CN111707476A - Longitudinal driving capability detection method for automatic driving automobile - Google Patents
Longitudinal driving capability detection method for automatic driving automobile Download PDFInfo
- Publication number
- CN111707476A CN111707476A CN202010432644.3A CN202010432644A CN111707476A CN 111707476 A CN111707476 A CN 111707476A CN 202010432644 A CN202010432644 A CN 202010432644A CN 111707476 A CN111707476 A CN 111707476A
- Authority
- CN
- China
- Prior art keywords
- driving
- longitudinal
- vehicle
- automobile
- data
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
- G01M17/00—Testing of vehicles
- G01M17/007—Wheeled or endless-tracked vehicles
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/15—Correlation function computation including computation of convolution operations
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/16—Matrix or vector computation, e.g. matrix-matrix or matrix-vector multiplication, matrix factorization
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Computational Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Mathematical Analysis (AREA)
- Mathematical Optimization (AREA)
- Pure & Applied Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Algebra (AREA)
- Databases & Information Systems (AREA)
- Software Systems (AREA)
- General Engineering & Computer Science (AREA)
- Computing Systems (AREA)
- Traffic Control Systems (AREA)
- Control Of Driving Devices And Active Controlling Of Vehicle (AREA)
Abstract
The invention relates to a longitudinal driving ability detection method for an automatic driving automobile, which comprises the following steps: acquiring heterogeneous group driving behavior data and constructing a longitudinal driving behavior feature library; aiming at the longitudinal safe distance keeping behavior of the vehicle, a driving capability evaluation index system is constructed by utilizing a longitudinal driving behavior feature library; constructing a longitudinal driving capability multi-factor analysis model by using a quantile regression method and a driving capability evaluation index system; and acquiring the running data of the automatic driving automobile, and carrying out quantitative detection on the driving capability by utilizing a longitudinal driving capability multi-factor analysis model. Compared with the prior art, the method has the advantages that the quantitative positioning of the driving capacity of the automatic driving automobile in human driver groups is realized, the problems that the existing driving capacity evaluation method cannot give consideration to grade division and interpretability are solved, and the evaluation result can support the optimization of the control algorithm of the automatic driving automobile.
Description
Technical Field
The invention relates to the field of automatic driving automobiles, in particular to a longitudinal driving capability detection method for an automatic driving automobile.
Background
Automatic driving is an important direction for a new round of scientific and technological innovation and industry development all over the world. At present, autopilot has entered into the practical development stage, still is in industrialization earlier stage, and autopilot drive test accident takes place occasionally, and the security problem that leads to the fact owing to the driving ability is not enough leads to public to reduce by a wide margin to autopilot's trust, consequently need carry out the driving ability analysis before the autopilot is used, and the autopilot method of application high driving ability goes to control autopilot to guarantee the security of independently dealing with the driving task in complicated traffic environment.
Disclosure of Invention
The invention aims to overcome the defects of the prior art and provide a longitudinal driving capability detection method for an automatic driving automobile.
The purpose of the invention can be realized by the following technical scheme:
a longitudinal driving ability detection method for an automatic driving automobile comprises the following steps:
step S1: acquiring heterogeneous group driving behavior data and constructing a longitudinal driving behavior feature library;
step S2: aiming at the longitudinal safe distance keeping behavior of the vehicle, a driving capability evaluation index system is constructed by utilizing a longitudinal driving behavior feature library;
step S3: constructing a longitudinal driving capability multi-factor analysis model by using a quantile regression method and a driving capability evaluation index system;
step S4: and acquiring the running data of the automatic driving automobile, and carrying out quantitative detection on the driving capability by utilizing a longitudinal driving capability multi-factor analysis model.
The step S1 includes:
step S11: acquiring heterogeneous group driving behavior data, including vehicle motion data, GPS driving track data and vehicle-mounted driving environment perception data;
step S12: fusing heterogeneous group driving behavior data, dividing driving segments, extracting the driving segments of another vehicle in the driving forward direction from the driving segments to obtain driving capability evaluation segments;
step S13: and calculating travel characteristics, running characteristics and road characteristics by taking the driving ability evaluation segment as a basic analysis unit, and constructing a longitudinal driving behavior characteristic library.
The travel characteristic indexes are duration of the driving ability assessment segment, whether peak travel is performed or not and whether night travel is performed or not, the operation characteristic indexes are average speed and standard deviation of speed, and the road characteristic indexes are main road types.
The driving ability evaluation index system comprises longitudinal risk exposure pexposureAnd longitudinal risk degree pseverityTwo indices, the longitudinal risk exposure pexposureComprises the following steps:
wherein, tauiIn order to collect the interval for the driving behavior data,iis tauiThe state of the vehicle in the time interval, d is the distance between the vehicle and another adjacent vehicle in the forward direction, and dmin(ti) At a minimum safe distance, d (t)i)＜dmin(ti) Time of flighti1, i is the total time period number, t is the duration of the driving ability evaluation segment, and exposure is the sum of the durations of the vehicle in the potential risk state;
the longitudinal risk degree pseverityComprises the following steps:
wherein, the term "visibility" refers to d (t) when the host vehicle is in a risk potential statei) And dmin(ti) The sum of the differences of (a).
D ismin(ti) Comprises the following steps:
wherein v ise(ti) And vl(ti) The speeds of the own vehicle and the forward adjacent other vehicle, a, respectively, during the period imax,aIs the maximum acceleration of the vehicle, amin,bMinimum deceleration of the vehicle, bmax,bρ is the reaction time of the driver of the vehicle/autonomous vehicle for the maximum deceleration of the forward adjacent other vehicle.
The longitudinal driving ability multi-factor analysis model is as follows:
wherein, yjEvaluating the longitudinal risk exposure p corresponding to the segment j for drivabilityexposureOr longitudinal risk degree pseverityP is the quantitative grade positioning quantile of the designated driving ability, xj＝(Durationj,Peakj,Nightj,Avg.Speedj,Std.Speedj,Roadj) To interpret the vectors, the items sequentially correspond to the duration of the driving ability assessment segment j, whether to travel at peak, whether to travel at night, average speed, standard deviation of speed, and main road type, respectively, βpIs a regression coefficient vector.
The regression coefficient vector βpSolving by weighted residual minimization, and calculating as follows:
where ρ ispIs the weight of the residual under p quantiles, and n is the number of samples.
The step S4 includes:
step S41: acquiring operation data of the automatic driving automobile, wherein the operation data comprises movement data of the automatic driving automobile, GPS (global positioning system) driving track data of the automatic driving automobile and vehicle-mounted driving environment detection data of the automatic driving automobile;
step S42: extracting a driving capability evaluation segment of the automatic driving automobile and calculating a basic characteristic index of driving behavior of the automatic driving automobile and a measured value of the driving capability evaluation index;
step S43: calculating driving ability evaluation index values corresponding to the quantitative grade positioning quantiles of the appointed driving ability for each driving ability evaluation segment of the automatic driving automobile based on the longitudinal driving ability multi-factor analysis model;
step S44: and comparing the driving ability evaluation index value with the driving ability evaluation index measured value, wherein the quantile interval where the measured value is located is the driving ability quantitative detection result.
Compared with the prior art, the invention has the following advantages:
(1) the method is oriented to the 'personification' operation characteristic evaluation requirement of the automatic driving automobile, constructs a driving behavior specific library of heterogeneous drivers, and provides a driving capability evaluation index system suitable for continuous and complex driving environments.
(2) Based on a quantile regression method, driving capacity grading reference values under different traveling, running and road conditions can be determined, and quantitative positioning of the driving capacity of the automatic driving automobile in a human driver group is realized; in addition, the relation between the influence factors such as travel, operation, roads and the like and the driving ability can be known through the quantile regression model result, and the problem that the existing driving ability evaluation method cannot give consideration to automatic classification and interpretability is solved; control of an autonomous vehicle may be optimized.
Drawings
Fig. 1 is a flow chart of the method for detecting the longitudinal driving ability of an autonomous vehicle according to the present invention.
Detailed Description
The invention is described in detail below with reference to the figures and specific embodiments. The present embodiment is implemented on the premise of the technical solution of the present invention, and a detailed implementation manner and a specific operation process are given, but the scope of the present invention is not limited to the following embodiments.
Examples
The invention provides a longitudinal driving ability detection method for an automatic driving automobile, which comprises the following steps:
step S1: acquiring heterogeneous group driving behavior data and constructing a longitudinal driving behavior feature library;
step S2: aiming at the longitudinal safe distance keeping behavior of the vehicle, a driving capability evaluation index system is constructed by utilizing a longitudinal driving behavior feature library;
step S3: constructing a longitudinal driving capability multi-factor analysis model by using a quantile regression method and a driving capability evaluation index system;
step S4: and acquiring the running data of the automatic driving automobile, and carrying out quantitative detection on the driving capability by utilizing a longitudinal driving capability multi-factor analysis model.
TABLE 1 basic characteristic index of driving behavior
| Dimension (d) of | Index (I) |
| Characteristic of trip | Duration, whether to go out at peak or not, and whether to go out at night |
| Operating characteristics | Mean velocity, standard deviation of velocity |
| Road characteristics | Main road type (road type with the highest ratio of distance traveled in the assessment segment) |
Further, the step S1 of acquiring heterogeneous group driving behavior data and constructing the longitudinal driving behavior feature library specifically includes:
(1.1) acquiring heterogeneous group driving behavior data, including vehicle motion data, GPS driving track data and vehicle-mounted driving environment perception data;
(1.2) multi-source driving data fusion: fusing vehicle motion data, GPS driving track data and vehicle-mounted driving environment perception data, and dividing driving segments;
(1.3) extraction of driving ability evaluation fragment: extracting a driving section in which another vehicle is detected to exist in the driving forward direction of the vehicle as a driving ability evaluation section;
(1.4) calculating a driving behavior basic characteristic index: and calculating travel characteristics, running characteristics and road characteristics by taking the driving ability evaluation segment as a basic analysis unit, and constructing a longitudinal driving behavior characteristic library. The driving behavior basic characteristic index is shown in table 1.
In step S2, a driving ability evaluation index system is constructed by using the longitudinal driving behavior feature library for the vehicle longitudinal safe distance maintenance behavior, specifically:
(2.1) for the vehicle longitudinal safe distance keeping behavior, judging the vehicle safe state through an RSS (Responsibility sensitive safety) model:
RSS is an IEEE autonomous vehicle safety decision model standard that defines the minimum safe distance of an autonomous vehicle from a forward-adjacent other vehicle:
wherein v ise(ti) And vl(ti) The speeds of the own vehicle and the forward adjacent other vehicle, a, respectively, during the period imax,aIs the maximum acceleration of the vehicle, amin,bMinimum deceleration of the vehicle, bmax,bFor the maximum deceleration of the preceding vehicle, ρ is the reaction time of the driver/autonomous vehicle, if the vehicle is adjacent to another vehicle in the forward direction, d (t)i)＜dmin(ti) Then the vehicle is in a potential risk state;
(2.2) constructing a driving ability evaluation index system based on an RSS model:
drivability assessment index system from longitudinal risk exposure pexposureAnd longitudinal risk degree pseverityThe method comprises the following steps:
wherein, tauiAcquiring intervals for driving behavior data;iis tauiVehicle state during time period, d (t)i)＜dmin(ti) Time of flighti1, the vehicle is in a potentially risky state; t is the duration of the driving ability evaluation segment; exposure is the distance d (t) between the host vehicle and another vehicle adjacent to the host vehicle in the forward direction in the driving ability evaluation segmenti)＜dmin(ti) I.e. sum of time periods that the vehicle is in a potential risk state, longitudinal risk exposure pexposureThe time length proportion of the vehicle in the potential risk state is obtained; "d" (t) when the host vehicle is in a potential risk state "visibilityi) And dmin(ti) Sum of the differences of (a), longitudinal risk degree pseverityD (t) is the time when the vehicle is in a potential risk statei) And dmin(ti) The average of the differences.
In step S3, a longitudinal driving ability multi-factor analysis model is constructed based on the quantile regression analysis method and the driving ability evaluation index system, specifically:
(3.1) selecting a driving ability quantization grade positioning quantile;
(3.2) constructing a longitudinal driving ability multi-factor analysis model based on a quantile regression method and a driving ability evaluation index system:
the longitudinal driving capability multi-factor analysis model comprises the following steps:
wherein x isj＝(Durationj,Peakj,Nightj,Avg.Speedj,Std.Speedj,Roadj) For explaining the vector, the items sequentially correspond to the duration of the driving ability evaluation segment j, whether the vehicle is going out at a peak or not, whether the vehicle is going out at night or not, the average speed, the speed standard deviation and the main road type respectively; response variable yjEvaluating the longitudinal risk exposure p corresponding to the segment j for drivabilityexposureOr longitudinal risk degree pseverityP is the designated quantile βpAs a vector of regression coefficients, KetongSolving for the minimum of the over-weighted residuals, namely:
where ρ ispIs the weight of the residual under p quantiles, and n is the number of samples. And solving the above formula to obtain the regression coefficient vector of the quantile model, namely obtaining the longitudinal driving capacity multi-factor analysis model.
Step S4 is to quantify the driving ability of the autonomous vehicle operation data using the longitudinal driving ability multi-factor analysis model:
(4.1) acquiring the running data of the automatic driving automobile, wherein the running data comprises the motion data of the automatic driving automobile, the GPS driving track data of the automatic driving automobile, the vehicle-mounted driving environment detection data of the automatic driving automobile and the like;
(4.2) extracting a driving capability evaluation segment of the automatic driving automobile and calculating a basic characteristic index of the driving behavior of the automatic driving automobile, a longitudinal risk exposure amount (measured value) of the automatic driving automobile and a longitudinal risk exposure degree value (measured value) of the automatic driving automobile according to a longitudinal driving behavior feature library construction method;
(4.3) calculating longitudinal risk exposure (estimated value) and longitudinal risk exposure degree value (estimated value) corresponding to each grade (quantile level) under corresponding trip, operation and road characteristics for each driving capacity evaluation segment of automatic driving based on a longitudinal driving capacity multi-factor analytical model;
(4.4) comparing the longitudinal risk exposure amount corresponding to each grade (quantile level) with the longitudinal risk exposure amount of the automatic driving automobile, the longitudinal risk exposure degree value corresponding to each grade (quantile level) and the longitudinal risk exposure degree of the automatic driving automobile, wherein the quantile interval where the longitudinal risk exposure amount of the automatic driving automobile and the longitudinal risk exposure degree of the automatic driving automobile are located is the quantitative grading result of the automatic driving ability; the corresponding grade of the longitudinal risk exposure represents the driving style under specific travel, operation and road characteristics, and the higher the quantile level is, the more aggressive the driving style is; the corresponding grade of the longitudinal risk degree represents the driving skill under specific trip, operation and road characteristics, and the higher the quantile level is, the lower the driving skill is.
The following description is made in connection with a specific alternative embodiment:
constructing a longitudinal driving behavior feature library based on heterogeneous group driving behavior data:
(1.1) acquiring heterogeneous group driving behavior data, including vehicle motion data, GPS driving track data, vehicle-mounted driving environment perception data and the like;
(1.2) multi-source driving data fusion: fusing vehicle motion data, GPS driving track data and vehicle-mounted driving environment perception data, and dividing driving segments;
(1.3) extraction of driving ability evaluation fragment: extracting a driving segment of another vehicle detected to exist in the driving forward direction of the vehicle, taking the driving segment as a driving ability evaluation segment, extracting 508 driving ability evaluation segments in total, and accumulating 4297km of driving behavior data;
TABLE 2 description of the basic characteristic indexes of specific driving behaviors
(1.4) calculating a driving behavior basic characteristic index: calculating travel characteristics, operation characteristics and road characteristics by taking the driving ability evaluation segment as a basic analysis unit; the concrete driving behavior basic characteristic index descriptions of the examples are shown in table 2.
And (II) aiming at the behavior of keeping the longitudinal safe distance of the vehicle, constructing a driving ability evaluation index system by utilizing a longitudinal driving behavior feature library:
(2.1) for the vehicle longitudinal safe distance keeping behavior, judging the vehicle safe state through an RSS (Responsibility sensitive safety) model:
RSS is an IEEE autonomous vehicle safety decision model standard that defines the minimum safe distance of an autonomous vehicle from a forward-adjacent other vehicle:
wherein v ise(ti) And vl(ti) The speeds of the own vehicle and the forward adjacent other vehicle, a, respectively, during the period imax,aIs the maximum acceleration of the vehicle, amin,bMinimum deceleration of the vehicle, bmax,bFor the maximum deceleration of the preceding vehicle, ρ is the reaction time of the driver/autonomous vehicle, if the vehicle is adjacent to another vehicle in the forward direction, d (t)i)＜dmin(ti) Then the vehicle is in a potential risk state;
(2.2) constructing a driving ability evaluation index system based on an RSS model:
drivability assessment index system from longitudinal risk exposure pexposureAnd longitudinal risk degree pseverityThe method comprises the following steps:
wherein, tauiAcquiring intervals for driving behavior data;iis tauiVehicle state during time period, d (t)i)＜dmin(ti) Time of flighti1, the vehicle is in a potentially risky state; t is the duration of the driving ability evaluation segment; exposure is the distance d (t) between the host vehicle and another vehicle adjacent to the host vehicle in the forward direction in the driving ability evaluation segmenti)＜dmin(ti) I.e. sum of time periods that the vehicle is in a potential risk state, longitudinal risk exposure pexposureThe time length proportion of the vehicle in the potential risk state is obtained; "d" (t) when the host vehicle is in a potential risk state "visibilityi) And dmin(ti) Sum of the differences of (a), longitudinal risk degree pseverityD (t) is the time when the vehicle is in a potential risk statei) And dmin(ti) The average of the differences.
And (III) constructing a longitudinal driving ability multi-factor analysis model based on a quantile regression analysis method and a driving ability evaluation index system:
(3.1) selecting a driving ability quantization grade positioning quantile, wherein quantiles 0.1, 0.25, 0.5, 0.75, 0.9 and 0.95 are used as representative quantiles for driving ability grade division in the embodiment;
(3.2) constructing a longitudinal driving capability multi-factor analysis model based on a quantile regression method and human driving behavior data:
wherein x isj＝(Durationj,Peakj,Nightj,Avg.Speedj,Std.Speedj,Roadj) For explaining the vector, respectively corresponding to the duration of the driving ability evaluation segment j, whether to travel at a peak or not, whether to travel at night, the average speed, the speed standard deviation and the main road type; q is model estimation; response variable yjEvaluating the longitudinal risk exposure p corresponding to the segment j for drivabilityexposureOr longitudinal risk degree pseverityP is the designated quantile βpFor the regression coefficient vector, the solution can be performed by weighted residual minimization, i.e.:
where ρ ispIs the weight of the residual under p quantiles, and n is the number of samples. The longitudinal driving ability multi-factor analysis model of the present embodiment includes 6 longitudinal driving ability analysis units.
And (IV) quantifying the driving ability by utilizing a longitudinal driving ability multi-factor analysis model aiming at the running data of the automatic driving automobile:
(4.1) acquiring the running data of the automatic driving automobile, wherein the running data comprises the motion data of the automatic driving automobile, the GPS driving track data of the automatic driving automobile, the vehicle-mounted driving environment detection data of the automatic driving automobile and the like;
(4.2) extracting a driving capability evaluation segment of the automatic driving automobile and calculating a basic characteristic index of the driving behavior of the automatic driving automobile, a longitudinal risk exposure amount (measured value) of the automatic driving automobile and a longitudinal risk exposure degree value (measured value) of the automatic driving automobile according to a longitudinal driving behavior feature library construction method;
(4.3) calculating longitudinal risk exposure (estimated value) and longitudinal risk exposure degree value (estimated value) corresponding to each grade (quantile level) under corresponding trip, operation and road characteristics for each driving capacity evaluation segment of automatic driving based on a longitudinal driving capacity multi-factor analytical model;
(4.4) comparing the longitudinal risk exposure amount corresponding to each grade (quantile level) with the longitudinal risk exposure amount of the automatic driving automobile, the longitudinal risk exposure degree value corresponding to each grade (quantile level) and the longitudinal risk exposure degree of the automatic driving automobile, wherein the quantile interval where the longitudinal risk exposure amount of the automatic driving automobile and the longitudinal risk exposure degree of the automatic driving automobile are located is the quantitative grading result of the automatic driving ability; the corresponding grade of the longitudinal risk exposure represents the driving style under specific travel, operation and road characteristics, and the higher the quantile level is, the more aggressive the driving style is; the corresponding grade of the longitudinal risk degree represents the driving skill under specific trip, operation and road characteristics, and the higher the quantile level is, the lower the driving skill is.
Table 3 shows the driving behavior basic feature indexes, the longitudinal risk exposure or the longitudinal risk degree measured values, and the quantile model estimation values of the 2 driving ability evaluation segments of the autonomous vehicle; by comparing the measured values with the model estimate values for the specified quantile levels, the relative level of quantified driveability can be determined. For example, the quantile intervals corresponding to the longitudinal risk exposure and the longitudinal risk degree of the driving ability evaluation segment 1 are all [25 ]th,50th) In a middle-lower section of the driver group, the automatic driving automobile shows a mild driving style and a middle-upper driving skill in the operation section, that is, under the same trip, operation and road conditions, the time proportion of the automatic driving automobile in a potential risk state in the operation process is relatively low, and the actual distance between the automatic driving automobile and the adjacent front automobile in an unsafe state and the RSS minimum safety are achievedThe overall distance gap is smaller and therefore the average risk level is lower. Quantile intervals corresponding to the longitudinal risk exposure and the longitudinal risk degree of the driving ability evaluation segment 2 are respectively [75 ]th,90th) And [50 ]th,75th) It is explained that the autonomous vehicle shows an aggressive driving style and a moderate driving skill in this driving segment. In addition, the longitudinal risk exposure and the longitudinal risk degree measured values of the driving ability evaluation segment 1 and the driving ability evaluation segment 2 are similar, and the classification standard and the classification result are different due to the difference of conditions such as travel and operation.
TABLE 3 automated Driving ability quantitative evaluation case
Through the steps, the driving capacity quantification of the automatic driving automobile is completed, parameters such as driving speed and the like can be improved according to the scene of the driving capacity evaluation segment 2, and the optimization control of the automatic driving automobile is realized.
The embodiment has the following advantages:
the method is oriented to the 'personification' operation characteristic evaluation requirement of the automatic driving automobile, constructs a driving behavior specific library of a heterogeneous driver, and provides a driving capability evaluation index system suitable for continuous and complex driving environments; based on a quantile regression method, driving capacity grading reference values under different traveling, running and road conditions can be determined, and quantitative positioning of the driving capacity of the automatic driving automobile in a human driver group is realized; in addition, the relation between the influence factors such as travel, operation, roads and the like and the driving ability can be known through the quantile regression model result, and the problem that the existing driving ability evaluation method cannot give consideration to automatic classification and interpretability is solved; control of an autonomous vehicle may be optimized.
Claims (8)
1. A longitudinal driving ability detection method for an automatic driving automobile is characterized by comprising the following steps:
step S1: acquiring heterogeneous group driving behavior data and constructing a longitudinal driving behavior feature library;
step S2: aiming at the longitudinal safe distance keeping behavior of the vehicle, a driving capability evaluation index system is constructed by utilizing a longitudinal driving behavior feature library;
step S3: constructing a longitudinal driving capability multi-factor analysis model by using a quantile regression method and a driving capability evaluation index system;
step S4: and acquiring the running data of the automatic driving automobile, and carrying out quantitative detection on the driving capability by utilizing a longitudinal driving capability multi-factor analysis model.
2. The method for detecting the longitudinal driving ability of the automatic driving vehicle according to claim 1, wherein the step S1 comprises:
step S11: acquiring heterogeneous group driving behavior data, including vehicle motion data, GPS driving track data and vehicle-mounted driving environment perception data;
step S12: fusing heterogeneous group driving behavior data, dividing driving segments, extracting the driving segments of another vehicle in the driving forward direction from the driving segments to obtain driving capability evaluation segments;
step S13: and calculating travel characteristics, running characteristics and road characteristics by taking the driving ability evaluation segment as a basic analysis unit, and constructing a longitudinal driving behavior characteristic library.
3. The method as claimed in claim 2, wherein the indexes of the travel characteristics are duration of the driving ability assessment segment, whether peak travel is performed, whether night travel is performed, the indexes of the operation characteristics are average speed and standard deviation of speed, and the indexes of the road characteristics are main road types.
4. The method of claim 1The method for detecting the longitudinal driving ability of the automatic driving automobile is characterized in that the driving ability evaluation index system comprises a longitudinal risk exposure pexposureAnd longitudinal risk degree pseverityTwo indices, the longitudinal risk exposure pexposureComprises the following steps:
wherein, tauiIn order to collect the interval for the driving behavior data,iis tauiThe state of the vehicle in the time interval, d is the distance between the vehicle and another adjacent vehicle in the forward direction, and dmin(ti) At a minimum safe distance, d (t)i)＜dmin(ti) Time of flighti1, i is the total time period number, t is the duration of the driving ability evaluation segment, and exposure is the sum of the durations of the vehicle in the potential risk state;
the longitudinal risk degree pseverityComprises the following steps:
wherein, the term "visibility" refers to d (t) when the host vehicle is in a risk potential statei) And dmin(ti) The sum of the differences of (a).
5. The method as claimed in claim 4, wherein d is a measure of the longitudinal drivability of the autonomous vehiclemin(ti) Comprises the following steps:
wherein v ise(ti) And vl(ti) The speeds of the own vehicle and the forward adjacent other vehicle, a, respectively, during the period imax,aIs the maximum acceleration of the vehicle, amin,bMinimum deceleration of the vehicle, bmax,bFor another vehicle adjacent in the forward directionρ is the reaction time of the driver of the vehicle/autonomous vehicle.
6. The method for detecting the longitudinal driving ability of the automatic driving automobile according to claim 1, wherein the longitudinal driving ability multi-factor analytic model is as follows:
wherein, yjEvaluating the longitudinal risk exposure p corresponding to the segment j for drivabilityexposureOr longitudinal risk degree pseverityP is the quantitative grade positioning quantile of the designated driving ability, xj＝(Durationj,Peakj,Nightj,Avg.Speedj,Std.Speedj,Roadj) To interpret the vectors, the items sequentially correspond to the duration of the driving ability assessment segment j, whether to travel at peak, whether to travel at night, average speed, standard deviation of speed, and main road type, respectively, βpIs a regression coefficient vector.
7. The method of claim 6, wherein the regression coefficient vector β is the vector of regression coefficientspSolving by weighted residual minimization, and calculating as follows:
where ρ ispIs the weight of the residual under p quantiles, and n is the number of samples.
8. The method for detecting the longitudinal driving ability of the automatic driving vehicle according to claim 1, wherein the step S4 comprises:
step S41: acquiring operation data of the automatic driving automobile, wherein the operation data comprises movement data of the automatic driving automobile, GPS (global positioning system) driving track data of the automatic driving automobile and vehicle-mounted driving environment detection data of the automatic driving automobile;
step S42: extracting a driving capability evaluation segment of the automatic driving automobile and calculating a basic characteristic index of driving behavior of the automatic driving automobile and a measured value of the driving capability evaluation index;
step S43: calculating driving ability evaluation index values corresponding to the quantitative grade positioning quantiles of the appointed driving ability for each driving ability evaluation segment of the automatic driving automobile based on the longitudinal driving ability multi-factor analysis model;
step S44: and comparing the driving ability evaluation index value with the driving ability evaluation index measured value, wherein the quantile interval where the measured value is located is the driving ability quantitative detection result.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010432644.3A CN111707476B (en) | 2020-05-20 | 2020-05-20 | Longitudinal driving capability detection method for automatic driving automobile |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010432644.3A CN111707476B (en) | 2020-05-20 | 2020-05-20 | Longitudinal driving capability detection method for automatic driving automobile |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111707476A true CN111707476A (en) | 2020-09-25 |
| CN111707476B CN111707476B (en) | 2021-07-20 |
Family
ID=72539116
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010432644.3A Active CN111707476B (en) | 2020-05-20 | 2020-05-20 | Longitudinal driving capability detection method for automatic driving automobile |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111707476B (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113111004A (en) * | 2021-01-11 | 2021-07-13 | 北京赛目科技有限公司 | Scene-independent unmanned driving simulation test evaluation method and device |
| CN113257045A (en) * | 2021-07-14 | 2021-08-13 | 四川腾盾科技有限公司 | Unmanned aerial vehicle control method based on large-scale fixed wing unmanned aerial vehicle electronic fence |
| CN117113045A (en) * | 2023-10-24 | 2023-11-24 | 交通运输部公路科学研究所 | Method for evaluating effectiveness of automatic driving positioning system |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109406166A (en) * | 2018-10-30 | 2019-03-01 | 百度在线网络技术（北京）有限公司 | Stage division, device, equipment, storage medium and the vehicle of unmanned vehicle |
| CN109443794A (en) * | 2018-10-26 | 2019-03-08 | 百度在线网络技术（北京）有限公司 | Evaluation method, device, computer equipment and the storage medium of automatic driving car |
| CN109784630A (en) * | 2018-12-12 | 2019-05-21 | 北京百度网讯科技有限公司 | Automatic Pilot level evaluation method, device, computer equipment and storage medium |
| CN109849816A (en) * | 2019-02-01 | 2019-06-07 | 公安部交通管理科学研究所 | A kind of autonomous driving vehicle driving ability evaluating method, apparatus and system |
| CN110782125A (en) * | 2019-09-23 | 2020-02-11 | 同济大学 | Road safety risk degree evaluation method for automatically driving automobile |
-
2020
- 2020-05-20 CN CN202010432644.3A patent/CN111707476B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109443794A (en) * | 2018-10-26 | 2019-03-08 | 百度在线网络技术（北京）有限公司 | Evaluation method, device, computer equipment and the storage medium of automatic driving car |
| CN109406166A (en) * | 2018-10-30 | 2019-03-01 | 百度在线网络技术（北京）有限公司 | Stage division, device, equipment, storage medium and the vehicle of unmanned vehicle |
| CN109784630A (en) * | 2018-12-12 | 2019-05-21 | 北京百度网讯科技有限公司 | Automatic Pilot level evaluation method, device, computer equipment and storage medium |
| CN109849816A (en) * | 2019-02-01 | 2019-06-07 | 公安部交通管理科学研究所 | A kind of autonomous driving vehicle driving ability evaluating method, apparatus and system |
| CN110782125A (en) * | 2019-09-23 | 2020-02-11 | 同济大学 | Road safety risk degree evaluation method for automatically driving automobile |
Non-Patent Citations (1)
| Title |
|---|
| 孙博华 等: "《驾驶人纵侧向驾驶能力评价方法研究》", 《北京理工大学学报》 * |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113111004A (en) * | 2021-01-11 | 2021-07-13 | 北京赛目科技有限公司 | Scene-independent unmanned driving simulation test evaluation method and device |
| CN113111003A (en) * | 2021-01-11 | 2021-07-13 | 北京赛目科技有限公司 | Scene-independent unmanned driving simulation test evaluation method and device |
| CN113111003B (en) * | 2021-01-11 | 2022-05-03 | 北京赛目科技有限公司 | Scene-independent unmanned driving simulation test evaluation method and device |
| CN113111004B (en) * | 2021-01-11 | 2022-05-17 | 北京赛目科技有限公司 | Scene-independent unmanned driving simulation test evaluation method and device |
| CN113257045A (en) * | 2021-07-14 | 2021-08-13 | 四川腾盾科技有限公司 | Unmanned aerial vehicle control method based on large-scale fixed wing unmanned aerial vehicle electronic fence |
| CN117113045A (en) * | 2023-10-24 | 2023-11-24 | 交通运输部公路科学研究所 | Method for evaluating effectiveness of automatic driving positioning system |
| CN117113045B (en) * | 2023-10-24 | 2024-01-26 | 交通运输部公路科学研究所 | Method for evaluating effectiveness of automatic driving positioning system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111707476B (en) | 2021-07-20 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111707476B (en) | Longitudinal driving capability detection method for automatic driving automobile | |
| EP2255347B9 (en) | Travel pattern information obtaining device, travel pattern information obtaining method, travel pattern information obtaining program and computer readable medium | |
| CN103562978B (en) | Vehicle data analysis method and vehicle data analysis system | |
| CN104809878B (en) | Method for detecting abnormal condition of urban road traffic by utilizing GPS (Global Positioning System) data of public buses | |
| CN108091137B (en) | Method and device for evaluating signal lamp control scheme | |
| CN110660270B (en) | Method for establishing vehicle collision risk evaluation model and collision risk evaluation method | |
| CN111695241A (en) | Method for determining length of accelerating lane of left-turn ramp confluence area based on VISSIM simulation | |
| EP2856452B1 (en) | Carriageway recognition | |
| JP7051424B2 (en) | Anomaly detection device | |
| CN113127466B (en) | Vehicle track data preprocessing method and computer storage medium | |
| CN110335467B (en) | Method for realizing highway vehicle behavior detection by using computer vision | |
| CN111942397A (en) | Dangerous driving behavior monitoring method and device and storage medium | |
| US9460614B2 (en) | Method and apparatus for determining traffic status | |
| CN117746640B (en) | Road traffic flow rolling prediction method, system, terminal and medium | |
| CN113611157A (en) | Method for estimating rear-end collision risk of vehicles on highway | |
| US20200257910A1 (en) | Method for automatically identifying parking areas and/or non-parking areas | |
| CN109344903A (en) | Urban road surfaces failure real-time detection method based on vehicle-mounted perception data | |
| CN116383678B (en) | Method for identifying abnormal speed change behavior frequent road sections of operating passenger car | |
| CN116740940A (en) | Severe weather high-impact road section risk prediction and safety management method, device and equipment | |
| JP2018181034A (en) | Travel supporting device, travel supporting method, and data structure therefor | |
| KR102228559B1 (en) | Method and system for seinsing fatigue state of driver | |
| US10618524B2 (en) | Method for determining a reference driving class | |
| CN114944083B (en) | Method for judging distance between running vehicle on expressway and front vehicle | |
| CN114692418B (en) | Centroid side slip angle estimation method and device, intelligent terminal and storage medium | |
| CN114822042B (en) | Information security test management system and method for vehicle-mounted terminal detection |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |