessment model to start calculation.

[0048] Specifically, hardware timing generation is achieved by an independent hardware crystal oscillator driving a counter, which outputs a timing pulse signal every 100 milliseconds as a global cycle reference clock. Dual-input AND logic enable uses the scene enable signal output from scene recognition, which is converted to a Boolean value and used as the first input to an AND gate. The 100-millisecond timing pulse output from the counter serves as the second input to the AND gate. Task output is achieved by performing a logical AND operation using the AND gate. A task trigger signal is output only when the 100-millisecond timing pulse arrives and the corresponding scene enable signal is valid. When the scene enable signal identifies an intersection scene, the AND gate outputs a task trigger signal every 100 milliseconds, triggering the intersection safety assessment module to start its operation. The multi-lane module has no enable signal and remains in sleep mode. When the scene enable signal identifies a multi-lane scene, the scene signal is inverted and connected to another set of AND gates, which outputs a task trigger signal every 100 milliseconds to trigger the multi-lane safety assessment module. The intersection module remains in sleep mode.

[0049] The hardware-level logic handles task start and stop, eliminating operating system task preemption interference. It strictly guarantees that the evaluation algorithm will run in a 100-millisecond cycle, and idle modules consume zero computing power.

[0050] In step S500, in response to the task trigger signal, either the intersection safety assessment model or the multi-lane safety assessment model is selected to start to perform a safety assessment calculation on the standardized input data, and obtain the risk level assessment result.

[0051] In this embodiment, the task trigger signal serves as the overall enable signal for the corresponding scenario evaluation module. The module starts operation only after receiving a valid enable pulse, and it is divided into two independent algorithms: intersection and multi-lane.

[0052] In one embodiment, the step of responding to the task trigger signal by selectively activating the intersection safety assessment model to perform a safety assessment calculation on the standardized input data to obtain a risk level assessment result includes: Based on the comparison between the yaw rate and the preset angular rate threshold, either the turning integral displacement formula or the linear uniform acceleration formula is selected to predict the trajectory coordinates of the vehicle and the obstacle in the future preset time period. Based on the trajectory coordinates, the distance between the vehicle and the obstacle is calculated at each time step to generate a distance matrix; The distance matrix is ​​subjected to dual threshold screening using a first physical collision threshold and a second danger threshold to determine valid collision targets. Based on the minimum collision time of the effective collision target, the normalized risk coefficient is calculated by substituting it into the piecewise exponential function to obtain the risk level assessment result.

[0053] Specifically, trajectory prediction uses a yaw rate of one-thousandth of a radian per second as the threshold. If the yaw rate exceeds this threshold, the vehicle is considered to be turning, and the integral displacement formula for turning is used; otherwise, the uniform acceleration formula for straight lines is used. The coordinates of the vehicle and all obstacles are predicted uniformly over the next three seconds, and distances are calculated hourly to generate a distance matrix. A dual-threshold screening system is used, setting a first physical collision threshold of 6.5 meters and a second danger threshold of 15 meters. If the distance is less than 6.5 meters, the collision time is directly locked. Between 6.5 and 15 meters, the trend of distance changes over the next three steps is examined; only if the distance continuously decreases is a valid collision considered. If the distance is greater than 15 meters, the risk target is directly eliminated. The minimum collision time for all obstacles is taken, and the risk value is calculated using a piecewise exponential function. The normalized output is categorized into high-risk, medium-risk, and low-risk levels.

[0054] In another embodiment, the step of responding to the task trigger signal by selectively activating a multi-lane safety assessment model to perform a safety assessment calculation on the standardized input data and obtaining a risk level assessment result further includes: Based on the vehicle lane markings and the total number of lanes on the road, switch the interactive multi-model filtering sub-model and refresh the lane affiliation probability to identify high-probability vehicles in the same lane as collision calculation targets. The lane curve is fitted based on discrete lane pixel coordinates, and the actual arc length of the curve is calculated by integral operation to replace the straight distance; the acceleration signal is compensated based on the historical vehicle speed difference results, and the collision time is corrected by superimposing the vehicle half length and the safety distance. The normalized risk coefficient is calculated based on the corrected collision time to obtain the risk level assessment result.

[0055] Specifically, the interactive multi-model adaptive lane probability relies on the vehicle's lane markings and the total number of lanes on the road to automatically switch between two-lane and three-lane interactive multi-model filtering sub-models. The single-target lane assignment probability is refreshed at 0.1 seconds, ensuring only vehicles with a high probability of entering the collision calculation. The cubic polynomial curve distance correction uses discrete lane pixels to fit a cubic polynomial, calculating the actual curve arc length through Simpson's integral, replacing the traditional Euclidean straight-line distance. Acceleration-free signal compensation uses the difference in historical vehicle speeds from the last three frames to solve for instantaneous acceleration, superimposed with a 2.4-meter vehicle half-length plus a 0.5-second driver reaction safety distance to correct the collision time, ultimately converting into high-risk, medium-risk, and low-risk normalized risk values.

[0056] In step S600, differentiated data collection and reporting are performed based on the risk level assessment results.

[0057] In this embodiment, the graded event triggering and reporting module collects normalized high-risk, medium-risk, and low-risk values ​​in real time, and controls the data collection window of the big data collection component according to the risk level.

[0058] In another embodiment, the step of performing differentiated data collection and reporting based on the risk level assessment result includes: when the risk level assessment result is high-risk, capturing and uploading all data within a fixed first preset time window; when the risk level assessment result is medium-risk, capturing and uploading data within a second preset time window, wherein the duration of the second preset time window is longer than that of the first preset time window; when the risk level assessment result is low-risk, storing risk logs locally without triggering real-time uploads; and at the rising edge of the risk level transition, packaging rule identifiers, timestamps, and risk tags and uploading them.

[0059] Specifically, in high-risk situations, a data collection command is immediately issued, capturing all bus and sensor data points from -15 seconds to +5 seconds and uploading it to the cloud data lake. In medium-risk situations, the data collection time domain is dynamically expanded to -20 seconds to +10 seconds to supplement redundant sensor data. In low-risk situations, risk logs are only marked in the local safety monitoring component's flash memory, without triggering real-time cloud uploads. Only the rising edge of the risk level transition is packaged with rule identifiers, timestamps, and risk tags and uploaded. Historical event samples are periodically fed back from the cloud for subsequent iterative optimization of model thresholds, forming the expected functional safety closed-loop optimization link.

[0060] The data collection time window is dynamically adjusted based on the normalized risk values ​​of high-risk, medium-risk, and low-risk, which reduces redundant reported data. The cloud-based backflow of labeled samples enables closed-loop iterative optimization of model parameters.

[0061] This application provides a scenario-adaptive safety assessment and scheduling method and related equipment based on multi-source road conditions. Through the collaborative work of an input preprocessing module, a multi-source road condition scenario recognition module, a scheduler module, a dual-scenario safety assessment calculation module, and a hierarchical data reporting module, the invention's objective is fully achieved, and the aforementioned beneficial effects are directly generated.

[0062] Specific application examples: This solution is based on Matlab / Simulink modeling. The Scheduler module is entirely built from counters and logic gates. The code is generated into automotive-grade C++ code using Embedded Coder and deployed in the vehicle safety monitoring component.

[0063] Example 1: Right turn at a crossroads; When the vehicle is turning right at a speed of 10 meters per second and the oncoming vehicle is turning left at 10 meters per second, and the distance to the stop line is 12 meters, falling within the preset distance threshold range, the scene recognition outputs a scene enable signal indicating the intersection scene. A hardware counter outputs a timing pulse every 100 milliseconds; when the logic AND gate meets the condition, it outputs a task trigger signal to activate the intersection assessment module. The module uses a trajectory prediction plus dual-threshold collision time algorithm to calculate the effective collision time, outputs a high-risk value, and triggers the full data upload. This embodiment accurately identifies intersection scenes through a multi-source road condition scene recognition module and precisely triggers the intersection safety assessment model through a Scheduler module, avoiding invalid calculations of multi-lane models, reducing computational load, and freeing it from the constraints of high-precision maps, thus improving the reliability of scene recognition.

[0064] Example 2: High-speed multi-lane cut-in lane change scenario; On a straight, high-speed road surface without traffic lights, stop lines, or pedestrian crossing signals, the scene recognition outputs a scene enable signal indicating a multi-lane scenario. A reverse logic AND gate outputs a task trigger signal to activate the multi-lane assessment module. An interactive multi-model algorithm calculates the probability of adjacent vehicles in the same lane, fits the straight lane with a cubic polynomial, and combines historical vehicle speeds to complete the acceleration of other vehicles. After correcting for collision time, a medium-risk assessment is output, and the data acquisition window is expanded to upload data. This embodiment accurately identifies multi-lane scenarios without intersection signs through a multi-source road condition scene recognition module. The interactive multi-model filtering and lane curve fitting in the multi-lane safety assessment model accurately calculate collision risk, avoiding misjudgments caused by excessively large straight-line distances in traditional methods, and improving the assessment accuracy in multi-lane scenarios.

[0065] Example 3: Overtaking maneuver by using the opposite lane on a mountain curve; In curves without intersection signs, scene enable signals indicate multi-lane scenarios, and the scheduling activates the multi-lane model. A cubic polynomial is used to fit the curve's lane curve, and integral calculations are performed to obtain the actual arc length. This avoids overestimating collision time due to excessive straight-line distance, accurately calculating collision risk and outputting low-risk data. The data is only stored locally and not reported in real-time. This embodiment solves the distortion problem of traditional Euclidean straight-line distance in curve scenarios by correcting the curve arc length in the multi-lane safety assessment model. The low-risk local marking strategy of the hierarchical reporting module effectively reduces invalid data uploads, saving vehicle-to-cloud communication bandwidth and cloud storage resources.

[0066] Compared with related technologies, the embodiments provided in this application have the following technical advantages: By adopting an automatic scene discrimination method based on multi-source road sign perception thresholds, the system is freed from the constraints of high-precision maps, improving the reliability of all-road-condition scene recognition and reducing the collision false positive and false negative rates. Through a scheduling architecture based on hardware counters and logical AND gates, the hardware logic is locked at a fixed period, and the timing stability meets functional safety requirements. The dynamic selection of dual models significantly reduces the computing load on the microcontroller unit, and existing low-to-mid-range automotive-grade chips can achieve full-function implementation in dual scenarios. Through standardized preprocessing and hierarchical reporting mechanisms, the two models reuse the same data source, reducing redundant reported data. Cloud-based feedback of labeled samples enables closed-loop iterative optimization of model parameters.

[0067] Reference Figure 2 In this embodiment of the application, a scene adaptive safety assessment and scheduling device based on multi-source road conditions is also provided. This scene adaptive safety assessment and scheduling device based on multi-source road conditions includes: The data acquisition module is used to acquire multi-source traffic condition perception data, which includes longitudinal distance data of road signs. The preprocessing module is used to perform standardized preprocessing on the multi-source road condition sensing data to obtain standardized input data; The scene recognition module is used to generate a scene enabling signal based on the comparison result between the longitudinal distance data of the road sign and the preset distance threshold. The scene enabling signal is used to identify intersection scenes or multi-lane scenes. The scheduling module is used to perform a logical AND operation between the scene enable signal and the periodic timing pulse generated based on the hardware counter to generate a task trigger signal; The assessment module is used to respond to the task trigger signal by selectively activating either the intersection safety assessment model or the multi-lane safety assessment model to perform safety assessment calculations on the standardized input data and obtain the risk level assessment result. The reporting module is used to perform differentiated data collection and reporting based on the risk level assessment results.

[0068] It is understood that the content of the scenario adaptive safety assessment and scheduling method embodiment based on multi-source road conditions in this application is applicable to the scenario adaptive safety assessment and scheduling device embodiment based on multi-source road conditions. The specific functions implemented by the scenario adaptive safety assessment and scheduling device embodiment based on multi-source road conditions are the same as those of the scenario adaptive safety assessment and scheduling method embodiment based on multi-source road conditions, and the beneficial effects achieved are also the same as those achieved by the scenario adaptive safety assessment and scheduling method embodiment based on multi-source road conditions.

[0069] This invention also provides an electronic device, comprising: At least one processor; At least one memory for storing at least one program; When the at least one program is executed by the at least one processor, the at least one processor performs the method described above.

[0070] It is evident that the content of the above method embodiments is applicable to this device embodiment. The specific functions implemented in this device embodiment are the same as those in the above method embodiments, and the beneficial effects achieved are also the same as those achieved in the above method embodiments.

[0071] Please see Figure 3 , Figure 3 The hardware structure of an electronic device according to another embodiment is illustrated. The electronic device includes: The processor can be implemented using a general-purpose CPU (Central Processing Unit), microprocessor, application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to achieve the technical solutions provided in the embodiments of this application. The memory can be implemented in the form of read-only memory (ROM), static storage device, dynamic storage device, or random access memory (RAM). The memory can store the operating system and other applications. When the technical solutions provided in the embodiments of this specification are implemented through software or firmware, the relevant program code is stored in the memory and called by the processor to execute the methods described in the embodiments of this application. Input / output interfaces are used to implement information input and output; The communication interface is used to enable communication and interaction between this device and other devices. Communication can be achieved through wired means (such as USB, Ethernet cable, etc.) or wireless means (such as mobile network, WIFI, Bluetooth, etc.). A bus is used to transfer information between various components of a device, such as processors, memory, input / output interfaces, and communication interfaces. The processor, memory, input / output interfaces, and communication interfaces communicate with each other within the device via a bus.

[0072] This application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described method.

[0073] It is understood that the content of the above method embodiments is applicable to this storage medium embodiment. The specific functions implemented in this storage medium embodiment are the same as those in the above method embodiments, and the beneficial effects achieved are also the same as those achieved in the above method embodiments.

[0074] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the above-described method.

[0075] It is understood that the content of the above method embodiments is applicable to the embodiments of this program product. The specific functions implemented by the embodiments of this program product are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0076] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0077] Those skilled in the art will understand that all or some of the steps in the methods disclosed above, as well as the functional modules / units in the systems and devices, can be implemented as software, firmware, hardware, or suitable combinations thereof.

[0078] The terms "first," "second," "third," "fourth," etc. (if present) in the specification and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0079] It should be understood that in this invention, "at least one (item)" refers to one or more, and "more than one" refers to two or more. "And / or" describes the relationship between related objects, indicating that three relationships can exist. For example, "A and / or B" can represent three cases: only A exists, only B exists, and both A and B exist simultaneously, where A and B can be singular or plural. The character " / " generally indicates that the preceding and following related objects are in an "or" relationship. "At least one (item) of the following" or similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one (item) of a, b, or c can represent: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, and c can be single or multiple.

[0080] In the several embodiments provided by this invention, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0081] 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.

[0082] Furthermore, 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. The integrated unit can be implemented in hardware or as a software functional unit.

[0083] If the integrated unit 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 the present invention, in essence, or the part that contributes to the prior art, or all or 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 multiple 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 described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing programs, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0084] The preferred embodiments of the present invention have been described above with reference to the accompanying drawings, but this does not limit the scope of the claims of the present invention. Any modifications, equivalent substitutions, and improvements made by those skilled in the art without departing from the scope and spirit of the present invention should be within the scope of the claims of the present invention.

Claims

1. A scenario-adaptive safety assessment and scheduling method based on multi-source road conditions, characterized in that, The method includes the following steps: Acquire multi-source traffic condition perception data, including road sign longitudinal distance data; The multi-source road condition perception data is preprocessed to obtain standardized input data; Based on the comparison result between the longitudinal distance data of the road sign and the preset distance threshold, a scene enable signal is generated. The scene enable signal is used to identify intersection scenes or multi-lane scenes. The scenario enable signal is logically ANDed with the periodic timing pulse generated based on the hardware counter to generate a task trigger signal. In response to the task trigger signal, either the intersection safety assessment model or the multi-lane safety assessment model is activated to perform a safety assessment calculation on the standardized input data to obtain a risk level assessment result. Based on the risk level assessment results, differentiated data collection and reporting will be implemented.

2. The method according to claim 1, characterized in that, The multi-source road condition perception data also includes vehicle motion parameters and obstacle motion parameters; the standardization preprocessing of the multi-source road condition perception data to obtain standardized input data includes: The heading angle in the vehicle motion parameters is converted to a unified angle system, and the difference correction is performed when the heading angle jumps across boundaries. Iterate through the obstacle motion parameters, identify and mark obstacles containing invalid data, assign default values ​​to obstacles containing invalid data and skip subsequent calculations; The obstacle distance matrix and collision time array are pre-initialized to provide a uniformly formatted input data source for the intersection safety assessment model and the multi-lane safety assessment model.

3. The method according to claim 1, characterized in that, The longitudinal distance data of road signs includes the longitudinal distance of traffic lights, the longitudinal distance of stop lines, and the longitudinal distance of pedestrian crossings; The step of generating a scene enable signal based on the comparison result between the longitudinal distance data of the road signs and a preset distance threshold includes: The longitudinal distance of the traffic lights is compared with a first preset distance threshold. The longitudinal distance of the stop line is compared with a second preset distance threshold. The longitudinal distance of the pedestrian crossing is compared with a third preset distance threshold. When any one of the following conditions is met: the longitudinal distance of the traffic light falls within the first preset distance threshold range, the longitudinal distance of the stop line falls within the second preset distance threshold range, or the longitudinal distance of the pedestrian crossing falls within the third preset distance threshold range, a scene enable signal is generated to identify the intersection scene. When the longitudinal distance of the traffic lights does not fall within the first preset distance threshold range, the longitudinal distance of the stop line does not fall within the second preset distance threshold range, and the longitudinal distance of the pedestrian crossing does not fall within the third preset distance threshold range, a scene enable signal is generated to identify the multi-lane scenario.

4. The method according to claim 1, characterized in that, The step of performing a logical AND operation between the scene enable signal and a periodic timing pulse generated based on a hardware counter to generate a task trigger signal includes: The periodic timing pulses are generated by driving a counter using a hardware crystal oscillator. The scenario enable signal is converted to Boolean and used as the first input of the AND gate, and the periodic timing pulse is used as the second input of the AND gate. The logical AND gate performs a logical AND operation, and outputs the task trigger signal when the scene enable signal is valid and the periodic timing pulse arrives. The task trigger signal is output to the input port of the safety assessment model for the corresponding scenario to trigger the intersection safety assessment model or the multi-lane safety assessment model to start calculation.

5. The method according to claim 1, characterized in that, In response to the task trigger signal, the intersection safety assessment model is selectively activated to perform a safety assessment calculation on the standardized input data, obtaining a risk level assessment result, including: Based on the comparison between the yaw rate and the preset angular rate threshold, either the turning integral displacement formula or the linear uniform acceleration formula is selected to predict the trajectory coordinates of the vehicle and the obstacle in the future preset time period. Based on the trajectory coordinates, the distance between the vehicle and the obstacle is calculated at each time step to generate a distance matrix; The distance matrix is ​​subjected to dual threshold screening using a first physical collision threshold and a second danger threshold to determine valid collision targets. Based on the minimum collision time of the effective collision target, the normalized risk coefficient is calculated by substituting it into the piecewise exponential function to obtain the risk level assessment result.

6. The method according to claim 1, characterized in that, The step of responding to the task trigger signal by selectively activating a multi-lane safety assessment model to perform a safety assessment calculation on the standardized input data and obtaining a risk level assessment result further includes: Based on the vehicle lane markings and the total number of lanes on the road, switch the interactive multi-model filtering sub-model and refresh the lane affiliation probability to identify high-probability vehicles in the same lane as collision calculation targets. The lane curve is fitted based on discrete lane pixel coordinates, and the actual arc length of the curve is calculated by integral operation to replace the straight distance. The acceleration signal is compensated based on historical vehicle speed difference results, and the collision time is corrected by superimposing vehicle half-length and safety distance. The normalized risk coefficient is calculated based on the corrected collision time to obtain the risk level assessment result.

7. The method according to claim 1, characterized in that, The step of performing differentiated data collection and reporting based on the risk level assessment results includes: When the risk level assessment result is high-risk, the full amount of data in the first preset time window is captured and uploaded. When the risk level assessment result is medium risk, the data of the second preset time window is extracted and uploaded, and the duration of the second preset time window is longer than that of the first preset time window. When the risk level assessment result is low risk, the risk log is stored locally without triggering real-time upload; At the rising edge of the risk level transition, package the rule identifier, timestamp, and risk label and upload them.

8. A scenario-adaptive safety assessment and scheduling device based on multi-source road conditions, characterized in that, include: The data acquisition module is used to acquire multi-source traffic condition perception data, which includes longitudinal distance data of road signs. The preprocessing module is used to perform standardized preprocessing on the multi-source road condition sensing data to obtain standardized input data; The scene recognition module is used to generate a scene enabling signal based on the comparison result between the longitudinal distance data of the road sign and the preset distance threshold. The scene enabling signal is used to identify intersection scenes or multi-lane scenes. The scheduling module is used to perform a logical AND operation between the scene enable signal and the periodic timing pulse generated based on the hardware counter to generate a task trigger signal; The assessment module is used to respond to the task trigger signal by selectively activating either the intersection safety assessment model or the multi-lane safety assessment model to perform safety assessment calculations on the standardized input data and obtain the risk level assessment result. The reporting module is used to perform differentiated data collection and reporting based on the risk level assessment results.

9. An electronic device comprising a processor and a memory, wherein the memory stores computer programs or instructions, characterized in that, When the computer program or instructions are executed by the processor, they implement the scenario-adaptive safety assessment and scheduling method based on multi-source road conditions as described in any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program or instructions stored thereon, characterized in that, When the computer program or instructions are executed by the processor, they implement the scenario-adaptive safety assessment and scheduling method based on multi-source road conditions as described in any one of claims 1 to 7.

Images