Driving ability evaluation data processing method and system, computer readable storage medium

Abstract

The application relates to a driving ability evaluation data processing method and system and a computer readable storage medium, wherein the method comprises the following steps: loading standardized test scene data; synchronously receiving multi-channel driving operation signals and vehicle motion state data, and adding a unified timestamp to each sampling frame to form a time sequence data stream; performing matching operation on the basis of the position, speed and driving state of a host vehicle and a trigger condition in a dynamic event layer, and outputting an event trigger signal when a preset trigger logic is met; creating a corresponding event level data record unit in response to the event trigger signal, writing a state snapshot at the trigger moment, and detecting a first operation response moment of a driver within a preset observation window, and calculating a response time index; and outputting a driving ability evaluation result on the basis of the event level data record unit and the whole process time sequence data stream. The application realizes time alignment of multi-source data, conditional triggering of dangerous events and automatic extraction of a response time correlation parameter in a standardized test scene.

Description

Technical Field

[0001] This invention relates to the field of driving behavior data processing technology, specifically to a driving ability evaluation data processing method, a driving ability evaluation data processing system, and a computer-readable storage medium applied to standardized driving test scenarios. Background Technology

[0002] In driving simulation tests or semi-realistic driving tests, to assess a driver's perception and reaction capabilities when facing sudden risks, it is typically necessary to apply repeatable scenario stimuli to the subject during the test and record the subject's operational responses after the stimuli appear. However, existing technologies generally suffer from the following shortcomings: First, the timing of event triggering in driving simulation scenarios lacks stable control. Different subjects may trigger the same event at different speeds, locations, or driving states, leading to inconsistent testing conditions. Specifically, while existing driving simulation systems can create virtual roads and traffic events, their focus is primarily on simulation presentation or equipment integration, lacking an integrated data processing workflow that automatically detects event triggering conditions, event-level recording, and first-time response. When different subjects pass through the same location at different speeds or in different driving states, the timing of event triggering and the degree of risk exposure vary significantly, making the test results lack cross-comparison.

[0003] Secondly, the lack of high-precision time alignment between different data sources makes it difficult to accurately establish the temporal correlation between the "event occurrence time and the driver's first response time." Training and evaluation schemes based on videos or manual annotations typically rely on post-event analysis. The video frame rate is usually 25 to 30 frames per second, with a temporal resolution of approximately 33 to 40 milliseconds. This makes it impossible to accurately capture the timing of driving operations at the millisecond level, and it is also difficult to ensure consistency among different evaluators.

[0004] Third, most solutions only record the entire test process or rely on video playback afterward, lacking data slicing and automatic extraction mechanisms for individual dangerous events, making it difficult to stably output comparable response time parameters.

[0005] Furthermore, while assessment methods based on eye tracking, EEG, or questionnaires can reflect some cognitive states, they are costly, complex, and unsuitable for directly establishing a complete response chain from dangerous events to vehicle control operations. For example, Chinese Patent Publication No. CN118115034A discloses a method and system for assessing driver situational awareness effectiveness based on driving scenario simulation. This method collects eye movement data using an eye tracker, EEG signals using an EEG analyzer, and SART and SAGAT questionnaire scores using a questionnaire acquisition module, and then comprehensively calculates the situational awareness effectiveness score. This method relies on dedicated external equipment such as eye trackers and EEG analyzers, which are costly and complex to operate, making it unsuitable for large-scale screening or routine assessment scenarios. In addition, the SAGAT assessment method requires randomly pausing the simulator during driving and having the subject answer questions, disrupting the continuity of the driving process and failing to record a complete driving operation chain.

[0006] For example, Chinese patent publication CN118887847A discloses a driving simulation device and its V2X warning function test control system, including a steering driving simulation platform, a force feedback direct drive motor, and a simulation software modeling system. However, the purpose of this system is to verify whether the V2X warning function works properly, rather than to evaluate the human driver's perception, judgment, and operational capabilities when facing danger. It does not establish a quantitative time correlation between the event triggering time and the driver's operation response time, nor does it define quantitative indicators such as danger perception time or danger reaction time.

[0007] For example, Chinese Patent Publication No. CN117948950A discloses a traffic safety training system, method, terminal, and storage medium. This system uses a surveillance camera to record video data of the driver during driving and marks the video data as events when preset conditions are triggered. However, this technology relies primarily on video as the data recording method and on manual review or semi-automatic annotation as the main data analysis method. It cannot establish a precise temporal correlation between event triggering and driver response in real time; moreover, it does not employ a fixed preset scenario script, making it impossible to ensure that different subjects face completely identical testing conditions.

[0008] Therefore, a new data processing method is needed to address the technical challenges of how to achieve high-precision synchronization of multi-source driving data and scenario events in standardized driving test scenarios, how to trigger dangerous events when real threat conditions are met, and how to automatically extract the temporal relationship between event triggering and the first operational response. Summary of the Invention

[0009] The purpose of this invention is to provide a driving ability evaluation data processing method, a driving ability evaluation data processing system, and a computer-readable storage medium applicable to standardized driving test scenarios, solving the problems of unstable scenario event triggering conditions, insufficient time alignment accuracy between event and operation data, and difficulty in automatically extracting event-level response parameters in the prior art.

[0010] The technical solution of this invention is as follows: A method for processing driving ability evaluation data includes the following steps: S1. Load standardized test scenario data, which includes at least static road layer data and dynamic event layer data, and the dynamic event layer data includes multiple event records and triggering conditions corresponding to each event record; S2. Simultaneously receive multi-channel driving operation signals and vehicle motion status data from the driving operation signal acquisition terminal, and add a unified timestamp to each sampling frame to form a time-series data stream; S3. Based on the time-series data stream, extract the position and motion parameters of the main vehicle in real time, and perform matching calculations with the triggering conditions of the event records that have not yet been triggered in the dynamic event layer data. When the matching result meets the preset triggering logic, output the event triggering signal of the corresponding event. S4. In response to the event trigger signal, create a corresponding event-level data recording unit, write a state snapshot at the event trigger time, and detect the driver's first operation response time based on the time-series data stream within a preset observation window after the event trigger, and calculate at least one response time index between the event trigger time and the first operation response time. S5. Output driving capability evaluation results based on the event-level data recording unit and the full-process time-series data stream.

[0011] As a preferred technical solution of the present invention, the static road layer data includes structured data of road topology, lane attributes and speed limit rules; each event record in the dynamic event layer includes at least an event number, event category, three-dimensional coordinate range of the trigger area, set of trigger conditions, trigger condition logic, event participants and their movement trajectories, and termination conditions.

[0012] As a preferred embodiment of the present invention, the triggering conditions include at least two of the following: position conditions, speed conditions, and driving state conditions. The preset triggering logic is a combined conditional logic formed by combining at least two of the position conditions, speed conditions, and driving state conditions using AND or OR logic. Specifically, the position condition is used to determine whether the center point of the main vehicle enters the three-dimensional coordinate range of the trigger area corresponding to the event; the speed condition is used to determine whether the instantaneous speed of the main vehicle is within a preset speed range; and the driving state condition is used to determine whether the current driving state matches the preset state identifier recorded in the event.

[0013] As a preferred technical solution of the present invention, the detection of the driver's first operation response time in step S4 includes: detecting the times when the throttle signal drops below a first preset threshold, the brake signal rises above a second preset threshold, and the steering wheel angle change exceeds a third preset threshold within a preset observation window after the event is triggered; and determining the earliest sampling time that meets the threshold condition as the first operation response time only when the corresponding signal satisfies the threshold condition for N consecutive frames, where N is an integer greater than or equal to 2.

[0014] Furthermore, the response time index includes at least the danger perception time and the danger reaction time; the danger perception time is the difference between the moment when the throttle signal is first detected to drop above a first preset threshold and the moment the event is triggered, and the danger reaction time is the difference between the moment when the brake signal is first detected to rise above a second preset threshold and the moment the event is triggered.

[0015] As a preferred technical solution of the present invention, the state snapshot includes at least the three-dimensional coordinates of the main vehicle at the time of event triggering, instantaneous speed, steering wheel angle, throttle opening, brake opening and current driving state; the event-level data recording unit is also used to record the operation type sequence, operation amplitude peak, collision state and event end marker within the event observation window.

[0016] As a preferred technical solution of the present invention, an event isolation rule is set between adjacent events; when the event-level data recording unit of the first event has not yet reached the end condition, the second event is marked as a delayed trigger event; wherein, the end condition includes at least the main vehicle leaving the event influence area, a collision occurring, or the observation window timeout.

[0017] As a preferred embodiment of the present invention, the multi-channel driving operation signal includes at least a steering wheel angle signal, throttle opening signal, brake opening signal, gear status signal, turn signal status signal, and seat belt status signal; the vehicle motion status data includes at least the vehicle's three-dimensional coordinates, vehicle speed, and collision status; the accuracy of the unified timestamp is no greater than 10 milliseconds, and the sampling frequency is no less than 100Hz.

[0018] As a preferred technical solution of the present invention, the driving ability evaluation result includes event-level response parameters output by the event-level data recording unit and full-process driving behavior parameters output by the full-process time-series data stream; wherein, the event-level response parameters include at least the hazard perception time, hazard reaction time and collision result corresponding to each event.

[0019] Furthermore, it also includes: based on the event-level response parameters and the full-process driving behavior parameters, calculating operational indicators, tactical indicators and strategic indicators respectively through a pre-set rule engine, and generating a comprehensive evaluation result.

[0020] Furthermore, the comprehensive evaluation result is obtained through weighted scoring calculation; the weighted scoring calculation adopts a full score deduction system, assigning a first weight deduction value to indicators of serious risk behaviors, assigning a second weight deduction value lower than the first weight to indicators of general operational errors, and mapping continuous variable indicators to deduction values ​​in segments according to thresholds.

[0021] Furthermore, the operational indicators include the number of times the accelerator was pressed suddenly, the number of times the brake was applied suddenly, the number of times the lane was changed consecutively, the number of times the turn signal was used incorrectly, and the number of times the turn signal was not used; the tactical indicators include the number of times the red light was run, the percentage of speeding, the average hazard perception time, the average hazard reaction time, the number of collisions, and the number of collisions with pedestrians; the strategic indicators include the accuracy of navigation and whether the destination was reached.

[0022] As a preferred technical solution of the present invention, a basic reaction capability test step is included before step S1: receiving operation signal data collected in the scenario of multiple traffic light intersections on a straight road, and calculating basic reaction capability parameters based on the difference between the time of traffic light state change and the time of the driver's first release of the accelerator, the time of the first application of the brake, the time of the first application of the accelerator and the time of the first release of the brake.

[0023] The present invention also provides a data processing system for evaluating driving ability, including a processor, a memory, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to perform the following: The scene data loading module is configured to load standardized test scene data; the standardized test scene data includes at least static road layer data and dynamic event layer data, and the dynamic event layer data includes multiple event records and trigger conditions corresponding to each event record. The data acquisition and synchronization module is configured to synchronously receive multi-channel driving operation signals and vehicle motion status data from the driving operation signal acquisition terminal, and to attach a unified timestamp to each sampling frame to form a time-series data stream. The event trigger determination module is configured to extract the position and motion parameters of the main vehicle in real time based on the time-series data stream, and perform matching calculations with the triggering conditions of the event records that have not yet been triggered in the dynamic event layer data. When the matching result meets the preset triggering logic, the corresponding event trigger signal is output. The event recording and response analysis module is configured to create a corresponding event-level data recording unit in response to the event trigger signal, write a state snapshot at the event trigger time, and detect the driver's first operation response time based on the time-series data stream within a preset observation window after the event trigger, and calculate at least one response time index between the event trigger time and the first operation response time. The evaluation result generation module is configured to output driving capability evaluation results based on the event-level data recording unit and the full-process time-series data stream.

[0024] The present invention also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, causes the processor to perform the driving ability evaluation data processing method as described in the above technical solution.

[0025] According to the above-described solution, the beneficial effects of this invention are as follows: 1. By synchronizing multi-channel driving operation signals and vehicle motion status data through a unified timestamp, a common time reference can be provided for the time-series correlation calculation between events and operations, thereby improving the accuracy of parameter calculation.

[0026] 2. By using a conditional triggering mechanism that combines position, speed, and driving status constraints, similar hazardous events can be activated under more similar risk exposure conditions in different subjects, thereby improving the consistency and cross-sectional comparability of test results.

[0027] 3. By using the event-level data recording unit to slice and record individual hazardous events, and automatically extracting the first operation response time in the observation window after the event is triggered, it can stably obtain event-level parameters such as hazard perception time and hazard reaction time, avoiding reliance on manual review.

[0028] 4. By controlling event isolation and termination conditions, the interference of overlapping adjacent events on parameter extraction can be reduced, and the auditability and verifiability of event-level analysis results can be improved. Attached Figure Description

[0029] Figure 1 This is a flowchart of the method of the present invention.

[0030] Figure 2 This is a flowchart of the event triggering condition matching process in the method of the present invention.

[0031] Figure 3 This is a flowchart illustrating the driver's response time and response time calculation process in the method of this invention.

[0032] Figure 4 This is a schematic diagram of the data structure of the event-level data recording unit in the method of the present invention.

[0033] Figure 5 This is a flowchart of the multi-level index calculation and comprehensive evaluation process in the method of this invention.

[0034] Figure 6 This is a schematic diagram of the system structure of the present invention. Detailed Implementation

[0035] The technical solution of the present invention will now be clearly and completely described with reference to the accompanying drawings and specific embodiments. The following embodiments are only for illustrating the technical solution of the present invention and are not intended to limit it. All other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0036] The driving ability evaluation data processing method provided by this invention is applied to a processor communicatively connected to a driving operation signal acquisition terminal. In a preferred embodiment, the driving operation signal acquisition terminal is a simulated driving cockpit equipped with multiple driving operation components and corresponding sensors. The simulated driving cockpit includes at least driving operation components such as a steering wheel, accelerator pedal, brake pedal, gear shift lever, turn signal control lever, and seat belt. Each operation component is equipped with a corresponding sensor, and the sensor signals are aggregated to an industrial control computer via a control bus communication interface. The industrial control computer is the device containing the processor executing the data processing method of this invention. In other embodiments, the driving operation signal acquisition terminal may also be a testing device capable of outputting corresponding driving operation signals and vehicle motion state data.

[0037] See attached document Figure 6 The system architecture shown illustrates that the data processing method of this invention involves the collaborative operation of the following functional modules: a scene data loading module, a data acquisition and synchronization module, an event trigger determination module, an event recording and response analysis module, and an evaluation result generation module. Each module runs on the processor as a software program and exchanges data through shared memory or inter-process communication mechanisms.

[0038] like Figures 1 to 5 As shown, a driving ability evaluation data processing method of the present invention includes the following steps: Step S1: Load standardized test scenario data The processor loads standardized test scenario data from memory. The test scenario data consists of two logical layers: static road layer data and dynamic event layer data.

[0039] The static road layer data is loaded once before the test begins and remains unchanged during the test. The static road layer data includes structured data of road topology, lane attributes, and speed limit rules. Specifically, the road topology defines the geometry, connectivity, and road type identification of each road segment; lane attributes define the number of lanes, lane width, and lane marking type, including dashed lines, solid lines, and double solid lines; and speed limit rules define the legal speed limits for each road segment. In a preferred embodiment, the road topology of the static road layer covers at least five road types: urban arterial roads, residential roads, school roads, curved and sloping roads, and elevated expressways, to ensure that the test covers the driver's adaptability in different road environments. The test scenario data also includes preset navigation route data, which covers all the above road types from start to finish, and is used for the calculation of subsequent strategic indicators.

[0040] The dynamic event layer data includes multiple event records. Each event record includes at least the following fields: event number, event category, 3D coordinate range of the trigger area, set of trigger conditions, trigger condition logic, participating objects and their initial positions and trajectories, and termination condition. Specifically, the event number is an integer and globally unique, used to identify each event record; the event category is an enumeration type, which can take values ​​such as dangerous scenario, caution scenario, or other scenarios; the 3D coordinate range of the trigger area defines the spatial region where the event is activated; the set of trigger conditions contains one or more trigger conditions; the trigger condition logic defines the combination relationship between multiple trigger conditions, taking values ​​of "AND" or "OR" logic; the participating objects and their initial positions and trajectories define the type, location, and preset movement path of the virtual traffic participants; and the termination condition defines the criteria for determining the termination of the event, including leaving the area, collision, or timeout.

[0041] In a preferred embodiment, the dynamic event layer includes at least nine event types distributed across different road types: lateral vehicle conflict events, non-motorized vehicle sudden departure events, pedestrian crossing events, main road vehicle failure to yield events, sudden lane change events of vehicles queuing ahead, road obstacle events, side target interference events while waiting at a standstill, high-speed continuous lane change events, and navigation to target area parking events. Typical scenarios for lateral vehicle conflict include: a vehicle turning left at a residential intersection while the vehicle is going straight; a non-motorized vehicle suddenly exiting the road is typically a motorcycle suddenly appearing when turning right at a residential T-junction, or a bicycle suddenly appearing when turning right at an urban intersection; a pedestrian suddenly entering the road is typically a child suddenly entering the lane in a school zone; vehicles failing to yield on the main road are typically vehicles in the main lane failing to slow down when merging into the main road from a parking lot; vehicles suddenly changing lanes while queuing at a traffic light are typically a vehicle suddenly changing lanes, exposing a new danger; road obstruction is typically a disabled vehicle stopped in the driving lane; lateral interference while waiting at a stationary position is typically a motorcycle cutting in from the side while waiting at a red light; continuous lane changing on highways is typically a continuous lane change required on an elevated road; and navigation to a target area requires the subject to reach a designated parking space according to navigation instructions.

[0042] Because the test scenario data is a pre-fixed dataset rather than randomly generated by an algorithm, all subjects face the exact same road topology, traffic facility layout, event types, and trigger points. The direct technical benefit of this feature is that it eliminates the interference of differences in the testing environment on the test results, making the test results of different subjects comparable across different groups.

[0043] Step S2: Synchronously receive multi-channel driving operation signals The processor uses a unified clock source as a reference and synchronously receives multi-channel driving operation signals and vehicle motion status data from the driving operation signal acquisition terminal at a preset sampling frequency. It also adds a unified timestamp to each sampling frame to form a time-series operation data stream.

[0044] The multi-channel driving operation signals include at least: steering wheel angle signal, throttle opening signal, brake opening signal, gear position signal, turn signal, and seat belt position signal. Specifically, the steering wheel angle signal is output by an angle encoder or rotary potentiometer, ranging from ±900 degrees; the throttle opening signal is output by a linear displacement sensor or Hall sensor, ranging from 0 to 100%; the brake opening signal is output by a linear displacement sensor or pressure sensor, ranging from 0 to 100%; the gear position signal is output by a gear position switch, providing discrete state values ​​such as P, R, N, D, and S; the turn signal is output by a toggle switch, indicating left turn, right turn, or off; and the seat belt position signal is output by a buckle status sensor, providing a Boolean value indicating whether the seat belt is fastened or not. Vehicle motion state data includes at least the vehicle's three-dimensional coordinates, vehicle speed, and collision status. The vehicle's three-dimensional coordinates are calculated and output in real time by the vehicle dynamics engine; the collision status is output by the collision detection engine, including collision markers and collision object identifiers.

[0045] In a preferred embodiment, the sampling frame frequency is not less than 100Hz, and the accuracy of the unified timestamp is not greater than 10 milliseconds. The data acquisition service reads data from each channel in a polling or interrupt manner. For continuous signals such as steering wheel angle, throttle opening, and brake opening, data is continuously acquired at a frequency of not less than 100Hz. For discrete signals such as gear position, turn signal position, and seat belt position, data is reported and recorded with the same timestamp when the position changes.

[0046] Through the aforementioned multi-channel synchronous acquisition mechanism, the system can accurately restore a complete snapshot of the driver's entire operating channels at any time; providing a data foundation with precise time alignment for subsequent event-level data analysis and multi-level index calculation. Compared with the existing technology that relies on video frame rate (25 to 30 frames per second, time resolution of about 33 to 40 milliseconds) or physiological signals for indirect inference, the data acquisition accuracy and directness are significantly improved.

[0047] like Figure 2 As shown, step S3: Event-triggered matching operation The processor extracts the vehicle's position coordinates and motion parameters in real time based on the time-series data stream, and performs matching calculations with the trigger conditions of each event record in the dynamic event layer. The motion parameters include instantaneous speed and current driving state.

[0048] Triggering conditions include at least two of the following: position conditions, speed conditions, and driving state conditions. The preset triggering logic is a combined conditional logic formed by combining at least two of the above conditions using "AND" or "OR" logic. Specifically, the position condition is used to determine whether the center point of the main vehicle enters the three-dimensional coordinate range of the trigger area corresponding to the event; the speed condition is used to determine whether the instantaneous speed of the main vehicle is within the preset speed range; and the driving state condition is used to determine whether the current driving state matches the preset state identifier recorded in the event. Driving states include going straight, turning left, turning right, changing lanes, waiting at a red light, and merging into the main road.

[0049] Regarding the specific implementation of driving state recognition: the current driving state can be jointly determined by the road segment marker where the vehicle is located, the steering angular velocity, the rate of change of lateral displacement, the distance to the stop line, and the vehicle speed threshold. For example, when the vehicle is at an intersection junction and the steering wheel angle changes continuously beyond the steering threshold, it is determined to be in a left or right turn state; when the vehicle is within the same road segment and the lateral displacement crosses the lane center line, it is determined to be in a lane change state; when the vehicle is within a preset distance range before the stop line and the vehicle speed is less than a preset low-speed threshold, it is determined to be in a waiting state at a red light state.

[0050] The matching operation is performed each time a frame of data is updated. The processor traverses all untriggered event records in the dynamic event layer, performing spatial matching and conditional logic judgment on each event record in sequence. Spatial matching is used to determine whether the vehicle's coordinates are within the trigger zone, and conditional logic judgment is used to determine whether the speed and driving state meet the trigger condition set for the event. When both judgments meet the preset trigger logic for the event record, the event trigger signal is output, and the preset virtual traffic participants and their movement trajectories in the event record are invoked.

[0051] Taking the incident of a child running across a school road as an example: the triggering conditions for this event include two items. The first is a location condition, requiring the vehicle to enter the school road trigger zone; the second is a speed condition, requiring an instantaneous speed greater than or equal to 20 kilometers per hour. The triggering condition logic is "AND," meaning both conditions must be met simultaneously. When the subject's vehicle enters the preset trigger zone of the school road and the vehicle speed meets the requirements, the system triggers the event, generating a virtual child running across the road at a preset position in front of the vehicle. The child's trajectory is a uniform movement from the sidewalk towards the road.

[0052] The direct technical effect of this conditional triggering mechanism is that a dangerous event is only activated when the vehicle is within a specific location range and meets specific speed and driving state requirements. This ensures that the vehicle is in a state of motion where the event poses a real threat when it is triggered, avoiding the damage to the diagnostic value of the test caused by invalid triggers. Simultaneously, this mechanism guarantees that different subjects encounter the same risk stimulus at the same decision-making point, strengthening the consistency and comparability of test results.

[0053] like Figure 3 and Figure 4 As shown, step S4: Create event-level data recording units and extract the first response time. In response to the event trigger signal output in step S3, the processor creates the corresponding event-level data recording unit and performs the following operations: First, a snapshot of the state at the moment the event is triggered is written to the event-level data recording unit. The state snapshot includes at least the three-dimensional coordinates of the main vehicle, instantaneous speed, steering wheel angle, throttle opening, brake opening, and current driving state at the moment the event is triggered.

[0054] Then, within a preset observation window after the event is triggered, the driver's initial operational response time is detected based on the time-series data stream. Specifically, the following three types of signal changes are detected within the preset observation window after the event is triggered: (1) Detect the first moment when the throttle signal drops from its current value to exceed a first preset threshold; (2) Detect the first moment when the braking signal rises from zero to exceed the second preset threshold; (3) The first moment when the change in steering wheel angle exceeds the third preset threshold.

[0055] To avoid misjudging the response time due to sensor jitter or short-term false triggering, this invention employs a continuous frame confirmation mechanism: the moment is recognized as the first operation response time only when the corresponding signal meets the threshold condition for N consecutive frames, where N is an integer greater than or equal to 2. In a preferred embodiment, N is 3, meaning that the corresponding signal must meet the threshold condition for 3 consecutive frames (equivalent to 30 milliseconds at a 100Hz sampling frequency) before a response can be confirmed. The preset observation window can be set to 1 to 5 seconds, preferably 3 seconds.

[0056] In a preferred embodiment, the first preset threshold is a decrease in throttle opening of ≥10%, the second preset threshold is an increase in brake opening of ≥5%, and the third preset threshold is a change in steering wheel angle of ≥15 degrees.

[0057] Based on the above detection results, response time indicators are calculated. Response time indicators include at least hazard perception time and hazard reaction time. Hazard perception time is equal to the difference between the moment the throttle signal is first detected to drop above a first preset threshold and the moment the event is triggered. Physically, it represents the time interval from the occurrence of the hazardous event to the driver's first perceptual reaction of releasing the throttle, reflecting the driver's visual and attentive acquisition speed of hazard information. Hazard reaction time is equal to the difference between the moment the brake signal is first detected to rise above a second preset threshold and the moment the event is triggered. Physically, it represents the time interval from the occurrence of the hazardous event to the driver's first application of the brake pedal, reflecting the comprehensive reaction speed from perception to decision-making to execution.

[0058] The event-level data recording unit also records the sequence of operation types, peak operation amplitude, collision status, and event end marker within the event observation window. The peak operation amplitude refers to the maximum value of each operation channel during the event, such as the maximum brake opening.

[0059] Regarding event-level data slicing: After an event is triggered, the time of the event trigger is used as the starting point for the slice, recording relevant data in the preset backtracking window before the event trigger and the observation window after the event trigger. The event-level data recording unit is sealed when the main vehicle leaves the event's affected area, a collision occurs, or the observation window times out.

[0060] Regarding adjacent event isolation: Event isolation rules are set between adjacent events. If the event-level data recording unit of the previous event has not yet met the termination condition, the subsequent event is marked as a delayed-triggered event and enters the delayed-triggered queue to avoid multiple events overlapping and causing confusion in the attribution of the first response. The termination condition includes at least the following: the main vehicle leaving the event's affected area, a collision occurring, or the observation window timeout.

[0061] To illustrate the content of the event-level data recording unit using a specific event as an example: Event number 003, event name "Child running out of school zone", event category "dangerous scenario", trigger time 12:03:45:320, vehicle speed at trigger time 42.3 km / h, first release of accelerator 12:03:46:080 (danger perception time 0.760 seconds), first application of brakes 12:03:46:540 (danger reaction time 1.220 seconds), first steering 12:03:46:890, peak brake opening 78%, collision result "no collision".

[0062] The system can quantitatively distinguish three fundamentally different driving states, enabling high-resolution diagnosis of driving ability deficiencies. The first is "driver does not perceive danger," characterized by no detected operational response; the second is "driver perceives danger but reacts slowly," characterized by a response time exceeding the normal range; and the third is "driver perceives danger and responds promptly," characterized by a response time within the normal range.

[0063] like Figure 5 As shown, step S5: Output driving ability evaluation results. The processor outputs driving capability evaluation results based on event-level data recording units and full-process time-series data streams. The driving capability evaluation results include event-level response parameters output from the event-level data recording units and full-process driving behavior parameters output from the full-process time-series data streams. Among these, the event-level response parameters include at least the hazard perception time, hazard reaction time, and collision outcome corresponding to each event.

[0064] In a preferred embodiment, the processor also uses a pre-built rule engine to calculate driving capability indicators at three levels: operational indicators, tactical indicators, and strategic indicators, based on event-level response parameters and full-process driving behavior parameters, and generates a comprehensive evaluation result.

[0065] Operational indicators reflect the accuracy and standardization of driver actions, including the following specific indicators: Number of times the accelerator was pressed suddenly, counted by scanning the entire time-series data stream and counting the number of times the accelerator signal change rate exceeded a preset acceleration threshold; Number of times the brake was applied suddenly, counted by counting the number of times the brake signal change rate exceeded a preset braking threshold; Number of times the driver changed lanes consecutively, counted by detecting the number of times the driver crossed two or more lane dividing lines consecutively within a preset distance; Number of times the driver used the turn signal incorrectly, counted by comparing the turn signal direction with the actual turning operation direction in a time-series logic; Number of times the turn signal was not used, counted by detecting whether the turn signal was off during lane changes or turns.

[0066] Tactical indicators reflect a driver's perception and judgment abilities and rule compliance in traffic situations, including the following specific indicators: number of red light violations, calculated by logically determining whether the vehicle's center point has crossed the stop line when the traffic light is red; speeding percentage, calculated by dividing the speeding distance by the total distance of the speed-limited section and multiplying by 100%; average hazard perception time, calculated as the arithmetic mean of the hazard perception times for each hazard event; average hazard reaction time, calculated as the arithmetic mean of the hazard reaction times for each hazard event; number of collisions, calculated as the number of collisions with virtual vehicles; and number of collisions with pedestrians, calculated as the number of collisions with virtual pedestrians or non-motorized vehicle riders.

[0067] Strategic indicators reflect the driver's ability to plan and execute the overall driving task, including the following specific indicators: navigation accuracy, which is calculated by mapping the actual driving trajectory of the main vehicle to the road segment identifier sequence in the road topology in chronological order, and then matching it segment by segment with the road segment sequence corresponding to the preset navigation route; whether the designated destination has been reached, which is determined by judging whether the distance between the coordinates of the main vehicle's final parking position and the coordinates of the preset destination is less than a preset distance threshold.

[0068] Regarding the specific implementation of route matching and navigation accuracy: The actual driving trajectory of the main vehicle can be mapped to a sequence of road segment identifiers in the road topology in chronological order. This sequence is then matched segment by segment with the road segment sequence corresponding to the preset navigation route to calculate the navigation accuracy. Specifically, at each sampling moment, the current road segment identifier is determined based on the three-dimensional coordinates of the main vehicle. The chronologically ordered sequence of road segment identifiers is compared with the road segment sequence of the preset navigation route. The proportion of the distance of the matching road segments to the total navigation route distance is calculated, which is the navigation accuracy.

[0069] The three levels of indicators correspond to different cognitive levels of driving behavior. Operational indicators measure the driver's hand-foot coordination, tactical indicators measure the driver's safety awareness and emergency response capabilities, and strategic indicators measure the driver's route planning and goal execution capabilities.

[0070] In a preferred embodiment, the comprehensive evaluation result is obtained through a weighted scoring calculation. The weighted scoring calculation uses a full-score deduction system, with the initial full score set at 100 points. The scoring rules are divided into three categories based on the severity of the risk of the indicators: Serious risk behaviors are assigned a first-weighted deduction value. Serious risk behaviors include at least collisions with pedestrians, running red lights, and not wearing seatbelts on highways. In a preferred embodiment, 20 points are deducted for each collision with a pedestrian, 10 points for each running a red light, and 10 points for not wearing a seatbelt on a highway. Highways can be defined as sections with a speed limit of 60 kilometers per hour or higher.

[0071] General operational errors are assigned a second-weighted deduction value, lower than the first weight. In a preferred embodiment, failure to use a turn signal incurs a deduction of 2 points, sudden braking or sudden acceleration incurs a deduction of 1 point, and continuous lane changes incur a deduction of 3 points.

[0072] Continuous variable indicators are mapped to deduction values ​​based on threshold segments. Continuous variable indicators include at least the average hazard perception time, the average hazard reaction time, and the percentage of speeding. In a preferred embodiment, the threshold segments for the average hazard perception time are: 0 points deducted for 0.8 seconds or less, 3 points deducted for 0.8 to 1.5 seconds, 6 points deducted for 1.5 to 2.5 seconds, and 10 points deducted for greater than 2.5 seconds; the threshold segments for the average hazard reaction time are: 0 points deducted for 1.2 seconds or less, 3 points deducted for 1.2 to 2.0 seconds, 6 points deducted for 2.0 to 3.0 seconds, and 10 points deducted for greater than 3.0 seconds.

[0073] The overall score equals 100 points minus the sum of all deductions, with a minimum overall score of 0. A competency level determination is output based on a comparison between the overall score and a preset level threshold. In a preferred embodiment, the competency level determination includes three levels: an overall score of 85 to 100 indicates a pass, an overall score of 70 to 84 indicates a recommendation for retesting, and an overall score below 70 indicates high risk.

[0074] Because the scoring rules are pre-configured in the rule engine, the results are completely consistent when scoring the same test data of the same subject at different times and locations. The scoring process is fully automated and standardized, eliminating subjective bias between different raters in human scoring, and the test results are auditable and verifiable.

[0075] In a preferred embodiment, a basic reaction ability test step is included before step S1 to measure the subject's basic reaction ability separately before the formal driving ability test.

[0076] The reaction ability test scenario is a series of traffic light intersections set up on a straight road. The subject drives on the straight road and passes through multiple traffic light intersections sequentially. The system receives operational signal data collected in this scenario and calculates basic reaction ability parameters based on the differences between the times of traffic light changes and the times the driver first releases the accelerator, first applies the brake, first applies the accelerator, and first releases the brake. Specifically, these parameters include: For scenarios where the traffic light changes from green to red: the deceleration accelerator reaction time is equal to the difference between the moment the traffic light changes from green to red and the moment the driver first releases the accelerator; the deceleration braking reaction time is equal to the difference between the moment the traffic light changes from green to red and the moment the driver first presses the brake pedal.

[0077] For scenarios where the traffic light changes from red to green: the accelerator response time is equal to the difference between the moment the traffic light changes from red to green and the moment the driver first presses the accelerator pedal; the braking response time is equal to the difference between the moment the traffic light changes from red to green and the moment the driver first releases the brake pedal.

[0078] Simultaneously, the number of erroneous operations that deviated from the expected operation type was counted, including: incorrect accelerator release, incorrect brake application, incorrect brake release, and incorrect accelerator application. These erroneous operation counts reflect the driver's operational stability in a simple stimulus-response task. The reaction ability test results serve as reference data for the subject's baseline reaction ability and are included in the individual test report.

[0079] In a preferred embodiment, after the evaluation results are output in step S5, individual test report data and group statistical report data are also generated.

[0080] Individual test report data should include at least the following: basic information of the subject, reaction ability test results, detailed records of each event, specific values ​​of each operational indicator, specific values ​​of each tactical indicator, specific values ​​of each strategic indicator, a list of major erroneous behaviors, comprehensive score and ability level determination conclusion, and an index of scene replay segments.

[0081] The group statistical report data should include at least the following: the average danger perception time and average danger reaction time of different subjects under the same event, the error rate and collision rate of each event, the distribution of the overall score, and the proportion of people in each ability level.

[0082] In summary, the method of this invention includes fixed datasets for a static road layer and a dynamic event layer, ensuring that all subjects face identical road topology, traffic facilities, and event triggering conditions, eliminating environmental differences and achieving horizontal comparability of test results. This method uses a unified clock source and high sampling frequency to acquire multi-channel operation signals such as steering wheel, accelerator, and brake, as well as vehicle motion states, forming a time-series operation data stream with millisecond-level precision, providing a precisely aligned data foundation for event-level analysis. Based on the real-time position and motion parameters of the main vehicle, this method combines position, speed, driving state, and other conditional logic for matching, activating dangerous events only under truly threatening motion states, avoiding invalid triggers, and enhancing test consistency and diagnostic value. This method records a state snapshot after event triggering and detects the first operation response time through a continuous frame confirmation mechanism, calculating the danger perception time and danger reaction time, quantitatively distinguishing between three driving states: "no danger perceived," "perceived but slow reaction," and "perceived and responded promptly," achieving high-resolution capability defect diagnosis.

[0083] like Figure 6 As shown, the present invention also provides a data processing system for evaluating driving ability, including a processor, a memory, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it performs the following: The scene data loading module is configured to load standardized test scene data; the standardized test scene data includes at least static road layer data and dynamic event layer data, and the dynamic event layer data includes multiple event records and the trigger conditions corresponding to each event record.

[0084] The data acquisition and synchronization module is configured to synchronously receive multi-channel driving operation signals and vehicle motion status data from the driving operation signal acquisition terminal, and to attach a unified timestamp to each sampling frame to form a time-series data stream.

[0085] The event trigger determination module is configured to extract the position and motion parameters of the main vehicle in real time based on the time-series data stream, and perform matching calculations with the triggering conditions of the event records that have not yet been triggered in the dynamic event layer data. When the matching result meets the preset triggering logic, the corresponding event trigger signal is output.

[0086] The event recording and response analysis module is configured to create a corresponding event-level data recording unit in response to the event trigger signal, write a state snapshot at the event trigger time, and detect the driver's first operation response time based on the time-series data stream within a preset observation window after the event trigger, and calculate at least one response time index between the event trigger time and the first operation response time.

[0087] The evaluation result generation module is configured to output driving capability evaluation results based on the event-level data recording unit and the full-process time-series data stream.

[0088] The processor can be a general-purpose processor, including a central processing unit (CPU) and a network processor, or it can be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices. The memory can include volatile memory, such as random access memory (RAM), or non-volatile memory, such as flash memory, hard disk drive (HDD), or solid-state drive (SSD).

[0089] In a preferred embodiment, the processor is the central processing unit of the industrial control computer, and the memory consists of the industrial control computer's RAM and hard disk. The industrial control computer communicates with the various sensors in the simulated cockpit via a CAN bus or USB-HID protocol interface, receiving multi-channel driving operation signals and vehicle motion status data. The industrial control computer simultaneously runs a vehicle dynamics engine and a visual rendering engine, forming a real-time closed loop from driver operation to screen update, with a total latency of no more than 100 milliseconds.

[0090] The present invention also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, causes the processor to perform the data processing method of any of the above embodiments.

[0091] The computer-readable storage medium can be a tangible, non-transitory storage medium, including but not limited to disks, optical discs, and semiconductor memories. The computer program can be downloaded to the device where the processor resides, installed, and run, or it can be distributed via a network. By storing a computer program that implements the data processing method of this invention, this computer-readable storage medium enables any device equipped with a corresponding processor to load and execute the method, resulting in the same technical effects as described in the method embodiments.

[0092] The above embodiments are merely preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

[0093] Those skilled in the art will understand that the order of steps in the above embodiments does not imply a strict execution order limitation. The data acquisition in step S2 and the event triggering matching operation in step S3 are performed in parallel and in real time during actual execution. The specific implementation of each step can be adjusted according to actual needs, as long as it does not depart from the technical solution defined by the claims of this invention.

[0094] The specific values ​​listed in the above embodiments, such as sampling frequency of 100Hz, timestamp accuracy of 10 milliseconds, throttle drop threshold of 10%, brake rise threshold of 5%, steering wheel change threshold of 15 degrees, full score of 100 points, various deduction values ​​and level thresholds, are all example values ​​in the preferred embodiments. Those skilled in the art can adjust the above values ​​within a reasonable range according to actual testing needs, and such adjustments do not depart from the protection scope of the present invention.

Claims

1. A method for processing driving ability evaluation data applied to standardized driving test scenarios, characterized in that, Includes the following steps: S1. Load pre-fixed standardized test scenario data, which includes at least static road layer data and dynamic event layer data. The dynamic event layer data includes multiple event records and trigger conditions corresponding to each event record. Each event record includes at least an event number, event category, three-dimensional coordinate range of the trigger area, set of trigger conditions, trigger condition logic, and the initial position and preset motion trajectory of the event participants. S2. Using a unified clock source as a reference, multi-channel driving operation signals and vehicle motion status data are synchronously received from the driving operation signal acquisition terminal at a sampling frequency of not less than 100Hz, and a unified timestamp with an accuracy of not more than 10 milliseconds is added to each sampling frame to form a time-series data stream; the multi-channel driving operation signals include at least steering wheel angle signal, throttle opening signal and brake opening signal; S3. Based on the time-series data stream, extract the main vehicle's position, instantaneous speed, and current driving state in real time, and perform matching operations with the triggering conditions of untriggered event records in the dynamic event layer data; the triggering conditions include at least two of the following: position conditions, speed conditions, and driving state conditions. The position condition is used to determine whether the main vehicle's center point enters the three-dimensional coordinate range of the trigger area corresponding to the event. The speed condition is used to determine whether the main vehicle's instantaneous speed is within a preset speed range. The driving state condition is used to determine whether the current driving state matches the preset state identifier of the event record. When the preset triggering logic formed by combining at least two conditions using "AND" logic and "OR" logic is satisfied, output the event trigger signal of the corresponding event, and call the preset event participants and their motion trajectories in the event record. S4. In response to the event trigger signal, create a corresponding event-level data recording unit, write a state snapshot at the event trigger time, the state snapshot including at least the three-dimensional coordinates of the main vehicle, instantaneous speed, steering wheel angle, throttle opening, brake opening, and current driving state at the event trigger time; and within a preset observation window after the event trigger, detect the moment when the throttle signal drops above a first preset threshold and the moment when the brake signal rises above a second preset threshold based on the time-series data stream, and determine the earliest sampling moment that meets the threshold condition as the first operation response moment of the corresponding channel only when the corresponding signal meets the threshold condition for N consecutive frames, where N is an integer greater than or equal to 2; and calculate the danger perception time and danger reaction time, the danger perception time being the difference between the moment when the throttle signal drops above the first preset threshold is first detected and the event trigger time, and the danger reaction time being the difference between the moment when the brake signal rises above the second preset threshold is first detected and the event trigger time; S5. Output driving ability evaluation results based on the event-level data recording unit and the full-process time-series data stream. The driving ability evaluation results include at least the hazard perception time, hazard reaction time and collision result corresponding to each event.

2. The driving ability evaluation data processing method applied to standardized driving test scenarios according to claim 1, characterized in that, The static road layer data includes structured data on road topology, lane attributes, and speed limit rules.

3. The driving ability evaluation data processing method applied to standardized driving test scenarios according to claim 1, characterized in that, The event-level data recording unit is also used to record the sequence of operation types, peak operation amplitude, collision status, and event end marker within the event observation window.

4. The driving ability evaluation data processing method applied to standardized driving test scenarios according to claim 1, characterized in that, Event isolation rules are set between adjacent events; when the event-level data recording unit of the first event has not yet reached the end condition, the second event is marked as a delayed trigger event; wherein, the end condition includes at least the main vehicle leaving the event influence area, a collision occurring, or the observation window timeout.

5. The driving ability evaluation data processing method applied to standardized driving test scenarios according to claim 1, characterized in that, The driving ability evaluation results also include full-process driving behavior parameters based on the output of the full-process time-series data stream.

6. A data processing system for evaluating driving ability in standardized driving test scenarios, characterized in that, It includes a processor, a memory, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the computer program, implements: The scene data loading module is configured to load pre-fixed standardized test scene data; the standardized test scene data includes at least static road layer data and dynamic event layer data, the dynamic event layer data includes multiple event records and trigger conditions corresponding to each event record; each event record includes at least event number, event category, three-dimensional coordinate range of trigger area, set of trigger conditions, trigger condition logic, and initial position and preset motion trajectory of the event participants; The data acquisition and synchronization module is configured to synchronously receive multi-channel driving operation signals and vehicle motion status data from the driving operation signal acquisition terminal at a sampling frequency of not less than 100Hz, based on a unified clock source, and add a unified timestamp with an accuracy of not more than 10 milliseconds to each sampling frame to form a time-series data stream. The multi-channel driving operation signals include at least steering wheel angle signal, throttle opening signal, and brake opening signal; The event trigger determination module is configured to extract the main vehicle's position, instantaneous speed, and current driving state in real time based on the time-series data stream, and perform matching calculations with the trigger conditions of untriggered event records in the dynamic event layer data. The trigger conditions include at least two of the following: position conditions, speed conditions, and driving state conditions. The position condition is used to determine whether the main vehicle's center point enters the three-dimensional coordinate range of the trigger area corresponding to the event. The speed condition is used to determine whether the main vehicle's instantaneous speed is within a preset speed range. The driving state condition is used to determine whether the current driving state matches the preset state identifier of the event record. When the preset trigger logic formed by combining at least two conditions using "AND" logic and "OR" logic is satisfied, the corresponding event trigger signal is output, and the preset event participants and their motion trajectories in the event record are invoked. The event recording and response analysis module is configured to create a corresponding event-level data recording unit in response to the event trigger signal, write a state snapshot at the event trigger time, the state snapshot including at least the three-dimensional coordinates of the main vehicle, instantaneous speed, steering wheel angle, throttle opening, brake opening, and current driving state at the event trigger time; and within a preset observation window after the event trigger, detect the moment when the throttle signal drops above a first preset threshold and the moment when the brake signal rises above a second preset threshold based on the time-series data stream, and determine the earliest sampling moment that meets the threshold condition as the first operation response moment of the corresponding channel only when the corresponding signal meets the threshold condition for N consecutive frames, where N is an integer greater than or equal to 2; and calculate the danger perception time and danger reaction time, the danger perception time being the difference between the moment when the throttle signal drops above the first preset threshold is first detected and the event trigger time, and the danger reaction time being the difference between the moment when the brake signal rises above the second preset threshold is first detected and the event trigger time; The evaluation result generation module is configured to output driving ability evaluation results based on the event-level data recording unit and the full-process time-series data stream. The driving ability evaluation results include at least the hazard perception time, hazard reaction time and collision result corresponding to each event.

7. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, causes the processor to perform the driving ability evaluation data processing method for a standardized driving test scenario as described in any one of claims 1 to 5.

Images