A driver road driving ability evaluation method based on big data

Abstract

This invention discloses a method for assessing driver's road driving ability based on big data, comprising: multi-dimensional collection of driving behavior data from instructors and trainees; setting standard event thresholds to identify and distinguish normal and abnormal behaviors; screening indicators in which the instructor's performance is significantly better than the trainee's during the interaction process, and quantifying these process-oriented behavioral indicators using the standard event thresholds; determining optimal weights with the instructor's score being better than the trainee's score as the optimization objective; and after determining the optimal weights for each behavioral indicator, calculating the total score for each item using weighted averages to achieve the assessment of the driver's road driving ability. This invention provides an effective tool for driver training through a scientific, objective, and comprehensive evaluation system, helping to improve trainees' driving skills and safety awareness, enhance the quality of training and testing, and ultimately improve road traffic safety.

Description

Technical Field

[0001] This invention relates to the field of motor vehicle driver training and testing technology, and in particular to a method for assessing a driver's road driving ability based on big data. Background Technology

[0002] In driver's road driving skills tests, most countries use outcome-oriented evaluation indicators, such as turn signals (failure to turn off turn signals after turning), crossing lane lines (wheels touching the edge of the road), stopping (stopping midway through the test), driving route (not following the prescribed route), speed (insufficient duration of maximum speed), gear matching (failure to smoothly shift gears as instructed), and braking (failure to apply the brakes to decelerate). For the scoring of the third part of the driving test, points are typically deducted based on the severity of the student's mistakes in each item, with scores ranging from 5 to 100 points. For example, a lurching motion at the start deducts 5 points, while unstable steering control deducts 100 points. However, this approach pays less attention to specific progress during training and lacks a scientific and quantitative evaluation of training effectiveness based on driving behavior data. This outcome-oriented evaluation method focuses primarily on the final test result, neglecting specific progress and continuous improvement in driving behavior during training. This evaluation method may not fully reflect the skill enhancement and behavioral changes of drivers during training.

[0003] Therefore, how to provide a method for assessing drivers' road driving ability based on big data has become a technical problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0004] In view of this, the purpose of this invention is to provide a method for assessing drivers' road driving ability based on big data. Through a scientific, objective, and comprehensive evaluation system, it provides an effective tool for driver training, helps to improve trainees' driving skills and safety awareness, enhances the quality of training and examinations, and ultimately improves road traffic safety.

[0005] The present invention solves the technical problem by adopting the following technical solution:

[0006] A method for assessing a driver's road driving ability based on big data includes the following steps:

[0007] S1, acquire training behavior data, and collect multi-dimensional driving behavior data of instructors and students;

[0008] S2, Determine the standard event threshold. Set the standard event threshold to identify and distinguish between normal and abnormal behavior;

[0009] S3, determine the behavioral indicator scores, screen the indicators in the interaction between the coach and the student where the coach and the student have significant differences and the coach is better than the student, and quantify these process behavioral indicators by using standard event thresholds.

[0010] S4. Determine the optimal indicator weights, with the optimization objective being that the coach's score is better than the student's score;

[0011] S5, calculate the project score. After determining the optimal weight of each behavioral indicator, calculate the total score of each project in a weighted manner to achieve the assessment of the driver's road driving ability.

[0012] Furthermore, in step S2, one or more standard event thresholds are determined by analyzing the distribution of driving behavior data, and the data is divided into different intervals based on the standard event thresholds.

[0013] Furthermore, the behavioral indicators are evaluated from three dimensions: safety, stability, and efficiency.

[0014] Furthermore, the safety measures include maximum steering wheel rotation speed, number of passive braking events, speed variation coefficient, number of rapid accelerations, number of rapid decelerations, and indicators such as failure to use turn signals and insufficient turn signal duration.

[0015] Furthermore, the stability includes training trajectory score, acceleration smoothness, starting deceleration position, maximum absolute value of steering wheel angle in the first 100 meters, number of positive and negative changes in steering wheel angle, and duration of speed and gear mismatch.

[0016] Furthermore, the efficiency includes an average speed indicator.

[0017] Furthermore, the calculation methods for the scores of each behavioral indicator are as follows:

[0018] Coefficient of variation of speed: It is an important indicator for measuring the stability of vehicle speed changes, and is the ratio of the standard deviation of speed to the average speed;

[0019] Number of positive and negative changes in steering wheel angle: Treat the smaller values ​​of steering wheel angle [-10°, 10°] as 0, calculate whether there is a positive or negative change in steering wheel angle within every 2 seconds, and if so, count it as 1 time, and count the total number of times for all 2-second time windows;

[0020] Acceleration and lateral acceleration smoothness: Calculate the Fourier transform of acceleration and obtain the frequency, calculate the energy of the frequency components, and define the cutoff frequency as 0.05; calculate the total energy and low-frequency energy, and the ratio of low-frequency energy to total energy is the lateral acceleration smoothness.

[0021] Starting deceleration position: Calculate the cumulative distance based on the speed. Continuous deceleration within 0.4s is considered deceleration. Calculate the position where the first deceleration occurs. If no deceleration position is found, it is considered that no deceleration has occurred.

[0022] Training trajectory score: Round the x-coordinate to the nearest integer, determine whether the y-coordinate is within the selected distance range, and count the percentage of points within the range, which is the training trajectory score.

[0023] The present invention discloses a method for assessing a driver's road driving ability based on big data, which has the following beneficial effects:

[0024] 1) Based on multi-dimensional data collection of driver behavior, this invention constructs a data-driven, process-oriented evaluation index system. This system evaluates learner driving behavior from three dimensions: safety, stability, and efficiency. Evaluation indicators such as maximum steering wheel speed, training trajectory score, and average speed comprehensively reflect learners' driving skills and safety levels, providing them with immediate feedback and personalized improvement suggestions. This helps learners adjust their driving behavior promptly and improve learning efficiency. This method can provide personalized training suggestions for each driver based on the evaluation results, helping them improve their driving skills. Simultaneously, by identifying undesirable driving behaviors, this method helps improve road safety and reduce traffic accidents.

[0025] 2) Based on multi-dimensional data collection during driver training, this invention constructs a standard event threshold evaluation method. By determining indicator thresholds, scores are assigned to items according to these thresholds, thereby obtaining scores for each indicator and determining the overall score for each item in Subject 3. This evaluation method comprehensively considers the student's performance on various evaluation indicators, providing a comprehensive driving ability score that more fairly reflects the student's actual driving level. The evaluation method not only focuses on passing the exam but also emphasizes cultivating good driving habits and safety awareness in students, helping them maintain safe driving behavior long-term after obtaining their driver's license. Attached Figure Description

[0026] Figure 1 This is a flowchart of the method of the present invention;

[0027] Figure 2 A schematic diagram showing the initial deceleration positions of the instructor and trainee.

[0028] Figure 3 This is a schematic diagram illustrating the trajectory characteristics of the coach and trainees in this invention;

[0029] Figure 4 This is a schematic diagram illustrating the similarity index features between the coach's and student's trajectories in this invention. Detailed Implementation

[0030] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0031] refer to Figure 1 The present invention discloses a method for assessing a driver's road driving ability based on big data, comprising the following steps:

[0032] S1 acquires training behavior data, collecting multi-dimensional driving behavior data from both instructors and students. The algorithm's input data includes the number of rapid accelerations and decelerations for different scenarios, training trajectories, acceleration smoothness, longitudinal acceleration smoothness, lateral acceleration smoothness, vehicle speed data, duration of speed-gear mismatch, number of sudden lane changes, and the number of positive and negative steering wheel angle changes for both instructors and students. First, the driving behavior time-series data is converted into distance coordinate data with a granularity of 1 meter.

[0033] S2, Determining Standard Event Thresholds: In data analysis and behavioral assessment, setting standard event thresholds is a crucial method for identifying and distinguishing between normal and abnormal behavior. One or more standard event thresholds are determined by analyzing the distribution of the data, dividing the data into different intervals. Once a data point exceeds these preset limits, it is assigned a different score based on the degree of deviation.

[0034] S3 identifies behavioral indicator scores. During the coach-student interaction, indicators where the coach significantly outperforms the student are selected. These process-oriented behavioral indicators are then quantified using standardized event thresholds to more accurately assess the coach's teaching effectiveness and the student's learning progress. The scoring system can be based on the differences between the coach and student, and whether the coach's performance meets preset standards.

[0035] S4. Determine the optimal indicator weights. In order to optimize the coach's teaching effect, it is necessary to determine the key indicators and assign weights accordingly to reflect the importance of the indicator in the overall teaching effect. The optimal weights are determined with the coach's score being better than the student's score as the optimization goal.

[0036] S5. Calculate the project score. After determining the optimal weight of each behavioral indicator, calculate the total score of each project in a weighted manner to reflect its degree of influence on the project. Determine the intervals of excellent, good, average and poor based on the distribution of scores, thereby providing a basis for improving teaching methods and enhancing learning efficiency.

[0037] The behavioral indicators are evaluated from three dimensions: safety, stability, and efficiency. Safety includes indicators such as maximum steering wheel speed, number of passive braking incidents, speed variation coefficient, number of rapid accelerations, number of rapid decelerations, failure to use turn signals, and insufficient turn signal duration. Stability includes indicators such as training trajectory score, acceleration smoothness, initial deceleration position, maximum absolute value of steering wheel angle in the first 100 meters, number of positive and negative steering wheel angle changes, and duration of speed and gear mismatch. Efficiency includes the average speed indicator. The evaluation indicators for each item are shown in Appendix Tables 1-9. The number of undesirable driving behaviors is also an important indicator for evaluating driving behavior in Subject 3.

[0038] Table 1 shows the evaluation index system for straight-line driving.

[0039]

[0040] Tables 1 and 2 show the evaluation index system for 100-meter acceleration and deceleration parking.

[0041]

[0042] Tables 2 and 3 show the evaluation index system for lane changing.

[0043]

[0044] Tables 3 and 4 show the evaluation index system for pedestrian crossings.

[0045]

[0046] Tables 4 and 5 show the evaluation index system for passing practice.

[0047]

[0048] Tables 5 and 6 show the evaluation index system for overtaking practice.

[0049]

[0050] Tables 6 and 7 show the evaluation index system for U-turn practice.

[0051]

[0052] Tables 7 and 8 show the evaluation index system for parallel parking.

[0053]

[0054] Table 8

[0055] Table 9 shows the evaluation index system for left turns at intersections.

[0056]

[0057] Table 9

[0058] The technical solution has been further optimized, and the calculation methods for the scores of each behavioral indicator are as follows:

[0059] Coefficient of variation of speed: It is an important indicator for measuring the stability of vehicle speed changes, and is the ratio of the standard deviation of speed to the average speed;

[0060] Number of positive and negative changes in steering wheel angle: Treat the smaller values ​​of steering wheel angle [-10°, 10°] as 0, calculate whether there is a positive or negative change in steering wheel angle within every 2 seconds, and if so, count it as 1 time, and count the total number of times for all 2-second time windows;

[0061] Acceleration and lateral acceleration smoothness: Calculate the Fourier transform of acceleration and obtain the frequency, calculate the energy of the frequency component (square of the amplitude), and define the cutoff frequency as 0.05; calculate the total energy and low-frequency energy, and the ratio of low-frequency energy is the acceleration smoothness.

[0062] Starting deceleration position: Calculate the cumulative distance based on the speed. Continuous deceleration within 0.4s is considered deceleration. Calculate the position where the first deceleration occurs. If no deceleration position is found, it is considered that no deceleration has occurred.

[0063] Training trajectory score: Round the x-coordinate to the nearest integer, determine whether the y-coordinate is within the selected distance range, and count the percentage of points within the range, which is the training trajectory score.

[0064] This invention is a road driving behavior assessment method developed based on the driving skills evaluation test for Subject 3. By collecting vehicle operation and driving behavior data, setting thresholds for each indicator based on their distribution, and determining optimal weights, the method assesses the driver's ability and operational level from the dimensions of safety, stability, and efficiency. The road driving skills test training is divided into nine sub-items: straight driving, 100-meter acceleration and deceleration, lane changing, crossing pedestrian crossings, meeting oncoming traffic, overtaking, U-turns, parking at the roadside, and left turns at intersections, involving a total of 74 indicators. Each sub-item includes multiple sub-items, each sub-item involves multiple evaluation items, and each evaluation item includes multiple driving behavior indicators. This invention aims to construct a process-oriented road driving skills evaluation indicator system based on driver control behavior data collected from real vehicles, and to develop a big data-based driver road driving behavior assessment method based on this indicator system. In terms of examination and training, a process-oriented road driving skills evaluation index system is adopted. It comprehensively evaluates drivers' driving operation skills during the training process from three dimensions: safety, stability, and efficiency. By determining the optimal weight, setting index thresholds, and determining the weighted total score, the evaluation process comprehensively, efficiently, and holistically reflects the driver's overall driving skills and operation level. It better captures every detail and behavior of the driver during the process and provides more comprehensive, real-time, and personalized feedback, thereby helping drivers to continuously improve and enhance their skills.

[0065] This invention provides an effective tool for driver training through a scientific, objective, and comprehensive evaluation system, which helps to improve trainees' driving skills and safety awareness, enhance the quality of training and testing, and ultimately improve road traffic safety.

[0066] Example

[0067] Taking the evaluation index of the starting deceleration position in the pedestrian crossing project as an example, such as Figure 2 , Figure 3 and Figure 4 As shown. The cumulative distance is calculated based on speed. Continuous deceleration within 0.4 seconds is considered deceleration, and the location of the first deceleration is calculated; if no deceleration location is found, it is considered that no deceleration occurred. The dataset is divided into units of meters. After removing outliers, the dataset ranges from 0 to 100 meters, as detailed below:

[0068] 1) Feature analysis and indicator determination: Describe the distribution of driving behavior characteristics of instructors and students, and screen indicators that show significant differences between instructors and students and are superior to those of students.

[0069] 2) Determine the threshold: Set the threshold for the indicator based on the distribution, and give different scores for different measures that exceed the limit.

[0070] The initial deceleration position is between -1m and 6.5m, excellent, 100.

[0071] Initial deceleration position is 6.5m-11.8m, good, 70

[0072] The initial deceleration point is between 11.8m and 15.6m, in the middle, at 50.

[0073] The initial deceleration point was more than 15.6m away, poor, 0.

[0074] 3) Weight Calculation: The weight of each indicator is calculated from the dimensions of safety, stability, and efficiency. The weight calculation of each indicator for the pedestrian crossing project is shown in Table 10:

[0075]

[0076] Table 10

[0077] 4) Overall Score Calculation: The total score is calculated using a weighted average, and the distribution is used to determine the Excellent, Good, Average, and Poor score ranges. The total score for each item is calculated using the following formula:

[0078] Total project score = Σ indicator score * weight.

[0079] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for evaluating a driver's road driving ability based on big data, characterized in that, Includes the following steps: S1, acquire training behavior data, and collect multi-dimensional driving behavior data of instructors and students; S2, Determine the standard event threshold. Set the standard event threshold to identify and distinguish between normal and abnormal behavior; S3, determine the behavioral indicator scores, screen the indicators in the interaction between the coach and the student where the coach and the student have significant differences and the coach is better than the student, and quantify these process behavioral indicators by using standard event thresholds. S4. Determine the optimal indicator weights, with the optimization objective being that the coach's score is better than the student's score; S5, Calculate the project score. After determining the optimal weight of each behavioral indicator, calculate the total score of each project in a weighted manner to achieve the assessment of the driver's road driving ability. The calculation methods for the scores of each behavioral indicator in step S3 are as follows: Coefficient of variation of speed: It is an important indicator for measuring the stability of vehicle speed changes, and is the ratio of the standard deviation of speed to the average speed; Number of positive and negative changes in steering wheel angle: Treat the smaller values ​​of steering wheel angle [-10°, 10°] as 0, calculate whether there is a positive or negative change in steering wheel angle within every 2 seconds, and if so, count it as 1 time, and count the total number of times for all 2-second time windows; Acceleration smoothness and lateral acceleration smoothness: Perform Fourier transform on the acceleration signal and lateral acceleration signal respectively, obtain the frequency, calculate the energy of each frequency component, and define the cutoff frequency as 0.05; calculate the total energy and low-frequency energy, and use the ratio of low-frequency energy to total energy as the corresponding smoothness index; Starting deceleration position: Calculate the cumulative distance based on the speed. Continuous deceleration within 0.4s is considered deceleration. Calculate the position where the first deceleration occurs. If no deceleration point is found, it is considered that no deceleration has occurred; Training trajectory score: Round the x-coordinate to the nearest integer, determine whether the y-coordinate is within the selected distance range, and count the percentage of points within the range, which is the training trajectory score. 2.The method of claim 1, wherein, In step S2, one or more standard event thresholds are determined by analyzing the distribution of driving behavior data, and the data is divided into different intervals according to the standard event thresholds. 3.The method of claim 2, wherein, The behavioral indicators are evaluated from three dimensions: safety, stability, and efficiency.

4. The method for assessing a driver's road driving ability based on big data according to claim 3, characterized in that, The safety metrics include maximum steering wheel rotation speed, number of passive braking events, speed variation coefficient, number of rapid accelerations, number of rapid decelerations, failure to use turn signals, and insufficient turn signal duration.

5. The method for assessing a driver's road driving ability based on big data according to claim 4, characterized in that, The stability includes training trajectory score, acceleration smoothness, starting deceleration position, maximum absolute value of steering wheel angle in the first 100 meters, number of positive and negative changes in steering wheel angle, and duration of speed and gear mismatch.

6. The method for assessing a driver's road driving ability based on big data according to claim 5, characterized in that, The efficiency includes the average speed indicator.

Images