CN110069894A - A kind of objective mapping test method for intelligent automobile traffic coordinating - Google Patents
A kind of objective mapping test method for intelligent automobile traffic coordinating Download PDFInfo
- Publication number
- CN110069894A CN110069894A CN201910384651.8A CN201910384651A CN110069894A CN 110069894 A CN110069894 A CN 110069894A CN 201910384651 A CN201910384651 A CN 201910384651A CN 110069894 A CN110069894 A CN 110069894A
- Authority
- CN
- China
- Prior art keywords
- normalized
- vehicle
- result
- tested
- tested vechicle
- 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.)
- Pending
Links
- 238000013507 mapping Methods 0.000 title claims abstract description 52
- 238000010998 test method Methods 0.000 title claims abstract description 15
- 238000012360 testing method Methods 0.000 claims abstract description 63
- 238000011156 evaluation Methods 0.000 claims abstract description 47
- 238000013210 evaluation model Methods 0.000 claims abstract description 28
- 238000012549 training Methods 0.000 claims abstract description 20
- 238000013528 artificial neural network Methods 0.000 claims abstract description 15
- 238000013461 design Methods 0.000 claims abstract description 12
- 238000004458 analytical method Methods 0.000 claims abstract description 7
- 230000008859 change Effects 0.000 claims description 54
- 230000001133 acceleration Effects 0.000 claims description 51
- 230000003993 interaction Effects 0.000 claims description 25
- 230000000007 visual effect Effects 0.000 claims description 16
- 238000012545 processing Methods 0.000 claims description 15
- 238000010606 normalization Methods 0.000 claims description 13
- 230000008676 import Effects 0.000 claims description 10
- 238000004364 calculation method Methods 0.000 claims description 8
- 210000002569 neuron Anatomy 0.000 claims description 7
- 238000004088 simulation Methods 0.000 claims description 4
- 239000013598 vector Substances 0.000 claims description 4
- 239000011521 glass Substances 0.000 claims description 3
- 230000013011 mating Effects 0.000 claims description 3
- 239000004973 liquid crystal related substance Substances 0.000 claims description 2
- 235000013399 edible fruits Nutrition 0.000 claims 1
- 238000000034 method Methods 0.000 abstract description 20
- 238000002474 experimental method Methods 0.000 abstract description 2
- 230000002452 interceptive effect Effects 0.000 description 8
- 230000008569 process Effects 0.000 description 8
- 238000011160 research Methods 0.000 description 4
- 238000010586 diagram Methods 0.000 description 3
- 241001269238 Data Species 0.000 description 2
- 238000012512 characterization method Methods 0.000 description 2
- 238000012886 linear function Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000012795 verification Methods 0.000 description 2
- 241000239290 Araneae Species 0.000 description 1
- 208000010201 Exanthema Diseases 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000004422 calculation algorithm Methods 0.000 description 1
- 238000013278 delphi method Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 201000005884 exanthem Diseases 0.000 description 1
- 238000013100 final test Methods 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 230000035699 permeability Effects 0.000 description 1
- 238000011158 quantitative evaluation Methods 0.000 description 1
- 206010037844 rash Diseases 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/01—Detecting movement of traffic to be counted or controlled
- G08G1/0104—Measuring and analyzing of parameters relative to traffic conditions
- G08G1/0137—Measuring and analyzing of parameters relative to traffic conditions for specific applications
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- General Health & Medical Sciences (AREA)
- Mathematical Physics (AREA)
- Data Mining & Analysis (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Computational Linguistics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Chemical & Material Sciences (AREA)
- Analytical Chemistry (AREA)
- Computer Hardware Design (AREA)
- Geometry (AREA)
- Traffic Control Systems (AREA)
Abstract
The present invention relates to a kind of objective mapping test method for intelligent automobile traffic coordinating, this method includes determining that traffic coordinating objectively evaluates index system；Easy driver/occupant is built in ring test platform, training sample data and verifying sample data are tested and acquired to design experiment；Training sample data are exported as training input, expert analysis mode as target after inciting somebody to action wherein normalized, obtain objective mapping evaluation model by the way that BP neural network is trained；It is verified using validity of the verifying sample data to the mapping model that training obtains, is finally realized with this objective mapping evaluation model and quantitative assessment is carried out to intelligent automobile traffic coordinating.Compared with existing intelligent assessment technique, uncertain, the high-efficient, testing cost that the present invention has the characteristics that be avoided that evaluation criterion disunity is brought to test result is low.
Description
Technical field
The present invention relates to vehicle testing assessment technique field, more particularly, to a kind of for intelligent automobile traffic coordinating
Objective mapping test method.
Background technique
With the development of intelligent automobile technology, permeability of the intelligent automobile on real road is stepped up, with the mankind
Invisible collaboration and game will be present in driving, in turn results in traffic coordinating problem.Currently, intelligent automobile test evaluation
Research is all directed to driveability of the vehicle under certain external condition, environment itself, the phase without being directed to traffic coordinating
Research is closed, there is an urgent need to a kind of test evaluation methods of objective quantitative to be evaluated.
For the intelligence evaluation of intelligent automobile, main evaluation method has: a kind of method is qualitative evaluation: 1) based on spider
Pessimistic concurrency control；2) it is based on turing test.Another method is quantitative assessment: 1) TOPSIS comprehensive evaluation and gray association evaluation side
Method；2) Field Using Fuzzy Comprehensive Assessment；3) entropy cost function evaluation assessment.Such method there are evaluation criterion disunity, evaluate low efficiency,
The disadvantages of test evaluation is at high cost.
In order to realize quantitatively evaluating, and avoids uncertain to evaluation result bring by evaluation criterion disunity and evaluate
The problem of low efficiency, it is necessary to which one kind is studied for the objective mapping evaluation method of intelligent automobile traffic coordinating.
Summary of the invention
It is an object of the present invention to overcome the above-mentioned drawbacks of the prior art and provide one kind to be directed to intelligent automobile
The objective mapping test method of traffic coordinating.
The purpose of the present invention can be achieved through the following technical solutions:
A kind of objective mapping test method for intelligent automobile traffic coordinating, comprising the following steps:
Step 1: establishing the objective mapping evaluation model of trained traffic coordinating；
Step 2: design interaction is tested and carries out emulation or real steering vectors；
Step 3: collecting sample data are simultaneously normalized；
Step 4: normalized result obtained in step 3 is input in the objective mapping evaluation model in step 1
Obtain traffic coordinating evaluation test result；
Step 5: being joined according to traffic coordinating evaluation test result auxiliary software mating to practical autonomous driving vehicle
Number design.
Further, the step 1 include it is following step by step:
Step 101: traffic coordinating is objective reflects for the index building that objectively evaluates for choosing for intelligent automobile traffic coordinating
Penetrate evaluation model；
Step 102: design interaction is tested and carries out emulation or real steering vectors；
Step 103: collecting test sample data simultaneously carries out data normalized；
Step 104: being inputted the data result selected part sample data of normalized as training, with expert analysis mode
As a result desired output, the training objective mapping evaluation model of traffic coordinating are used as；
Step 105: the remaining sample data in selecting step 104 in the data result of normalized is to verify traffic association
The objective mapping evaluation model performance of tonality.
Further, the index that objectively evaluates in the step 101 includes that global, occupant and other traffic participants drive
Three different visual angles of member are used to characterize the rule layer of traffic coordinating and the multiple of indicator layer objectively evaluate index, it may be assumed that tested
Vehicle lane change time, tested vechicle lane change position, lane change success rate, the decaying of background vehicle speed, two vehicle minimum lateral spacing, two vehicles are lateral
Relative velocity, two vehicle minimum space headways, two vehicle minimum time headways, tested vechicle maximum yaw velocity, tested vechicle are averagely vertical
To acceleration and tested vechicle maximum side acceleration.
Further, the step 102 specifically includes: selecting typical condition as test scene, the test scene packet
It includes ring road mouth to import and overtake other vehicles, designs various forms of interaction tests, and carry out by way of l-G simulation test or real train test
It tests, the scenario building software in the l-G simulation test includes Prescan, VTD, Carla and Unity, the wherein reproduction at visual angle
Equipment includes liquid crystal display and VR glasses, and wherein virtual vehicle controller includes driving simulator and pilot model.
Further, the normalized in the step 103 specifically includes:
A. the tested vechicle lane change time, x is used1It indicates: when the tested lane change time being greater than the normal lane change time, tested vechicle lane change
Time normalization processing result is 1, when the tested lane change time being less than the normal lane change time, at tested vechicle lane change time normalization
Manage the ratio that result is tested vechicle lane change time and normal lane change time；
B. tested vechicle lane change position, uses x2It indicates: the calculation formula of tested vechicle lane change place normalization processing result are as follows:
In formula, X2Indicate tested vechicle lane change place normalization processing result, P1Indicate the start position in remittance section, P2Table
Show the final position for importing section；
C. lane change success rate uses x3It indicates: the calculation formula of lane change success rate normalized result are as follows:
In formula, X3Indicate tested vechicle lane change place normalization processing result；
D. background vehicle speed is decayed: when background vehicle is not braked, then background vehicle speed decaying normalized result is
0, when background vehicle has deceleration behavior, the calculation formula of background vehicle speed decaying normalized result are as follows:
In formula, X4Indicate background vehicle speed decaying normalized as a result,Indicate that background vehicle is flat during interaction
Equal speed, V indicate background vehicle interaction device speed minimum value,Indicate average speed of the background vehicle before interaction；
E. two vehicle minimum lateral spacing, uses x5It indicates: when two vehicle minimum lateral spacing are greater than comfortable spacing X0When, two vehicles are most
Small lateral spacing normalized result is 1, when two vehicle minimum lateral spacing are less than comfortable spacing X0When, between two vehicle minimum laterals
It is x away from normalized result5/X0；
F. the laterally opposed speed of two vehicles, uses x6It indicates: when the laterally opposed speed of two vehicles is greater than comfortable relative velocity, two vehicles
Laterally opposed speed normalized result is 1, when the laterally opposed speed of two vehicles is less than comfortable relative velocity, the lateral phase of two vehicles
It is the ratio of two vehicles laterally opposed speed and comfortable relative velocity to speed normalized result；
G. two vehicle minimum space headway, uses x7It indicates: when two vehicle minimum space headways are greater than comfortable space headway, two vehicles
Minimum space headway normalized result is 1, when two vehicle minimum space headways are less than comfortable space headway, two vehicles most trolley
Head spacing normalized result is the ratio of two vehicle minimum space headways and comfortable space headway；
H. two vehicle minimum time headway, uses x8It indicates: when two vehicle minimum time headways are less than 0, when two vehicle minimum headstocks
It is 1 away from normalized result, when two vehicle minimum time headways are greater than 0, two vehicle minimum time headway normalized results
For x8/ T, T indicate safe bus head when away from；
I. tested vechicle maximum yaw velocity, uses x9It indicates: when tested vechicle maximum yaw velocity is greater than maximum yaw angle
When speed, tested vechicle maximum yaw velocity normalized result is 1, when tested vechicle maximum yaw velocity is less than maximum cross
When pivot angle speed, tested vechicle maximum yaw velocity normalized result is x9/ω0, ω0Indicate maximum yaw velocity；
J. tested vechicle is averaged longitudinal acceleration, uses x10It indicates: when tested vechicle is averaged longitudinal acceleration most greater than human body receiving
When big longitudinal acceleration, the tested vechicle longitudinal acceleration normalized result that is averaged is 1, when the tested vechicle longitudinal acceleration that is averaged is small
When human body bears maximum longitudinal acceleration, tested vechicle is averaged longitudinal acceleration normalized result as the average longitudinal direction of tested vechicle
Acceleration and human body bear the ratio of maximum longitudinal acceleration；
K. tested vechicle maximum side acceleration, uses x11It indicates: being born most when tested vechicle maximum side acceleration is greater than human body
When big side acceleration, tested vechicle maximum side acceleration normalized result is 1, when tested vechicle maximum side acceleration is small
When human body bears maximum side acceleration, tested vechicle maximum side acceleration normalized result is that tested vechicle is maximum lateral
Acceleration and human body bear ratio when maximum side acceleration.
Further, the step 104 specifically includes: taking part sample data, objectively evaluates finger with normalized
Data are marked as training input, using expert's subjective scoring data as desired output, are trained to obtain visitor by neural network
See mapping model.
Further, the calculation formula of the hidden layer neuron number in the neural network are as follows:
In formula, m indicates input layer number, and n indicates output layer neuron number, and α takes the integer between 0~10.
Further, the step 4 specifically includes: writing script tune using MATLAB Neural Network Toolbox or matlab
The objective mapping evaluation model obtained with training, the test sample data after inputting normalized, is emulated to obtain traffic
Harmony objectively evaluates result.
Compared with prior art, the invention has the following advantages that
1, the present invention has filled up the blank of corresponding field test evaluation, and can carry out for intelligent automobile traffic coordinating high
The objective mapping test evaluation of effect, specific provided with set, there are many objective mappings of the traffic coordinating of index parameter to evaluate mould
Type specifically includes global, three different visual angles of occupant and other traffic participants driver for characterizing traffic coordinating
Rule layer and the multiple of indicator layer objectively evaluate index, it may be assumed that the tested vechicle lane change time, tested vechicle lane change position, lane change success
Rate, the decaying of background vehicle speed, two vehicle minimum lateral spacing, the laterally opposed speed of two vehicles, two vehicle minimum space headways, two vehicles are minimum
Time headway, tested vechicle maximum yaw velocity, tested vechicle are averaged longitudinal acceleration and tested vechicle maximum side acceleration, therefore
It is stronger for the specific aim of intelligent automobile, more help the design for being suitable for the practical auxiliary software of intelligent automobile；
2, the present invention can effectively avoid evaluation criterion disunity uncertain to evaluation result bring, specifically gather vapour
The highest achievement data of the degree of correlation actually generated in vehicle traveling, comprising: global, occupant and other traffic participants driver three
A different visual angle is used to characterize the rule layer of traffic coordinating and the multiple of indicator layer objectively evaluate index, it may be assumed that tested vechicle becomes
Road time, tested vechicle lane change position, lane change success rate, the decaying of background vehicle speed, two vehicle minimum lateral spacing, two vehicles are laterally opposed
The average longitudinal direction of speed, two vehicle minimum space headways, two vehicle minimum time headways, tested vechicle maximum yaw velocity, tested vechicle adds
Speed and tested vechicle maximum side acceleration, final testing result uniformity is high, and traditional test evaluation method is to comment each time
Valence requires to ask expert estimation, so the different conditions of the different and same experts of expert can all bring not evaluation result
Together.So evaluation criterion disunity, the present invention because being to remove evaluation traffic coordinating using the evaluation model that sample training obtains,
Evaluation model will not become, so evaluation criterion is unified, eliminate the uncertainty of evaluation result；
3, the present invention can evaluate the abswolute level of intelligent automobile traffic association property, and improve evaluation efficiency, reduce
Test evaluation cost, the present invention is provided with set, and there are many objective mapping evaluation model of traffic coordinating of index parameter, tradition
In test evaluation method: test evaluation each time requires multiple experts and participates in, and allows expert to give a mark evaluation index one by one, institute
It is at high cost with test evaluation, evaluate low efficiency；The present invention carries out automatic quantitative evaluation to traffic coordinating using test data, keeps away
Exempt from traditional test evaluation and needed the participation of multidigit expert, therefore reduced overall testing cost, improves evaluation efficiency；
4, of the present invention can cover for the objective mapping test evaluation method of traffic coordinating high level is driven automatically
The test evaluation for sailing automobile, due to being provided with set in the present invention, there are many objective mapping evaluations of the traffic coordinating of index parameter
Model, high-level autonomous driving vehicle refer to: the autonomous driving vehicle of L3, L4 and L5 rank in automatic Pilot SAE classification,
(L3 or more is high-level autonomous driving vehicle), thus can satisfy the above rank of SAE L3 autonomous driving vehicle it is comprehensive
Close performance evaluation.
Detailed description of the invention
Fig. 1 is the objective mapping test flow chart of intelligent automobile traffic coordinating in the present invention；
Fig. 2 is that tested vechicle imports position view in the embodiment of the present invention；
Fig. 3 is a kind of two lane highway ring road mouths remittance scene in the embodiment of the present invention；
Fig. 4 is that multi-angle of view reproduces schematic diagram in the embodiment of the present invention；
Fig. 5 is BP neural network topological structure schematic diagram in the embodiment of the present invention；
Fig. 6 is BP neural network training fitting result schematic diagram in the embodiment of the present invention.
Specific embodiment
Following will be combined with the drawings in the embodiments of the present invention, and technical solution in the embodiment of the present invention carries out clear, complete
Site preparation description, it is clear that described embodiment is a part of the embodiments of the present invention, rather than whole embodiments.Based on this hair
Embodiment in bright, those of ordinary skill in the art's every other reality obtained without making creative work
Example is applied, all should belong to the scope of protection of the invention.
Embodiment
As shown in Figure 1, a kind of objective mapping test method for intelligent automobile traffic coordinating of the invention, main
Implementation steps are:
S1 establishes objective mapping evaluation model and verifies its validity for traffic coordinating；
S2 carries out quantitative assessment to traffic coordinating using the effective objective mapping evaluation model established；
Finally, parameter is carried out according to traffic coordinating evaluation test result auxiliary software mating to practical autonomous driving vehicle
Design.
Wherein establishing the objective mapping evaluation model of traffic coordinating, detailed process is as follows:
S1.1 selection characterization intelligent automobile traffic coordinating objectively evaluates index.
For AEV traffic coordinating sex expression, the degree of involvement from AEV in human environment, AEV are to microcosmic traffic system respectively
It cuts in terms of influence three of the influence, AEV of system to human comfort, research equipment intelligence, high efficiency and comfort three
Interpretational criteria.According to figure, Sentos thinks, from macro, microcosmic angle, i.e., global (visual angle I), from vehicle occupant (visual angle II) and other traffic
Three different visual angles of participant driver (visual angle III), the method combined using analytic approach and Delphi method consider purpose
Property, comparativity and operability, determine that 11 objectively evaluate index (as shown in table 1): tested vechicle lane change time, tested vechicle lane change
Position, lane change success rate, the decaying of background vehicle speed, two vehicle minimum lateral spacing, the laterally opposed speed of two vehicles, two vehicle minimum headstocks
Spacing, two vehicle minimum time headways, tested vechicle maximum yaw velocity, tested vechicle are averaged longitudinal acceleration and tested vechicle maximum side
To acceleration；
Table 1: intelligent automobile traffic coordinating objective examination's index name information table in the present invention
| Index name | Dimension | Index name | Dimension | Index name | Dimension |
| The tested vechicle lane change time | s | Two vehicle minimum lateral spacing | m | Tested vechicle maximum yaw velocity | rad/s |
| Background vehicle lane change position | m | The laterally opposed speed of two vehicles | km/h | Tested vechicle is averaged longitudinal acceleration | m/s2 |
| Lane change success rate | % | Two vehicle minimum space headways | m | Tested vechicle maximum side acceleration | m/s2 |
| The decaying of background vehicle speed | km/h | Two vehicle minimum time headways | s | Expert opinion | Grade |
S1.2 design experiment simultaneously carries out emulation testing.
The present embodiment is described so that expressway ring road mouth imports scene as an example.In order to obtain covering interaction as much as possible
The sample data of situation is provided with 30 groups of different types of interbehaviors, integration factor are as follows: background vehicle is with the constant vehicle of 60km/h
Speed traveling, the driving style of 4 groups of tests be it is conservative, understand judge whether oneself should give way according to the driving status of tested vechicle；
Remaining test driving style be all it is radical, ignore the driving status of tested vechicle, keep constant speed traveling.Tested vechicle with
The constant speed of 60km/h, which is driven into, imports section.During remittance, planning aspect shows as conservative or radical, decision-making level table
It is now resolute, rash or hesitation, quiet run, anxious acceleration sudden turn of events road or the anxious slow lane change of deceleration is shown as in terms of tracing control,
Importing position has the remittance of A point, the remittance of B point, the remittance of C point and D point to import (as shown in Figure 2).
Wherein, since tested vechicle is autonomous driving vehicle or mankind's driving mapping evaluation objective to traffic coordinating
The research of method be do not have it is influential, therefore the present embodiment select the mankind driving as tested vechicle.
The problems such as in view of efficiency, cost, safety, the present embodiment selection are carried out in a simulated environment, develop driving
Member/occupant's assemblage on-orbit test platform.
The present invention selects Prescan simulation software, to import surveying on the spot for ring road to G15 Shen Hai high speed Shanghai certain entrance of section
Based on measured data, the two lane highway ring road mouths for constructing standard import scene (as shown in Figure 3).Scene is divided into three
Section: 1) accelerating sections of main road 200m and ring road branch 80m can speed up two interactive vehicles to test pre-set velocity and keep steady
Determine to interaction locations.2) the remittance section of 200m is interacted for meeting two vehicles in different location.3) the surplus section of 200m is come
Guarantee the integrality of primary interaction test.
Using 1920 × 1080 resolution ratio, 23.8 inches of Dell's displays reproduce the multi-angle of view visual field (as shown in Figure 4), include
Microcosmic traffic participant visual angle, microcosmic occupant visual angle and macroscopical visual angle, are in addition added to speed real-time display device, enable driver
It is enough to grasp this vehicle speed information in real time.
Using the driving simulator comprising steering wheel, accelerator, brake, clutch pedal and gear lever, pass through USB interface
It is connect with computer.The motion control of test carriage and background vehicle is realized by driving simulator.
Two drivers pass through the driver/occupant's assemblage on-orbit test platform built, different according to designed 30 groups
The interaction test of form, completes interactive testing task.
Expert is horizontal according to the experience and knowledge of oneself, carries out subjectivity to the interaction scenario of different form interaction test and comments
Valence obtains the traffic coordinating expert estimation result of each test sample.
By Prescan and Simulink associative simulation, 11 for obtaining characterization traffic coordinating objectively evaluate index number
According to.
S1.3 collecting test sample data simultaneously carries out data normalized.
By in Prescan Scene Editor add ideal transducer come measure two vehicle relative distances in interactive process,
The data such as relative velocity, lane change position, time；The state parameter of two vehicles itself is by the vehicle dynamic model in Simulink
Directly export.
The 11 objective indicator data collected are normalized, data normalization is that data contract in proportion
It puts, is allowed to fall into a small specific sections, so that achievement data is can participate in evaluation and calculate and facilitate processing.Treatment process is such as
Under:
A. the tested vechicle lane change time.Use x1It indicates, rule of thumb taking the normal lane change time is 2s.
1) work as x1When > 2s, normalized result is X1=1；
2) work as x1When < 2s, normalized result is X1=x1/2。
B. tested vechicle lane change position.Use x2It indicates.Data normalization processing result is as follows:
Wherein, wherein P1For the start position for importing section, P2For the final position for importing section；
C. lane change success rate.Use x3It indicates, normalized result is as follows:
D. background vehicle speed is decayed.Refer to during the interaction of two vehicles, the degree that background vehicle speed reduces.
1) when background vehicle is not braked, then X4=0；
2) when background vehicle has deceleration behavior, processing result is as follows:
Wherein, V is minimum value of the background vehicle during interaction；It is average speed of the background vehicle during interaction；It is average speed of the background vehicle before interaction.
E. two vehicle minimum lateral spacing.Use x5It indicates, when two vehicles rule of thumb being taken to be respectively at two lane center line positions
Two vehicle lateral spacings be comfortable spacing X0, then normalized is as follows:
1) work as x5>X0When, normalized result X5=1；
2) work as x5<X0When, normalized result
F. the laterally opposed speed of two vehicles.Use x6It indicates, is rule of thumb analyzed with data, occupant is universal receptible comfortable
Relative velocity is 20km/h.Normalized is done as standard.
1) work as x6> 20km/h, normalized result are X6=1；
2) work as x6< 20km/h, normalized result are
G. two vehicle minimum space headway.Use x7It indicates, rule of thumb, 2 vehicle body lengths is set as between the comfortable headstock of standard
Away from Δ, normalized is done as standard.
1) work as x7When > Δ, normalized result X7=1；
2) work as x7When < Δ, normalized result
H. two vehicle minimum time headway.Use x8Indicate, rule of thumb using T=2s as when safe bus head away from as mark
Standard carrys out normalized data.
1) work as x8When < 0, illustrate that two vehicles will not bump against, then X8=1；
2) work as x8When > 0, normalized result
I. tested vechicle maximum yaw velocity.Use x9It indicates, rule of thumb learns, the maximum yaw angle speed under limiting condition
Degree is ω0=1rad/s does normalized as standard.
1) work as x9>ω0When, then normalize result X9=1；
2) work as x9<ω0When, then normalize result
J. tested vechicle is averaged longitudinal acceleration.Use x10It indicates, the maximum longitudinal acceleration that human body maximum can be born is one
G, i.e. 10m/s2.Data normalization processing is carried out as standard.
1) work as x10>10m/s2When, then normalize result X10=1；
2) work as x10<10m/s2When, then normalize result
K. tested vechicle maximum side acceleration.Use x11It indicates, the maximum side acceleration that human body maximum can be born is one
G, i.e. 10m/s2.Data normalization processing is carried out as standard.
1) work as x11>10m/s2When, then normalize result X11=1；
2) work as x11<10m/s2When, then normalize result
S1.4 randomly selects 22 groups of sample datas in step S1.3 after normalized and expert analysis mode result to train
Objective mapping evaluation model.
The present invention selects BP neural network (as shown in Figure 5) to establish objective indicator measured value and expert analysis mode result to train
Numerical relationship model.
Ordinary priority considers 3 layers of BP neural network (i.e. 1 hidden layer), is obtained by increasing the neuron number of hidden layer
Obtain lower training error.The number of requirement, the input and output of hidden layer neuron number and problem has relationship, probable ranges
Can there is following formula to obtain:
Wherein, m is input layer number, and n is output layer neuron number, and α is the integer between 0~10.
The common transmission function of BP neural network has S type nonlinear function, linear function and threshold value type function.Every layer of biography
Delivery function can be not quite similar.It commonly enters layer and S type nonlinear function Logsig, Tansig is respectively adopted as biography in hidden layer
Delivery function, output layer use linear function Purelin, come the numerical value scaling of any range before keeping, are convenient for and sample object
Output valve is made comparisons.
The training function of BP neural network selects the speed of service fast, and convergence rate quickly, and can be with seldom the number of iterations
Reach the target error of requirement, training precision also very high Trainlm algorithm.
It is inputted using 22 groups in the sample data after the normalized randomly selected as training, expert analysis mode result is
Desired output is trained to obtain objective mapping using the BP neural network tool box (or writing matlab script) in Matlab
Evaluation model (as shown in Figure 6).
Other 8 groups of sample datas in S1.5 selecting step S1.3 verify the validity of objective mapping evaluation model.
Using other 8 groups in the sample data after normalized as verifying sample, it is objective to obtain with training
Mapping evaluation model is calculated the model evaluation of 8 groups of samples as a result, contrast model evaluation result and expert analysis mode result are tested
The validity for the objective mapping evaluation model that card training obtains.Objective mapping evaluation model verification result comparison is as shown in table 2.
Table 2: objective mapping evaluation model verification result comparing result in the embodiment of the present invention
The effective objective mapping evaluation model established with step S1, tests intelligent automobile traffic coordinating
Evaluation.Detailed process is as follows for it:
S2.1 design interaction is tested and carries out emulation testing.The present embodiment devise two groups interaction test: 1, tested vechicle with
The constant speed of 65km/h, which is driven into, imports section.During remittance, it is radical that planning aspect is shown as, and shows as in terms of tracing control
It is anxious to accelerate sudden turn of events road；Background vehicle with the constant speed drive of 65km/h, in interactive process can according to the traveling behavior of other side and
Adjust the driving behavior of oneself.2, tested vechicle is driven into the constant speed of 65km/h and imports section.In interactive process, traveling row
It is depending on the circumstances；Background vehicle, can be according to the traveling row of other side in interactive process with the constant speed drive of 65km/h
For and adjust oneself driving behavior.
Two drivers pass through the driver/occupant's assemblage on-orbit test platform built, according to designed 2 groups not similar shapes
The interaction test of formula, completes interactive testing task.
S2.2 measured in interactive process by adding ideal transducer in Prescan Scene Editor two vehicles it is opposite away from
From data such as, relative velocity, lane change position, times；The state parameter of two vehicles itself is by the dynamics of vehicle mould in Simulink
It is directly exported in type.The 11 objective indicator data collected are carried out such as the normalized in step S1.
The effective objective mapping evaluation model that S2.3 calls training to obtain in MATLAB Neural Network Toolbox, input
2 groups of test sample data after normalized, are emulated, and available traffic coordinating objectively evaluates result (such as table 3
It is shown).
Table 3: the objective mapping model evaluation result of intelligent automobile traffic coordinating in the embodiment of the present invention
Embodiment 2
Scenario building software uses Prescan in embodiment 1, and visual angle reproduction equipment uses display, virtual vehicle control
Method uses driving simulator, and in addition to this VTD, Carla and Unity also can be used in scenario building software, and wherein visual angle reproduces
VR glasses also can be used in equipment, and pilot model also can be used in virtual vehicle control method.
The above description is merely a specific embodiment, but scope of protection of the present invention is not limited thereto, any
Those familiar with the art in the technical scope disclosed by the present invention, can readily occur in various equivalent modifications or replace
It changes, these modifications or substitutions should be covered by the protection scope of the present invention.Therefore, protection scope of the present invention should be with right
It is required that protection scope subject to.
Claims (8)
1. a kind of objective mapping test method for intelligent automobile traffic coordinating, which comprises the following steps:
Step 1: establishing the objective mapping evaluation model of trained traffic coordinating；
Step 2: design interaction is tested and carries out emulation or real steering vectors；
Step 3: collecting sample data are simultaneously normalized；
Step 4: normalized result obtained in step 3 being input in the objective mapping evaluation model in step 1 and is obtained
Traffic coordinating evaluation test result；
Step 5: parameter being carried out according to traffic coordinating evaluation test result auxiliary software mating to practical autonomous driving vehicle and is set
Meter.
2. a kind of objective mapping test method for intelligent automobile traffic coordinating according to claim 1, feature
Be, the step 1 include it is following step by step:
Step 101: the objective mapping of index building traffic coordinating that objectively evaluates chosen for intelligent automobile traffic coordinating is commented
Valence model；
Step 102: design interaction is tested and carries out emulation or real steering vectors；
Step 103: collecting test sample data simultaneously carries out data normalized；
Step 104: being inputted the data result selected part sample data of normalized as training, with expert analysis mode result
As desired output, the training objective mapping evaluation model of traffic coordinating；
Step 105: the remaining sample data in selecting step 104 in the data result of normalized is to verify traffic coordinating
Objective mapping evaluation model performance.
3. a kind of objective mapping test method for intelligent automobile traffic coordinating according to claim 2, feature
It is, the index that objectively evaluates in the step 101 includes that global, occupant and other traffic participants driver three are different
Visual angle is used to characterize the rule layer of traffic coordinating and the multiple of indicator layer objectively evaluate index, it may be assumed that tested vechicle lane change time, quilt
Measuring car lane change position, lane change success rate, the decaying of background vehicle speed, two vehicle minimum lateral spacing, the laterally opposed speed of two vehicles, two vehicles
Minimum space headway, two vehicle minimum time headways, tested vechicle maximum yaw velocity, tested vechicle are averaged longitudinal acceleration and tested
Vehicle maximum side acceleration.
4. a kind of objective mapping test method for intelligent automobile traffic coordinating according to claim 2, feature
It is, the step 102 specifically includes: selects typical condition as test scene, the test scene includes that ring road mouth imports
With overtake other vehicles, design various forms of interaction tests, and tested by way of l-G simulation test or real train test, the emulation
Scenario building software in test includes Prescan, VTD, Carla and Unity, and wherein the reproduction equipment at visual angle includes liquid crystal
Show device and VR glasses, wherein virtual vehicle controller includes driving simulator and pilot model.
5. a kind of objective mapping test method for intelligent automobile traffic coordinating according to claim 2, feature
It is, the normalized in the step 103 specifically includes:
A. the tested vechicle lane change time, x is used1It indicates: when the tested lane change time being greater than the normal lane change time, the tested vechicle lane change time
Normalized result is 1, when the tested lane change time being less than the normal lane change time, tested vechicle lane change time normalization processing knot
Fruit is the ratio of tested vechicle lane change time and normal lane change time；
B. tested vechicle lane change position, uses x2It indicates: the calculation formula of tested vechicle lane change place normalization processing result are as follows:
In formula, X2Indicate tested vechicle lane change place normalization processing result, P1Indicate the start position in remittance section, P2It indicates to converge
Enter the final position in section；
C. lane change success rate uses x3It indicates: the calculation formula of lane change success rate normalized result are as follows:
In formula, X3Indicate tested vechicle lane change place normalization processing result；
D. background vehicle speed is decayed: when background vehicle is not braked, then background vehicle speed decaying normalized result is 0, when
When background vehicle has deceleration behavior, the calculation formula of background vehicle speed decaying normalized result are as follows:
In formula, X4Indicate background vehicle speed decaying normalized as a result,Indicate average speed of the background vehicle during interaction
Degree, V indicate background vehicle interaction device speed minimum value,Indicate average speed of the background vehicle before interaction；
E. two vehicle minimum lateral spacing, uses x5It indicates: when two vehicle minimum lateral spacing are greater than comfortable spacing X0When, two vehicle minimum sides
It is 1 to spacing normalized result, when two vehicle minimum lateral spacing are less than comfortable spacing X0When, two vehicle minimum lateral spacing are returned
One changes processing result as x5/X0；
F. the laterally opposed speed of two vehicles, uses x6Indicate: when the laterally opposed speed of two vehicles is greater than comfortable relative velocity, two vehicles are lateral
Relative velocity normalized result is 1, when the laterally opposed speed of two vehicles is less than comfortable relative velocity, the laterally opposed speed of two vehicles
Spend the ratio that normalized result is two vehicles laterally opposed speed and comfortable relative velocity；
G. two vehicle minimum space headway, uses x7Indicate: when two vehicle minimum space headways are greater than comfortable space headway, two vehicles are minimum
Space headway normalized result is 1, when two vehicle minimum space headways are less than comfortable space headway, between two vehicle minimum headstocks
It is the ratio of two vehicle minimum space headways and comfortable space headway away from normalized result；
H. two vehicle minimum time headway, uses x8It indicates: when two vehicle minimum time headways are less than 0, two vehicle minimum time headway normalizings
Changing processing result is 1, and when two vehicle minimum time headways are greater than 0, two vehicle minimum time headway normalized results are x8/ T,
T indicate safe bus head when away from；
I. tested vechicle maximum yaw velocity, uses x9It indicates: when tested vechicle maximum yaw velocity is greater than maximum yaw velocity
When, tested vechicle maximum yaw velocity normalized result is 1, when tested vechicle maximum yaw velocity is less than maximum yaw angle
When speed, tested vechicle maximum yaw velocity normalized result is x9/ω0, ω0Indicate maximum yaw velocity；
J. tested vechicle is averaged longitudinal acceleration, uses x10Indicate: when tested vechicle is averaged, longitudinal acceleration is maximum greater than human body receiving to be indulged
When to acceleration, the tested vechicle longitudinal acceleration normalized result that is averaged is 1, when the tested vechicle longitudinal acceleration that be averaged is less than people
When body receiving maximum longitudinal acceleration, tested vechicle is averaged longitudinal acceleration normalized result as the average longitudinal acceleration of tested vechicle
Degree bears the ratio of maximum longitudinal acceleration with human body；
K. tested vechicle maximum side acceleration, uses x11It indicates: bearing maximum side when tested vechicle maximum side acceleration is greater than human body
When to acceleration, tested vechicle maximum side acceleration normalized result is 1, when tested vechicle maximum side acceleration is less than people
When body bears maximum side acceleration, tested vechicle maximum side acceleration normalized result, which is that tested vechicle is maximum, laterally to be accelerated
Degree bears ratio when maximum side acceleration with human body.
6. a kind of objective mapping test method for intelligent automobile traffic coordinating according to claim 2, feature
It is, the step 104 specifically includes: takes part sample data, achievement data is objectively evaluated as instruction using normalized
Practice input, using expert's subjective scoring data as desired output, is trained to obtain objective mapping model by neural network.
7. a kind of objective mapping test method for intelligent automobile traffic coordinating according to claim 6, feature
It is, the calculation formula of the hidden layer neuron number in the neural network are as follows:
In formula, m indicates input layer number, and n indicates output layer neuron number, and α takes the integer between 0~10.
8. a kind of objective mapping test method for intelligent automobile traffic coordinating according to claim 1, feature
It is, the step 4 specifically includes: writes script using MATLAB Neural Network Toolbox or matlab and training is called to obtain
Objective mapping evaluation model, the test sample data after inputting normalized are emulated to obtain that traffic coordinating is objective to be commented
Valence result.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910384651.8A CN110069894A (en) | 2019-05-09 | 2019-05-09 | A kind of objective mapping test method for intelligent automobile traffic coordinating |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910384651.8A CN110069894A (en) | 2019-05-09 | 2019-05-09 | A kind of objective mapping test method for intelligent automobile traffic coordinating |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110069894A true CN110069894A (en) | 2019-07-30 |
Family
ID=67370301
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910384651.8A Pending CN110069894A (en) | 2019-05-09 | 2019-05-09 | A kind of objective mapping test method for intelligent automobile traffic coordinating |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110069894A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112744223A (en) * | 2021-01-18 | 2021-05-04 | 北京智能车联产业创新中心有限公司 | Method and system for evaluating intersection performance of automatic driving vehicle |
| CN112977477A (en) * | 2021-02-26 | 2021-06-18 | 江苏大学 | Hybrid vehicle-vehicle cooperative convergence system and method based on neural network |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101118620A (en) * | 2007-09-18 | 2008-02-06 | 吉林大学 | Vehicle gear shifting quality evaluation method based on nerval net |
| CN103234749A (en) * | 2013-04-09 | 2013-08-07 | 杭州电子科技大学 | Automobile clutch control comfort evaluation method based on artificial neural network |
| CN106530717A (en) * | 2016-12-26 | 2017-03-22 | 长安大学 | Construction road section risk evaluating method based on cloud model |
| CN106991811A (en) * | 2017-05-03 | 2017-07-28 | 同济大学 | Expressway exit ring road upstream trackside road information Optimization Design based on drive simulation experiment porch |
| CN109572695A (en) * | 2018-11-08 | 2019-04-05 | 湖南汽车工程职业学院 | A kind of autonomous driving vehicle Car following control method and system |
| CN109572706A (en) * | 2018-12-12 | 2019-04-05 | 西北工业大学 | A kind of driving safety evaluation method and device |
-
2019
- 2019-05-09 CN CN201910384651.8A patent/CN110069894A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101118620A (en) * | 2007-09-18 | 2008-02-06 | 吉林大学 | Vehicle gear shifting quality evaluation method based on nerval net |
| CN103234749A (en) * | 2013-04-09 | 2013-08-07 | 杭州电子科技大学 | Automobile clutch control comfort evaluation method based on artificial neural network |
| CN106530717A (en) * | 2016-12-26 | 2017-03-22 | 长安大学 | Construction road section risk evaluating method based on cloud model |
| CN106991811A (en) * | 2017-05-03 | 2017-07-28 | 同济大学 | Expressway exit ring road upstream trackside road information Optimization Design based on drive simulation experiment porch |
| CN109572695A (en) * | 2018-11-08 | 2019-04-05 | 湖南汽车工程职业学院 | A kind of autonomous driving vehicle Car following control method and system |
| CN109572706A (en) * | 2018-12-12 | 2019-04-05 | 西北工业大学 | A kind of driving safety evaluation method and device |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112744223A (en) * | 2021-01-18 | 2021-05-04 | 北京智能车联产业创新中心有限公司 | Method and system for evaluating intersection performance of automatic driving vehicle |
| CN112744223B (en) * | 2021-01-18 | 2022-04-15 | 北京智能车联产业创新中心有限公司 | Method and system for evaluating intersection performance of automatic driving vehicle |
| CN112977477A (en) * | 2021-02-26 | 2021-06-18 | 江苏大学 | Hybrid vehicle-vehicle cooperative convergence system and method based on neural network |
| CN112977477B (en) * | 2021-02-26 | 2022-03-22 | 江苏大学 | Hybrid vehicle-vehicle cooperative convergence system and method based on neural network |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| JP6898101B2 (en) | Systems and methods for analyzing vehicle energy efficiency | |
| CN106874597B (en) | highway overtaking behavior decision method applied to automatic driving vehicle | |
| JP6726106B2 (en) | Vehicle motion behavior determination and/or optimization system | |
| CN112896170B (en) | Automatic driving transverse control method under vehicle-road cooperative environment | |
| CN110160804A (en) | A kind of test method of automatic driving vehicle, apparatus and system | |
| JP7053147B2 (en) | Systems and methods for analyzing the energy efficiency of automobiles, especially automobile equipment | |
| CN113287073B (en) | Automatic driving automobile simulator using network platform | |
| CN109583776A (en) | A kind of vehicle body-sensing evaluating method, device, electronic equipment, medium and vehicle | |
| CN110046833A (en) | A kind of traffic congestion auxiliary system virtual test system | |
| CN109446662A (en) | Generation method and device, the computer equipment and storage medium of vehicle simulation track | |
| Bareket et al. | Methodology for assessing adaptive cruise control behavior | |
| Niu et al. | Eco-driving versus green wave speed guidance for signalized highway traffic: a multi-vehicle driving simulator study | |
| CN110069894A (en) | A kind of objective mapping test method for intelligent automobile traffic coordinating | |
| CN109572706A (en) | A kind of driving safety evaluation method and device | |
| CN109387374A (en) | A kind of lane holding level evaluation method | |
| CN113918615A (en) | Simulation-based driving experience data mining model construction method and system | |
| CN105468888B (en) | A kind of appraisal procedure and device of motor racing control performance | |
| CN110264741A (en) | Road conditions detection method, device, equipment and medium based on motion sensor | |
| CN108447306A (en) | Real time position shares the analogy method of intelligent collision warning between No-shell culture conflict vehicle | |
| CN106446335A (en) | Road alignment quality assessment method in three-dimensional space | |
| Wang et al. | A method to automatic measuring riding comfort of autonomous vehicles: Based on passenger subjective rating and vehicle parameters | |
| Ma | Effects of vehicles with different degrees of automation on traffic flow in urban areas | |
| CN115718437A (en) | Simulation method and device for networking automatic driving | |
| Cacciabue et al. | Unified Driver Model simulation and its application to the automotive, rail and maritime domains | |
| CN114969092A (en) | Driving simulation system based on unreal engine and simulation experiment method |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190730 |
|
| RJ01 | Rejection of invention patent application after publication |