CN110414803A - The assessment method and device of automated driving system level of intelligence under different net connection degree - Google Patents

The assessment method and device of automated driving system level of intelligence under different net connection degree Download PDF

Info

Publication number
CN110414803A
CN110414803A CN201910611202.2A CN201910611202A CN110414803A CN 110414803 A CN110414803 A CN 110414803A CN 201910611202 A CN201910611202 A CN 201910611202A CN 110414803 A CN110414803 A CN 110414803A
Authority
CN
China
Prior art keywords
intelligence
tested
road
tested vechicle
net connection
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
Application number
CN201910611202.2A
Other languages
Chinese (zh)
Other versions
CN110414803B (en
Inventor
王建强
刘艺璁
郑讯佳
许庆
黄荷叶
李克强
崔明阳
林学武
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Tsinghua University
Original Assignee
Tsinghua University
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Tsinghua University filed Critical Tsinghua University
Priority to CN201910611202.2A priority Critical patent/CN110414803B/en
Publication of CN110414803A publication Critical patent/CN110414803A/en
Application granted granted Critical
Publication of CN110414803B publication Critical patent/CN110414803B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G06Q10/063—Operations research, analysis or management
    • G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
    • G06Q10/06393—Score-carding, benchmarking or key performance indicator [KPI] analysis

Landscapes

  • Business, Economics & Management (AREA)
  • Human Resources & Organizations (AREA)
  • Engineering & Computer Science (AREA)
  • Strategic Management (AREA)
  • Development Economics (AREA)
  • Economics (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Educational Administration (AREA)
  • Operations Research (AREA)
  • Marketing (AREA)
  • Game Theory and Decision Science (AREA)
  • Quality & Reliability (AREA)
  • Tourism & Hospitality (AREA)
  • Physics & Mathematics (AREA)
  • General Business, Economics & Management (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Traffic Control Systems (AREA)

Abstract

The invention discloses a kind of assessment methods of automated driving system level of intelligence under different net connection degree, this method comprises: S1, chooses the evaluation index of automated driving system level of intelligence；S2 obtains the quantitatively evaluating foundation of automated driving system level of intelligence according to the numerical value difference between the practical function amount and theoretical least action in test traffic process；S3 determines the variation range of quantitatively evaluating foundation, marks off the assessment interval that at least two evaluations are tested the level of intelligence grade of automated driving system, the corresponding level of intelligence grade of each assessment interval；S4 obtains multiple groups quantitatively evaluating of the tested automated driving system under different net connection degree according to data；S5, it is for statistical analysis according to data to quantitatively evaluating, it is evaluated according to level of intelligence of each statistic analysis result to tested automated driving system.The present invention can be three-dimensional and truly test and assess to level of intelligence of the automated driving system under different net connection degree.

Description

The assessment method and device of automated driving system level of intelligence under different net connection degree
Technical field
The present invention relates to a kind of intelligent network connection automobile, autonomous driving vehicle evaluation technology field more particularly to a kind of differences The assessment method and device of automated driving system level of intelligence under net connection degree.
Background technique
Intelligent and net connectionization is the inexorable trend of autonomous driving vehicle development, during netting the long-run development of connectionization, Net connection vehicle and non-net connection vehicle will be existed simultaneously in traffic.Different net connection degree are to influence autonomous driving vehicle in mixed traffic One key factor of level of intelligence, the assessment for carrying out level of intelligence under different net connection degree to automated driving system can take off Show the affecting laws of behind.In recent years, automatic Pilot intelligence test field technology continues to develop, (such as compared to conventional test methodologies Software Simulation Test, hardware-in―the-loop test, test scrnario testing, real vehicle road test), have stronger automation, can weigh The abilities such as multiple, validity, provide active platform for assessment of the automated driving system under different net connection degree.
The prior art does not consider that this key factor of difference net connection degree present in mixed traffic, is regarded as mostly Net connection degree is fixed as to 0 evaluation technology, the intelligent water of automated driving system under the different net connection degree that can not be used to test and assess It is flat.Even if there are still many problems for existing evaluation technology from the perspective of reference:
1. the prior art chooses the evaluation index of automated driving system level of intelligence not reasonable.U.S. DARPA (beauty National defence high-level plan research office, state) tissue three unmanned challenge matches with complete all gainers time-consuming length make For evaluation index, but single index is difficult to reflect the level of intelligence of automated driving system.Level of intelligence is layered and in each layer Grade is lower to carry out overall merit by classification of task index for selection, but this kind of overall target and indirectly embodies level of intelligence (such as peace Entirely, efficiently).
2. the prior art is doped with subjective factor in assessment, not objective enough.Mechanically press the subjective determination of task category Complexity, corresponding level of intelligence grade, but complexity difference of every generic task under different scenes is huge.Choosing every evaluation There is also subjective factors in the corresponding weight of index.
3. the prior art can carry out quantization assessment to driving trace, but its selected reference locus is based on static ideal geometry rail Mark and driver's driving trace, the difference by quantifying track to be measured and reference locus correspond to level of intelligence.But its selected reference Track is not suitable for dynamic traffic environment, can not embody safety objectively and efficiently.
4. the resulting comprehensive evaluation index of the prior art and method are inadequate in dealing with the randomness of net connection degree in testing.
Therefore, to solve the above problems, it is necessary to develop automated driving system level of intelligence under a kind of different net connection degree Assessment method and device.
Summary of the invention
The purpose of the present invention is to provide a kind of assessment methods of automated driving system level of intelligence under different net connection degree And device, can be three-dimensional and truly test and assess to level of intelligence of the automated driving system under different net connection degree.
To achieve the above object, the present invention provides a kind of assessment of automated driving system level of intelligence under different net connection degree Method, this method comprises the following steps:
S1 is based on traffic safety field theory, chooses evaluation index of the actuating quantity as automated driving system level of intelligence；
S2 is obtained according to the numerical value difference between the practical function amount and theoretical least action in test traffic process The quantitatively evaluating foundation of automated driving system level of intelligence, the quantitatively evaluating according to can with the practical function amount and it is theoretical most The variation of the numerical value difference of small actuating quantity and be monotonically changed；
S3 determines the variation model of the quantitatively evaluating foundation according to the acquisition modes of the quantitatively evaluating foundation in S2 It encloses, and marks off the assessment interval that at least two evaluations are tested the level of intelligence grade of automated driving system, each described survey Comment the corresponding level of intelligence grade in section；
S4 obtains the tested automated driving system and joins in different nets in the automatic Pilot intelligence test scene of setting Multiple groups quantitatively evaluating under degree is according to data；
S5, the quantitatively evaluating described in each group of S4 acquisition is for statistical analysis according to data, and according to each statistical analysis As a result the assessment interval in affiliated S3 evaluates the level of intelligence of the tested automated driving system.
Further, the quantitatively evaluating in S4 is specifically included according to the acquisition methods of data:
According to the driving parameters of the tested vechicle obtained in the automatic Pilot intelligence test scene, remaining road occupation The driving parameters and road environment test data of person obtain the practical function amount S in the actual measurement path of the tested vechicle traveling；
According to the reference path planned in advance in the automatic Pilot intelligence test scene remaining described road user And road environment test data, obtain the theoretical least action S* under the automatic Pilot intelligence test scene；
According under the automatic Pilot intelligence test scene the practical function amount and the theoretical least action it Between numerical value difference obtain the quantitatively evaluating according to data in conjunction with the acquisition modes of the quantitatively evaluating foundation in S2.
Further, the practical function amount S in S41 is obtained by following formula (1) to formula (7):
Gi=mig (4)
Fai=Eai·Mi·Pi (5)
For formula (1) into formula (7), A is the initial position of the driving path of tested vechicle i, and B is the driving path of tested vechicle i Final position, tAAt the time of correspondence for A, tBAt the time of correspondence for B, L is the Largrangian in the driving path of tested vechicle i, xi For the actual measurement path length travel of tested vechicle i, yiIt is displaced for the actual measurement path lateral of tested vechicle i,It is tested vechicle i along actual measurement road The longitudinal driving speed of diameter,The longitudinal acceleration in path is surveyed for the edge tested vechicle i,It is tested vechicle i along the cross in actual measurement path To travel speed, RiFor resistance field, GiFor the constant field of force, FaiThe risk of tested vechicle i is acted on for lane line or road boundary a Power, EaiFor positioned at (xa,ya) at the potential energy field that is formed of lane line or road boundary a in (xi,yi) at vector field strength；VjiFor it The potential energy that remaining road user j generates tested vechicle i, FjiThe risk that tested vechicle i is generated is acted on for remaining road user j Power, EjiFor remaining road user j formed kinetic energy field in (xi,yi) at vector field strength；miFor the quality of tested vechicle i；G is Acceleration of gravity, f are coefficient of rolling resistance, iαFor the gradient, CDiFor the air resistance coefficient of tested vechicle i, WiFor the windward side of tested vechicle i Product, λiFor the correction coefficient of rotating mass of tested vechicle i, PaFor the road impact factor at lane line a or road boundary, PiFor quilt Impact factor at measuring car i, PjFor the road impact factor at remaining road user j, D is lane width, raiFor from lane Line a or road boundary are directed toward the mass center (x of tested vechicle ii,yi) distance vector, MiFor the equivalent mass of tested vechicle i, MjFor remaining The equivalent mass of road user j, K are adjustment factor, rjiFor remaining road user j mass center (xj,yj) it is directed toward tested vechicle i matter The heart (xi,yi) the distance between vector, vjFor the velocity vector of remaining road user j, θjFor rjiWith vjAngle, a is lane Line or road boundary, b are the quantity of lane line or road boundary, and n is the quantity of remaining road user.
Further, the theory least action S* is obtained by following formula (16):
In formula,For the length travel of the reference path of tested vechicle i；For the lateral displacement of the reference path of tested vechicle i,It is tested vechicle i along the longitudinal driving speed of reference path,It is tested vechicle i along the cross running speed of reference path.
Further, the acquisition modes of the quantitatively evaluating foundation in S2 include:
The first situation: the quantitatively evaluating foundation can be with the numerical value of the practical function amount and theoretical least action The increase of difference and dullness becomes larger, be expressed as following formula (8):
Or
Second case: the quantitatively evaluating foundation can be with the numerical value of the practical function amount and theoretical least action The increase of difference and dullness becomes smaller, be expressed as following formula (9):
Or
Further, the multiple groups quantitatively evaluating in S4 is (y according to datak1,yk2 ...ykm), wherein k is a wherein net Serial number corresponding to connection degree, m are the quantity of different net connection vehicle distribution forms under the net connection degree；" S4 is obtained every in S5 Quantitatively evaluating described in one group is for statistical analysis according to data " method include: calculate each group described in quantitatively evaluating according to number According to average value, standard deviation, extreme value, frequency and frequency distribution feature or specific distribution fitting, and use distribution map and/or list The statistic analysis result is presented in form.
Further, in S5 the method for " quantitatively evaluating described in each group obtained S4 is for statistical analysis according to data " It specifically includes:
The quantitatively evaluating described in each group seeks its average value and extreme value according to data respectively, described tested automatic to obtain Control loop averagely can reach under different net connection degree and at least accessible level of intelligence grade or its intelligent water equality The mean value and lower limit of grade；
Wherein, the quantitatively evaluating is expressed as formula (17) according to the average value of data:
Under the first described situation, the quantitatively evaluating is according to the maximum that the extreme value of data is that formula (18) indicate ykmax:
ykmax=max { yk1,yk2,...,ykm} (18)
Under the second case, the quantitatively evaluating is according to the minimum that the extreme value of data is that formula (19) indicate ykmin:
ykmin=min { yk1,yk2,...,ykm} (19)。
Further, after S4 further include:
S4 is calculated net connection degree in the resulting practical function amount S and the automatic Pilot intelligence test scene by S6 The distribution form F of C and net connection vehicle, are stored as (C, F, S) form；
S7 joins vehicle by changing net connection degree C and net in the case of the automatic Pilot intelligence test scene is identical Distribution form F executes step S41, is tested the tested automated driving system and record the corresponding reality of test process Actuating quantity S.
The present invention also provides a kind of assessment device of automated driving system level of intelligence under different net connection degree, the intelligence Horizontal assessment device includes:
Information acquisition module is used to acquire the driving parameters of tested vechicle, remaining road in automatic Pilot intelligence test scene The driving parameters and road environment test data of road user；
Actuating quantity computing module, according to the information acquisition module obtain the tested vechicle driving parameters, remaining The driving parameters and road environment test data of road user obtain the reality in the actual measurement path of the tested vechicle traveling Actuating quantity, and according to the reference arm planned in advance in the automatic Pilot intelligence test scene remaining described road user Diameter and road environment test data obtain the theoretical least action under the automatic Pilot intelligence test scene；With
Statistical appraisal module is used to store the assessment interval for evaluating the level of intelligence grade of tested automated driving system, And the practical function amount and theoretical least action obtained according to the actuating quantity computing module, it obtains described be tested and drives automatically Multiple groups quantitatively evaluating of the system under different net connection degree is sailed according to data, and the quantitatively evaluating described in each group according to data into Row statistical analysis, and the assessment interval according to belonging to each statistic analysis result, to the intelligence of the tested automated driving system Level is evaluated.
Further, the actuating quantity computing module includes:
Tested vechicle practical function amount computing unit tests number according to the driving parameters of the tested vechicle and road environment According to calculating the practical function amount S in tested vechicle actual measurement path0；
Road constraint practical function amount computing unit, according to the row of the road environment test data and the tested vechicle Data are sailed, traffic safety field theory is based on, establish the static risk field of lane line, road boundary or static-obstacle thing, calculate vehicle Diatom, road boundary or static-obstacle thing are to road constraint practical function amount S in tested vechicle actual measurement path1；
Remaining road user practical function amount computing unit, according to the driving parameters and road of remaining road user Environmental testing data calculate the practical function amount S in remaining road user actual measurement path2；
Collection unit, according to its tested vechicle practical function amount computing unit, road constraint practical function amount computing unit Summarized with the respective calculated result of remaining road user practical function amount computing unit, the tested vechicle is calculated as follows The practical function amount S in the actual measurement path of traveling:
S=S0-S1-S2。
Beneficial effects of the present invention: the present invention popularizes necessary long-term of process for this automated driving system of mixed traffic Assessment under stage is based on automatic Pilot intelligent testing examination hall platform, provides automated driving system intelligence under a kind of different net connection degree Assessment method and device that can be horizontal.In specific assessment process, foundation acts on figureofmerit with more objective and interpretable side The level of intelligence of formula weight automated driving system, and can directly embody the reflection big key factor of level of intelligence two --- safety and height Effect.In the Grading assessment of level of intelligence, the prior art has been abandoned by the way of artificially assuming scene and task complexity, But it is objective reasonably from multi-angle pair by the numerical value discrepancy mappings between building practical function amount and theoretical least action Level of intelligence grade of the automated driving system in test scene is evaluated, and this assessment method is not limited to specifically test Scene, it is widely applicable.In addition, assessment method provided by the present invention is combined with statistical analysis, difference can be coped with well Under net connection degree due to net connection vehicle be distributed bring randomness, make its can also from whole angle to automated driving system in different nets The evaluating result of level of intelligence is more three-dimensional and true under connection degree.
Detailed description of the invention
Fig. 1 is the control logic structural schematic diagram that the present invention is based on automatic Pilot intelligent testing examination hall；
Fig. 2 is the process figure of automated driving system level of intelligence under different net connection degree in the present invention；
Fig. 3 is that automatic Pilot intelligent testing provided by the invention examination hall carries out the survey of no signal lamp intersection automated driving system The schematic diagram of a scenario commented；
Fig. 4 is that test scene generates the schematic diagram with reproduction under different net connection degree in the present invention；
Fig. 5 is the assessment device internal module schematic diagram of vehicle intelligent level provided by the invention.
Specific embodiment
In the accompanying drawings, using same or similar marked as same or similar element or with the same or similar functions Element.The embodiment of the present invention is described in detail with reference to the accompanying drawing.
The assessment method of automated driving system level of intelligence is based on automatic under difference net connection degree provided by the present embodiment Driving intelligent checkout area and implement, Fig. 1 shows the control logic structural schematic diagram in automatic Pilot intelligent testing examination hall, and Fig. 3 is shown Automatic Pilot intelligent testing provided in this embodiment examination hall carries out the scene of no signal lamp intersection automated driving system assessment Schematic diagram.As shown in figures 1 and 3, automatic Pilot intelligent testing examination hall can carry out information exchange with control centre 4.Automatic Pilot Intelligent testing examination hall is the test environment for netting connection completely, equipped with car networking communications facility, trackside communications facility 5, test road The equipment such as road 6 and differential GPS high-accuracy position system 7, these equipment say that collected various data feed back to control centre 4.Control centre 4 can planning to tested automatic driving vehicle 1 (being hereafter referred to as " tested vechicle i or tested vechicle 1 "), mould Quasi-, monitoring and long-range control.
Specifically, as shown in Fig. 2, automated driving system level of intelligence under difference net connection degree provided by the present embodiment Assessment method includes the following steps:
S1 is based on traffic safety field theory, chooses evaluation index of the actuating quantity as automated driving system level of intelligence.Make Dosage S had not only included the traffic risk variable of direct reflection safety, but also the product in time aspect including directly reflecting high efficiency Point, actuating quantity S can be formula (1):
In formula (1), S is actuating quantity of the tested vechicle in test traffic process, and test traffic process is referred to by test quilt Measuring car turns left from initial position A through no signal lamp intersection, and the stop line in directed overshoot lane, the stop bit of arrival Set the whole process of B.tAIt, can be right for the initial time (at the time of tested vechicle arrival initial position A is corresponded to) for testing traffic process Answer the initial conditions of test assignment.tBFor end time (the test traffic process tested vechicle arrival stop bit for testing traffic process At the time of setting B and correspond to), the termination condition of test assignment can be corresponded to.L is Lagrange of the tested vechicle in test traffic process Amount.
By selected evaluation index, the mode that S1 can be objective and interpretable quantifies the intelligent water of automated driving system It is flat.
Certainly, in complicated test traffic process, the traffic risk of tested vechicle includes that road constraint and remaining road make The venture influence that user generates it.It, can be by all kinds of traffic venture influences with Largrangian according to traffic safety field theory Form is quantified, then Largrangian L can be formula (2):
Gi=mig (4)
Fai=Eai·Mi·Pi (5)
For formula (1) into formula (7), A is the initial position of the driving path of tested vechicle i, and B is the driving path of tested vechicle i Final position, tAAt the time of correspondence for A, tBAt the time of correspondence for B；L is the Largrangian in the driving path of tested vechicle i；xi For 1 length travel of actual measurement path P of tested vechicle i, yiFor 1 lateral displacement of actual measurement path P of tested vechicle i,It is tested vechicle i along reality The longitudinal driving speed of path P 1 is surveyed,The longitudinal acceleration of path P 1 is surveyed for the edge tested vechicle i,It is tested vechicle i along actual measurement The cross running speed of path P 1, above-mentioned kinematics information can be obtained by GPS；RiFor resistance field, GiFor the constant field of force；FaiFor The risk active force of lane line or road boundary a to tested vechicle i, EaiFor positioned at (xa,ya) at lane line or road boundary a formed Potential energy field in (xi,yi) at vector field strength；VjiFor the potential energy that remaining road user j generates tested vechicle i, FjiFor remaining The risk active force that road user j generates tested vechicle i, EjiFor remaining road user j formed kinetic energy field in (xi,yi) The vector field strength at place；miFor the quality of tested vechicle i；G is acceleration of gravity, is generally taken as 9.81m/s2；F is coefficient of rolling resistance, It is determined according to tire and pavement behavior, is generally chosen between 0.015-0.02；iαFor the gradient, according to checkout area road geometry item Part determines；CDiFor the air resistance coefficient of tested vechicle i, is determined by tested vehicle face shaping, generally chosen between 0.25-0.5；Wi For the front face area of tested vechicle i, can be obtained according to tested vehicle geometry computations；λiIt converts for the gyrating mass of tested vechicle i Coefficient generally can be taken as 1.05 according to automobile theory relevant knowledge；LT,aFor the type of lane line a or road boundary, according to row The safe field theory of vehicle chooses analog value to its definition；PaFor the road impact factor at lane line a or road boundary, PiFor quilt Impact factor at measuring car i, PjFor the road impact factor at remaining road user j, with reference to having delivered correlative theses, one As each impact factor is set as 1；D is lane width, can be obtained by measurement checkout area lane width；raiFor from lane line a or Mass center (the x of road boundary direction tested vechicle ii,yi) distance vector；MiFor the equivalent mass of tested vechicle i, MjFor remaining road The equivalent mass of user j, equivalent mass are the functions of type of vehicle, real quality and travel speed, can refer to correlative theses It is calculated；K is adjustment factor, is generally set to 0.5 with reference to correlative theses；k1, k2And k3For constant coefficient, with reference to correlative theses one As be set to 1,1.2 and 45；rjiFor remaining road user j mass center (xj,yj) it is directed toward tested vechicle i mass center (xi,yi) between Distance vector；vjFor the velocity vector of remaining road user j；θjFor rjiWith vjAngle (being positive counterclockwise)；A is lane line Or road boundary；B is the quantity of lane line or road boundary；N is the quantity of remaining road user.
S2 is theoretically solved by the extreme value analysis to actuating quantity S, inherently there is a reference path P2, is corresponded to Theoretical least action S* is travelled according to reference path P2, the safety of corresponding vehicle and high efficiency reach most at this time Excellent state.In consideration of it, according to the numerical difference between the practical function amount S and theoretical least action S* in test traffic process It is different, the quantitatively evaluating of automated driving system level of intelligence is obtained according to y, which can be with the practical function according to y It measures the variation of the numerical value difference of S and theoretical least action S* and is monotonically changed.Such as: quantitatively evaluating can be with described according to y The becoming larger for numerical value difference of practical function amount S and theoretical least action S* and dullness becomes larger or quantitatively evaluating can be with according to y With the becoming larger for numerical value difference of the practical function amount S and theoretical least action S*, dullness becomes smaller." becoming larger and single herein Modulation is big " and " become larger and dullness becomes smaller " depend primarily on selected quantitatively evaluating can be according to quantitatively evaluating according to y according to y Acquisition form.
In one embodiment, quantitatively evaluating indicates an accepted way of doing sth (8) and formula (9) according to the acquisition form of y in the present embodiment One of mapping relationship f (S, S*) practical function amount S can be quantified and theoretical minimum made by the mapping relationship f (S, S*) Numerical value difference between dosage S*, i.e. quantitatively evaluating are according to y with the difference between practical function amount S and theoretical least action S* Change and be monotonically changed, scale and mapping relationship f (S, S*) are related.Mapping relationship f (S, S*) includes two kinds of situations:
The first situation: the quantitatively evaluating foundation can be with the numerical value of the practical function amount and theoretical least action The increase of difference and dullness becomes larger, be expressed as following formula (8):
Or
Second case: the quantitatively evaluating foundation can be with the numerical value of the practical function amount and theoretical least action The increase of difference and dullness becomes smaller, be expressed as following formula (9):
Or
Certainly, mapping relationship f (S, S*) can also be the other forms other than above-mentioned four kinds of forms, as long as meet volume Change the requirement that Appreciation gist y is monotonically changed with the change of divergence between practical function amount S and theoretical least action S*.
Preferably, S2 specifically includes following S21 and S22:
S21 calculates theory least action S* and its corresponding ginseng according to the practical function amount S in test traffic process Examine path P 2.According to the expression formula for the practical function amount S (functional) that formula (1) provides, variation is asked according to formula (10) and it is made to obtain 0, It is corresponding with specific reference to path P 2 (being the function of time) that theoretical least action S* can be solved, then reference path P2 is substituted into The specific value of theoretical least action S* can be obtained in formula (1):
S22, according to the theoretical least action S* that practical function amount S and S21 are obtained, by constructing automated driving system Acquisition form of the quantitatively evaluating of level of intelligence according to y, the quantization for establishing the level of intelligence grade for dividing automated driving system are commented Valence is according to y.In formula (8) and formula (9), it is clear that the numerical value difference between practical function amount S and theoretical least action S* passes through two The difference of person embodies.According to the definition of actuating quantity S, there must be practical function amount S to be greater than theoretical least action S*.
S3 determines variation model of the quantitatively evaluating according to y according to the acquisition modes of the quantitatively evaluating foundation in S2 It encloses, and marks off assessment interval (following several sections that at least two evaluations are tested the level of intelligence grade of automated driving system It is shown), the corresponding level of intelligence grade of each described assessment interval:
[y0,y1)[y1,y2)...[yn-1,yn]
Wherein, y0,y1,…,ynFor the endpoint value in each section.In test process, each endpoint value determines according to the following rules:
1. recruiting several drivers at random, under identical test scene, tested vechicle is manipulated by driver completely, is calculated Actuating quantity, quantitatively evaluating under different driver's manipulations is according to y, so as to calculate quantization Appreciation gist mean valueStandard deviation σhm；
2. being enabled when n is even numberWhen n is odd number, orderPreferably, n=5；
3. length equalization constructs each section, and is set as above-mentioned quantitatively evaluating establishing criteria difference σhmRatio value, i.e. yk- yk-1=kp·σhm,Siding-to-siding block length coefficient kpIt needs to choose in combination with degree of refinement；Preferably, kp=2；
4. adjusting the coverage area in ultra-Left section and ultra-Right section, mainly y0And yn, make it just and based on mapping relations Quantitatively evaluating is consistent according to the coverage area of y theoretically.
Such as: in the present embodiment, using the first situation and using the corresponding mapping relations in the left side, i.e. y= (S-S*)/S*.At this point, quantitatively evaluating is smaller according to y, level of intelligence is higher.In this embodiment, it is assumed thatσhm=0.25, Then quantitatively evaluating according to y range [0 ,+∞), and 5 sections, siding-to-siding block length coefficient k are divided into itp=2, each section pair A level of intelligence grade is answered, " very high ", " higher ", " medium ", " lower ", " very low ", following level of intelligence grade are followed successively by Shown in table 1:
Table 1
The section y [0,0.5) [0.5,1.0) [1.0,1.5) [1.5,2.0) [2.0,+∞)
Intelligent grade It is very high It is higher It is medium It is lower It is very low
S4 obtains the tested automatic Pilot in the automatic Pilot intelligence test scene of Fig. 2 to setting shown in Fig. 4 Multiple groups quantitatively evaluating of the system under different net connection degree is according to data.
Preferably, the quantitatively evaluating in S4 is specifically included according to the acquisition methods of data:
S41, by control centre 4, is generated and reproduction automatic Pilot intelligence test scene and change before each test Necessary test condition, to obtain the driving parameters of the tested vechicle under each test condition, the traveling of remaining road user Parameter and road environment test data.
" control centre 4 " can be monitored and control to automatic Pilot intelligent testing examination hall, and pass through automatic Pilot intelligence The feedback information of checkout area carries out data acquisition, constitutes the data center in test process." control centre 4 " can also be formed High-precision map has and communicates with the various types of vehicles in test scene and carry out the critical functions such as path planning to it.
Specifically, 4 software layer of control centre is according to the different classes of of the object of received data, and there are different software moulds Block is mated with, and each module operates independently of each other, is handled with merging optimal way.It is suitable for using to divide between module and module The communication mode of cloth, software scenario can be selected " ZeroMQ " and be used as communication pool, and " Publish-subscribe " of ZeroMQ is selected to work Mode.All modules can be all mounted on a ZeroMQ formula bus, and each module gets newest test information Afterwards, " subscription " can be carried out for the module of needs by its " publication " on the bus.It should be noted is that testing Vehicle is joined to the net in test scene before beginning and non-net connection vehicle has carried out different labels respectively, these are marked companion simultaneously The parameter transmitted with it is embodied on the bus, can play auxiliary " subscription " module screen information source it With.For example turn left to tested vechicle and tested by the level of intelligence of this process of no signal lamp intersection, to the test When automated driving system level of intelligence in scene is evaluated: when tested vechicle left-hand rotation passed through no signal lamp intersection, control Center 4 processed can precisely capture this information by its high-precision map and differential GPS positioning function, and terminate net connection degree Test process under conditions of this kind net connection vehicle distribution form, the information exchange of each intermodule of 4 software layer of control centre are whole therewith Only, but previous all test information have been stored by control centre 4, constitute the data center of the test process.
" remaining road user " mainly includes that net connection vehicle 2 and non-net join vehicle 3.
" necessary test condition " is preset a series of limited to net net the connection degree, net that connection vehicle accounting embodies Join the distribution form of vehicle and whether activates the independent sensing system of tested vechicle under the whole network connection degree.That is, in intelligence In energy checkout area, net connection degree and net connection vehicle distribution be can change, to establish different test conditions.
Each traffic element includes: that as shown in Figures 2 and 3, tested vechicle 1, net connection vehicle 2, non-net join vehicle in " test scene " 3, test road 6, test facilities and testing background object.In the present embodiment, test scene is set are as follows: tested vechicle 1 is by south orientation Straight close to no signal lamp intersection stop line, and safely by the crossing, left-hand rotation drives towards west side lane, crosses stopping in north Line simultaneously leaves the crossing.
In " test scene ", automated driving system to be tested and assessed is installed, and the automated driving system is connect on tested vechicle 1 In the centralized car networking for entering automatic Pilot intelligence test scene, it is allowed to pass through control according to itself algorithm and test condition The data center at center 4 processed obtains required external information.Net connection vehicle and non-net join vehicle, are not installed with to be tested and assessed oneself Dynamic control loop.Net connection vehicle 2 and non-net connection vehicle 3 are set as simulating the vehicle of pilot steering by control centre 4, and net connection vehicle 2 and non-net connection vehicle 3 can be communicated with control centre 4 and path planning is carried out to it.Net connection vehicle 2 and non-net join vehicle 3 difference be can be communicated and be exchanged with tested vechicle 1 by control centre 4 necessary information (such as position, speed plus Speed etc.).Test road 6 is the part road structure for the suitable test scene chosen in automatic Pilot intelligent testing examination hall, In In the present embodiment, test road 6 is right-angled intersection, and the hardware and software device for the instruction right of way signal such as remove traffic lights, structure Road is tested at no signal lamp intersection.Test facilities mainly include differential GPS base station 7, traffic camera, trackside communication set Standby 5 and communication equipment DSRC/LTE-V) etc..Testing background object mainly includes that no signal lamp intersection is nearby used for simulant building The background objects of object, block etc..
In each test, extracted from the test data that automatic Pilot intelligent testing examination hall feeds back to control centre 4 tested Driving parameters, net connection vehicle 2 and the non-net of vehicle join the driving parameters and road environment test data of vehicle 3.Wherein, " traveling Parameter " includes vertically and horizontally displacement, velocity and acceleration etc.." road environment test data " includes removing tested vechicle 1, net connection vehicle 2 With the routing information of remaining road user except non-net connection vehicle 3, surface roughness, the gradient, the wind-force of road environment, vehicle Diatom or road boundary position, static-obstacle object location and size etc..
Above-mentioned test data is all made of the interaction that centralized communication topological structure realizes information and data.Tested vechicle, net connection Vehicle and the non-net connection vehicle driving parameters that can extract itself from CAN, including be vertically and horizontally displaced, the ginseng such as speed, acceleration Number respectively be packaged, by DSRC or LTE-V communication technology real-time Transmission to trackside communications facility, and by trackside communications facility with The proprietary communication mode of control centre 4 is by each parameter feedback to control centre 4.Remaining road occupation is not considered in the present embodiment Person's (such as skimulated motion pedestrian, model sport cyclist and bicycle), but the validity of assessment method is had no effect on, and if depositing It, can be by the similarly communication technology and mode, by its routing information and parameter feedback to control centre in remaining road user 4.As for the test data of road environment, the surface roughness including road environment, road grade, wind-force size, direction and speed Degree, lane line or road boundary position, static-obstacle thing positions and dimensions size etc. are regarded as static data substantially, Thus at the beginning of test process, the test data of all road environments is packaged and is disposably passed by aforementioned communications techniques and mode Transport to control centre 4.
Moreover, during the test, tested vechicle needs to obtain by perception in the information auxiliary of his vehicle and road environment The intelligent decision of portion's automated driving system and control.On the one hand, some sensors are installed on tested vechicle, including camera, are swashed Optical radar, millimetre-wave radar, ultrasonic radar etc. meet automated driving system and want during the test to the hardware independently perceived It asks, and can realize being connected and disconnected from for software layer with interior tested automated driving system；On the other hand, the automated driving system Information, the automated driving system and 4 software of trackside communications facility and control centre may be obtained by the perceptive mode of net connection Layer tested vechicle module carry out information exchange, and 4 software layer tested vechicle module of control centre pass through again ZeroMQ formula bus with Other modules carry out information exchange, and the data at this time with non-net connection vehicle specific markers will not be allowed to be transferred to control centre 4 software layer tested vechicle modules, and then prevent this type of information to pass through control centre 4- trackside communication equipment-tested vechicle from source Communication link transmitted, also ensure that tested vechicle in test process can not get non-net connection vehicle by net connection mode Any information.
S42, after each test, according to the traveling of the tested vechicle obtained in the automatic Pilot intelligence test scene Parameter, the driving parameters of remaining road user and road environment test data obtain the actual measurement road of the tested vechicle traveling The practical function amount S of diameter P1.Wherein, using above-mentioned formula (1) and formula (2), actuating quantity S is the time integral of Largrangian L, is drawn The bright feed ration L of lattice includes three parts: the kinetic energy and resistance potential energy L of tested vechicle0, lane line, road boundary or static-obstacle thing are to quilt The potential energy V that measuring car generates1, total potential energy V that remaining road user generates tested vechicle2。
Wherein, " the kinetic energy and resistance potential energy L of tested vechicle0" calculation method is as follows:
By the driving parameters and part road environment test data of acquired tested vechicle, the static state in conjunction with tested vechicle is special Property parameter comprising quality, coefficient of rolling resistance, air resistance coefficient, front face area, correction coefficient of rotating mass etc. can calculate quilt The kinetic energy and resistance potential energy L of measuring car0, calculating formula such as following formula (11):
Wherein, " the potential energy V that lane line, road boundary or static-obstacle thing generate tested vechicle1" calculation method is as follows:
Based on traffic safety field theory, the static risk field of lane line, road boundary or static-obstacle thing can be established.According to The running data of acquired road environment test data and part tested vechicle, can calculate lane line, road boundary or static-obstacle The potential energy V that object generates tested vechicle1, calculating formula such as following formula (12):
Wherein, " total potential energy V that remaining road user generates tested vechicle2" calculation method is as follows:
Based on traffic safety field theory, the dynamic risk field of remaining road user can be established.Remaining road user packet Include net connection vehicle, non-net connection vehicle and vulnerable road user etc..According to his acquired vehicle and road environment test data, Total potential energy V that remaining road user generates tested vechicle can be calculated2, calculating formula such as following formula (13):
Shown in the calculation method of Largrangian L such as following formula (14):
L=L0-V1-V2 (14)
The time that task is completed according to tested vechicle in test process, the Largrangian L in test process is integrated, Under net connection degree and its distribution occasion that this time test can be calculated, tested vechicle surveys actuating quantity S corresponding to path P 1, practical Actuating quantity S calculating formula is following formula (15):
After determining selected this current test scene of no signal lamp intersection left-hand rotation of the present embodiment, control centre 4 Can test every time it is initial when reappear the test scene, the reproducing method of the test scene specifically includes:
It is preset when being not yet in there are some traffic elements (such as tested vechicle, net connection vehicle, non-net join vehicle) Initial testing conditions when, firstly, control centre 4 controls energy to the efficient monitoring of the automation of element each in checkout area by it Power carries out high-precision path planning for these traffic elements for being unsatisfactory for test condition, it is made as early as possible and accurately to arrive at symbol Close the position of initial testing conditions.Then, control centre 4 carries out high-precision attitude control to all traffic elements in place, makes It meets initial testing Gesture.Finally, control centre 4 is real to all traffic elements set in predeterminated position and posture Existing initial motion control makes each traffic element in scheduled initial position, with preset posture while reaching initial testing and wanted The kinematic parameter (such as speed, acceleration) asked.
Before being tested, as shown in figure 4, control centre 4 also needs the net connection to all vehicles in addition to tested vechicle Feature is set and is marked.The setting includes two dimensions with label.One of dimension is the setting of net connection degree: choosing Take a certain net connection degree (such as 50%), i.e. net connection vehicle scale shared in all vehicles in addition to tested vechicle.Its In another dimension be that net connection vehicle distribution form is set: due to only having Some vehicles for net connection vehicle, need to it in institute Have in vehicle and be allocated, forms of distribution selected by the present embodiment are to be randomly assigned, and meet the off line vehicle of mixed traffic environment point The unpredictability of cloth, and a form of label is carried out to the net connection vehicle distributed, and remaining non-net connection vehicle is carried out Another form of label.It should be noted that the present embodiment illustrates a kind of net connection under a kind of net connection degree with not repeating Vehicle distribution form, and under real conditions during a series of loop tests, the distribution form of net connection degree and net connection vehicle can Become rather than immobilizes.
S43, according to the reference planned in advance in the automatic Pilot intelligence test scene remaining described road user Path P 2 and road environment test data obtain the theoretical least action S* under the automatic Pilot intelligence test scene. Wherein, the theory least action S* is obtained by following formula (16):
In formula,For the length travel of the reference path P2 of tested vechicle i；For the transverse direction of the reference path P2 of tested vechicle i Displacement,It is tested vechicle i along the longitudinal driving speed of reference path P2,It is tested vechicle i along the cross running of reference path P2 Speed.
No matter tested vechicle should be deposited it should be noted that how the net connection degree and its distribution occasion in test scene change In a unified theoretical least action S*, this is the calculating side of the continuous reproduction and theoretical least action by test scene What formula was determined.
S44, according to the numerical value between the theoretical least action in the practical function amount and S43 surveyed in S42 Difference obtains quantitatively evaluating described in multiple groups according to data in conjunction with the acquisition modes of the quantitatively evaluating foundation in S2.Specifically Ground, to the corresponding one group of actuating quantity data (S of all nets connection vehicle distribution form under each net connection degreek1,Sk2 ...Skm) knot It rationally discusses least action S* and carries out difference degree calculating, export corresponding one group of quantitatively evaluating according to data.Each group of quantization Appreciation gist data pass through (yk1,yk2 ...ykm) presented, also, k is serial number corresponding to a wherein net connection degree, m is The quantity of different net connection vehicle distribution forms under the net connection degree.
S5, the quantitatively evaluating described in each group of S4 acquisition is for statistical analysis according to data, and according to each statistical analysis As a result the assessment interval in affiliated S3 evaluates the level of intelligence of the tested automated driving system, finally will be by Level of intelligence Grading assessment result of the automated driving system under different net connection degree is surveyed to print out in visual form.
In one embodiment, " the quantitatively evaluating foundation data progress statistical described in each group of S4 acquisition in S5 Analysis " method include:
Average value, standard deviation, extreme value, frequency and frequency distribution feature of the quantitatively evaluating described in calculating each group according to data Or specific distribution fitting, and the statistic analysis result is presented using distribution map and/or sheet format.Specifically, to each group The quantitatively evaluating seeks its average value and extreme value according to data respectively, to obtain the tested automated driving system in different nets It averagely can reach the mean value and lower limit at least accessible level of intelligence grade or its level of intelligence grade under connection degree.
Wherein, the quantitatively evaluating is expressed as formula (17) according to the average value of data:
In the quantitatively evaluating according to using the increasing with the practical function amount and the numerical value difference of theoretical least action In the case of big and dullness becomes larger, so the maximum of the quantitatively evaluating foundation determines the lower limit of level of intelligence grade, at this time The quantitatively evaluating is according to the maximum y that the extreme value of data is that formula (18) indicatekmax:
ykmax=max { yk1,yk2,...,ykm} (18)
In the quantitatively evaluating according to using the increasing with the practical function amount and the numerical value difference of theoretical least action In the case of big and dullness becomes smaller, so the minimum of the quantitatively evaluating foundation determines the lower limit of level of intelligence grade, at this time The quantitatively evaluating is according to the minimum y that the extreme value of data is that formula (19) indicatekmin:
ykmin=min { yk1,yk2,...,ykm} (19)
In one embodiment, " evaluating the level of intelligence of the tested automated driving system " in S5 is specifically According to the statistic analysis result of quantitatively evaluating foundation and level of intelligence grade interval, to tested automated driving system in different nets Level of intelligence under connection degree carries out multi-angle Grading assessment.Wherein, " multi-angle " focus includes: generally, from average meaning Adopted angle is tested automated driving system accessible level of intelligence grade, the quilt under different net connection degree under different net connection degree Survey at least accessible level of intelligence grade of automated driving system, tested intelligence of the automated driving system under different net connection degree The distribution characteristics etc. of hierarchical level.Particularly, it is based only upon tested vechicle under zero net connection degree and independently perceives tested automatic Pilot system Unite accessible level of intelligence grade, be based only upon under the whole network connection degree more vehicle nets connection and perceive tested automated driving systems and can reach Level of intelligence grade.
In one embodiment, after S4 further include:
S6, by S4 calculate gained tested vechicle no signal lamp intersection test process practical function amount S, with it is described from The distribution form F of net connection degree C and net connection vehicle, are stored in level of intelligence in the form of one group of data in dynamic driving intelligent test scene Assessment device statistical analysis module in data storage cell, storage form is as follows:
(C, F, S) (20)
In formula (20), it can be decimal form be also percents that C, which is net connection degree,；F is net connection vehicle distribution form number According to label, a kind of label correlating method need to be established, the different distributions form of vehicle is joined by data markers net.In the present embodiment, It is associated with net connection vehicle distribution form that serial number is first passed through in advance, then serial number can mark net connection vehicle distribution form；S is test process Practical function amount.
S7 joins vehicle by changing net connection degree C and net in the case of the automatic Pilot intelligence test scene is identical Distribution form F executes step S41, is tested the tested automated driving system and record the corresponding reality of test process Actuating quantity S.
As shown in figure 5, the present invention also provides a kind of assessment dresses of automated driving system level of intelligence under different net connection degree It sets, the assessment device of the level of intelligence is located on tested vechicle, after an EOT end of test, the level of intelligence installed on tested vechicle Assessment device information and date request will be carried out by roadside device to control centre 4, the data center of control centre 4 will be to All data that can be obtained by control centre 4 in this open test process of the assessment device of level of intelligence, whether net joins Vehicle driving parameters are also non-net connection vehicle driving parameters, and the specific markers of the two do not work at this time.
The assessment device of the level of intelligence include information acquisition module, actuating quantity computing module, statistical appraisal module and Result display module, in which:
Information acquisition module makes for acquiring the driving parameters of tested vechicle, remaining road in automatic Pilot intelligence test scene The driving parameters and road environment test data of user.
Driving parameters, remaining road for the tested vechicle that actuating quantity computing module is obtained according to the information acquisition module The driving parameters and road environment test data of user obtain the practical work of the actual measurement path P 1 of the tested vechicle traveling Dosage, and according to the reference path planned in advance in the automatic Pilot intelligence test scene remaining described road user P2 and road environment test data obtain the theoretical least action under the automatic Pilot intelligence test scene.
Statistical appraisal module is used to store the assessment interval for evaluating the level of intelligence grade of tested automated driving system, and The practical function amount and theoretical least action obtained according to the actuating quantity computing module, obtains the tested automatic Pilot system The multiple groups quantitatively evaluating united under different net connection degree is according to data, and the quantitatively evaluating described in each group is united according to data Meter analysis, and the assessment interval according to belonging to each statistic analysis result, to the level of intelligence of the tested automated driving system It is evaluated.
Result display module is for printing out in visual form tested automated driving system in different net connection degree Under level of intelligence Grading assessment result.
In one embodiment, the actuating quantity computing module include tested vechicle practical function amount computing unit, road about Beam practical function amount computing unit, remaining road user practical function amount computing unit and collection unit, in which:
Tested vechicle practical function amount computing unit will request by the communication technology and mode and obtain the data of control centre 4 The driving parameters and part road environment test data for the tested vechicle that center is stored recall the static characteristic ginseng of tested vechicle Number comprising quality, coefficient of rolling resistance, air resistance coefficient, front face area and correction coefficient of rotating mass etc. utilize following formula (21) to formula (24), the kinetic energy and resistance potential energy L that calculate each time step in tested vechicle actual measurement path P 1 can be obtained0, and it is right It is integrated, and the practical function amount S in the tested vechicle test process is calculated0:
Gi=mig (24)
Formula (21) is into formula (24), RiFor resistance field；GiFor the constant field of force；miFor the quality of tested vechicle i；xiFor tested vechicle i Length travel；yiFor the lateral displacement of tested vechicle i；G is acceleration of gravity；F is coefficient of rolling resistance；iαFor the gradient；CDiFor The air resistance coefficient of tested vechicle i；WiFor the front face area of tested vechicle i；λiFor the correction coefficient of rotating mass of tested vechicle i.
Road constraint practical function amount computing unit will request by the communication technology and mode and obtain the number of control centre 4 The running data of the road environment test data and part tested vechicle that are stored according to center is based on traffic safety field theory, can build The static risk field of vertical lane line, road boundary or static-obstacle thing, so as to calculate lane line, road boundary or static-obstacle Potential energy V of the object to the generation of each time step in tested vechicle actual measurement path P 11And it is integrated to obtain road constraint reality Border actuating quantity S1, calculation is as follows:
Fai=Eai·Mi·Pi (27)
In formula, EaiFor positioned at (xa,ya) at the potential energy field that is formed of lane line a or road boundary in (xi,yi) at vector field By force, LT,aFor the type of lane line a or road boundary, PaFor the road impact factor at lane line a or road boundary, D is lane Width, raiFor the mass center (x for being directed toward tested vechicle i from lane line a or road boundaryi,yi) distance vector, K is adjustment factor, Mi For the equivalent mass of tested vechicle i, PiFor the road impact factor at tested vechicle i, a is lane line or road boundary, and b is lane line Or the quantity of road boundary, n are the quantity of remaining road user, k2For constant coefficient.
Remaining road user only includes net connection vehicle and non-net connection vehicle.Remaining road user practical function amount calculates Unit will request by the communication technology and mode and obtain the net connection vehicle and non-net connection that the data center of control centre 4 is stored The driving parameters and road environment test data of vehicle are based on traffic safety field theory, can establish net connection vehicle and non-net connection The dynamic risk field of vehicle, so as to calculate net connection vehicle and non-net connection vehicle to each time step in tested vechicle actual measurement path P 1 Total potential energy V of long generation2And it is integrated to obtain remaining road user practical function amount S2, calculation is as follows:
In formula, VjiFor the potential energy that remaining road user j generates tested vechicle i, FjiIt is remaining road user j to quilt The risk active force that measuring car i is generated, EjiFor remaining road user j formed kinetic energy field in (xi,yi) at vector field strength, grad EjiFor the magnetic field gradient that remaining road user j generates tested vechicle i, K is adjustment factor, PiFor the shadow at tested vechicle i Ring the factor, PjFor the road impact factor at remaining road user j, MiFor the equivalent mass of tested vechicle i, MjFor remaining road The equivalent mass of user j, rjiFor remaining road user j mass center (xj,yj) it is directed toward tested vechicle i mass center (xi,yi) between away from From vector, VijFor the potential energy that tested vechicle i generates remaining road user j, vjFor the velocity vector of remaining road user j, θjFor rjiWith vjAngle,For vjWith the angle of x-axis, k1And k3For constant coefficient.
Collection unit according to its tested vechicle practical function amount computing unit, road constraint practical function amount computing unit and its The respective calculated result of remaining road user practical function amount computing unit is summarized, and the net connection journey of this time test can be calculated Under degree and its distribution occasion, tested vechicle surveys actuating quantity S corresponding to path P 1, and calculation is as follows:
Finally it is noted that the above embodiments are merely illustrative of the technical solutions of the present invention, rather than its limitations.This The those of ordinary skill in field is it is understood that be possible to modify the technical solutions described in the foregoing embodiments or right Part of technical characteristic is equivalently replaced；These are modified or replaceed, and it does not separate the essence of the corresponding technical solution originally Invent the spirit and scope of each embodiment technical solution.

Claims (10)

1. the assessment method of automated driving system level of intelligence under a kind of difference net connection degree, which is characterized in that including walking as follows It is rapid:
S1 is based on traffic safety field theory, chooses evaluation index of the actuating quantity as automated driving system level of intelligence；
S2 is obtained automatic according to the numerical value difference between the practical function amount and theoretical least action in test traffic process The quantitatively evaluating foundation of control loop level of intelligence, quantitatively evaluating foundation can be with the practical function amounts and theoretical minimum work The variation of the numerical value difference of dosage and be monotonically changed；
S3 determines the variation range of the quantitatively evaluating foundation according to the acquisition modes of the quantitatively evaluating foundation in S2, and Mark off the assessment interval that at least two evaluations are tested the level of intelligence grade of automated driving system, each described assessment interval A corresponding level of intelligence grade；
S4 obtains the tested automated driving system in different net connection degree in the automatic Pilot intelligence test scene of setting Under multiple groups quantitatively evaluatings according to data；
S5, the quantitatively evaluating described in each group of S4 acquisition is for statistical analysis according to data, and according to each statistic analysis result The assessment interval in affiliated S3 evaluates the level of intelligence of the tested automated driving system.
2. the assessment method of automated driving system level of intelligence, feature exist under difference net connection degree as described in claim 1 In the quantitatively evaluating in S4 is specifically included according to the acquisition methods of data:
According to the driving parameters of the tested vechicle obtained in the automatic Pilot intelligence test scene, remaining road user Driving parameters and road environment test data obtain the practical function amount S in the actual measurement path of the tested vechicle traveling；
According to the reference path planned in advance in the automatic Pilot intelligence test scene to remaining described road user and Road environment test data obtains the theoretical least action S* under the automatic Pilot intelligence test scene；
According between the practical function amount and the theoretical least action under the automatic Pilot intelligence test scene Numerical value difference obtains the quantitatively evaluating according to data in conjunction with the acquisition modes of the quantitatively evaluating foundation in S2.
3. the assessment method of automated driving system level of intelligence, feature exist under difference net connection degree as claimed in claim 2 In the practical function amount S in S41 is obtained by following formula (1) to formula (7):
Gi=mig (4)
Fai=Eai·Mi·Pi (5)
For formula (1) into formula (7), A is the initial position of the driving path of tested vechicle i, and B is the termination of the driving path of tested vechicle i Position, tAAt the time of correspondence for A, tBAt the time of correspondence for B, L is the Largrangian in the driving path of tested vechicle i, xiFor quilt The actual measurement path length travel of measuring car i, yiIt is displaced for the actual measurement path lateral of tested vechicle i,It is tested vechicle i along actual measurement path Longitudinal driving speed,The longitudinal acceleration in path is surveyed for the edge tested vechicle i,It is tested vechicle i along the lateral line in actual measurement path Sail speed, RiFor resistance field, GiFor the constant field of force, FaiRisk active force for lane line or road boundary a to tested vechicle i, Eai For positioned at (xa,ya) at the potential energy field that is formed of lane line or road boundary a in (xi,yi) at vector field strength；VjiFor remaining road The potential energy that user j generates tested vechicle i, FjiFor the risk active force that remaining road user j generates tested vechicle i, EjiFor The kinetic energy field that remaining road user j is formed is in (xi,yi) at vector field strength；miFor the quality of tested vechicle i；G is gravity acceleration Degree, f is coefficient of rolling resistance, iαFor the gradient, CDiFor the air resistance coefficient of tested vechicle i, WiFor the front face area of tested vechicle i, λiFor quilt The correction coefficient of rotating mass of measuring car i, PaFor the road impact factor at lane line a or road boundary, PiAt tested vechicle i Impact factor, PjFor the road impact factor at remaining road user j, D is lane width, raiFor from lane line a or road Mass center (the x of boundary direction tested vechicle ii,yi) distance vector, MiFor the equivalent mass of tested vechicle i, MjFor remaining road occupation The equivalent mass of person j, K are adjustment factor, rjiFor remaining road user j mass center (xj,yj) it is directed toward tested vechicle i mass center (xi,yi) The distance between vector, vjFor the velocity vector of remaining road user j, θjFor rjiWith vjAngle, a be lane line or road Boundary, b are the quantity of lane line or road boundary, and n is the quantity of remaining road user.
4. the assessment method of automated driving system level of intelligence, feature exist under difference net connection degree as claimed in claim 3 In the theory least action S* is obtained by following formula (16):
In formula,For the length travel of the reference path of tested vechicle i；For the lateral displacement of the reference path of tested vechicle i,For Tested vechicle i along reference path longitudinal driving speed,It is tested vechicle i along the cross running speed of reference path.
5. the assessment method of automated driving system level of intelligence, feature exist under difference net connection degree as claimed in claim 2 In the acquisition modes of the quantitatively evaluating foundation in S2 include:
The first situation: the quantitatively evaluating foundation can be with the numerical value difference of the practical function amount and theoretical least action Increase and dullness becomes larger, be expressed as following formula (8):
Second case: the quantitatively evaluating foundation can be with the numerical value difference of the practical function amount and theoretical least action Increase and dullness becomes smaller, be expressed as following formula (9):
6. the assessment side of automated driving system level of intelligence under the different net connection degree as described in any one of claims 1 to 5 Method, which is characterized in that the multiple groups quantitatively evaluating foundation data in S4 are (yk1,yk2...ykm), wherein k is that wherein a net joins journey The corresponding serial number of degree, m are the quantity of different net connection vehicle distribution forms under the net connection degree；" each group that S4 is obtained in S5 The quantitatively evaluating is for statistical analysis according to data " method include: calculate each group described in quantitatively evaluating according to data Average value, standard deviation, extreme value, frequency and frequency distribution feature or specific distribution fitting, and use distribution map and/or sheet format The statistic analysis result is presented.
7. the assessment method of automated driving system level of intelligence, feature exist under difference net connection degree as claimed in claim 6 In the method for " quantitatively evaluating described in each group of S4 acquisition is for statistical analysis according to data " specifically includes in S5:
The quantitatively evaluating described in each group seeks its average value and extreme value according to data respectively, to obtain the tested automatic Pilot System averagely can reach under different net connection degree and at least accessible level of intelligence grade or its level of intelligence grade Mean value and lower limit；
Wherein, the quantitatively evaluating is expressed as formula (17) according to the average value of data:
Under the first described situation, the quantitatively evaluating is according to the maximum y that the extreme value of data is that formula (18) indicatekmax:
ykmax=max { yk1,yk2,...,ykm} (18)
Under the second case, the quantitatively evaluating is according to the minimum y that the extreme value of data is that formula (19) indicatekmin:
ykmin=min { yk1,yk2,...,ykm} (19)。
8. the assessment method of automated driving system level of intelligence, feature exist under difference net connection degree as claimed in claim 6 In after S4 further include:
S6, by S4 calculate the resulting practical function amount S, with the automatic Pilot intelligence test scene in net connection degree C and The distribution form F of net connection vehicle, is stored as (C, F, S) form；
S7, in the case of the automatic Pilot intelligence test scene is identical, by the distribution for changing net connection degree C and net connection vehicle Form D executes step S41, is tested the tested automated driving system and record the corresponding practical function of test process Measure S.
9. the assessment device of automated driving system level of intelligence under a kind of difference net connection degree, which is characterized in that the intelligent water Flat assessment device is located on tested vechicle, comprising:
Information acquisition module, being used to acquire the driving parameters of tested vechicle, remaining road in automatic Pilot intelligence test scene makes The driving parameters and road environment test data of user；
Actuating quantity computing module, according to driving parameters, remaining road of the tested vechicle that the information acquisition module obtains The driving parameters and road environment test data of user obtain the practical function in the actual measurement path of the tested vechicle traveling Amount, and according to the reference path that remaining described road user is planned in advance in the automatic Pilot intelligence test scene with And road environment test data, obtain the theoretical least action under the automatic Pilot intelligence test scene；With
Statistical appraisal module is used to store the assessment interval for evaluating the level of intelligence grade of tested automated driving system, and The practical function amount and theoretical least action obtained according to the actuating quantity computing module, obtains the tested automatic Pilot system The multiple groups quantitatively evaluating united under different net connection degree is according to data, and the quantitatively evaluating described in each group is united according to data Meter analysis, and the assessment interval according to belonging to each statistic analysis result, to the level of intelligence of the tested automated driving system It is evaluated.
10. the assessment device of automated driving system level of intelligence, feature exist under difference net connection degree as claimed in claim 9 In the actuating quantity computing module includes:
Tested vechicle practical function amount computing unit, according to the driving parameters of the tested vechicle and road environment test data, meter Calculate the practical function amount S in tested vechicle actual measurement path0；
Road constraint practical function amount computing unit, according to the traveling number of the road environment test data and the tested vechicle According to, be based on traffic safety field theory, establish the static risk field of lane line, road boundary or static-obstacle thing, calculate lane line, Road boundary or static-obstacle thing are to road constraint practical function amount S in tested vechicle actual measurement path1；
Remaining road user practical function amount computing unit, according to the driving parameters and road environment of remaining road user Test data calculates the practical function amount S in remaining road user actual measurement path2；
Collection unit, according to its tested vechicle practical function amount computing unit, road constraint practical function amount computing unit and its The respective calculated result of remaining road user practical function amount computing unit is summarized, and the tested vechicle traveling is calculated as follows Actual measurement path practical function amount S:
S=S0-S1-S2。
CN201910611202.2A 2019-07-08 2019-07-08 Method and device for evaluating intelligent level of automatic driving system under different internet connection degrees Active CN110414803B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201910611202.2A CN110414803B (en) 2019-07-08 2019-07-08 Method and device for evaluating intelligent level of automatic driving system under different internet connection degrees

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201910611202.2A CN110414803B (en) 2019-07-08 2019-07-08 Method and device for evaluating intelligent level of automatic driving system under different internet connection degrees

Publications (2)

Publication Number Publication Date
CN110414803A true CN110414803A (en) 2019-11-05
CN110414803B CN110414803B (en) 2022-01-07

Family

ID=68360624

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201910611202.2A Active CN110414803B (en) 2019-07-08 2019-07-08 Method and device for evaluating intelligent level of automatic driving system under different internet connection degrees

Country Status (1)

Country Link
CN (1) CN110414803B (en)

Cited By (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110853393A (en) * 2019-11-26 2020-02-28 清华大学 Intelligent network vehicle test field data acquisition and fusion method and system
CN111611746A (en) * 2020-05-20 2020-09-01 中国公路工程咨询集团有限公司 Intelligent network networking test oriented database management system
CN113793074A (en) * 2021-02-20 2021-12-14 同济大学 Social cooperation evaluation method for automatic driving vehicle
CN113887037A (en) * 2021-09-28 2022-01-04 中国科学技术大学先进技术研究院 Automatic driving system evaluation method under network connection environment with different permeability
CN113960996A (en) * 2020-07-20 2022-01-21 华为技术有限公司 Planning method and device for obstacle avoidance path of driving device
CN114298518A (en) * 2021-12-22 2022-04-08 北京工业大学 Road risk evaluation index system under networked vehicle environment

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2000222322A (en) * 1999-01-28 2000-08-11 Matsushita Electric Ind Co Ltd Auto-pilot device and auto-pilot method and recording medium
CN107782564A (en) * 2017-10-30 2018-03-09 青岛慧拓智能机器有限公司 A kind of automatic driving vehicle evaluation system and method
CN108255151A (en) * 2017-12-15 2018-07-06 工业和信息化部计算机与微电子发展研究中心（中国软件评测中心） A kind of evaluation system of automatic driving vehicle
CN108304986A (en) * 2017-12-15 2018-07-20 工业和信息化部计算机与微电子发展研究中心（中国软件评测中心） A kind of assessment method of automatic driving vehicle behavior safety
CN109814533A (en) * 2019-01-30 2019-05-28 同济大学 A kind of intelligent network connection vehicle of multimode fusion is in ring test system

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2000222322A (en) * 1999-01-28 2000-08-11 Matsushita Electric Ind Co Ltd Auto-pilot device and auto-pilot method and recording medium
CN107782564A (en) * 2017-10-30 2018-03-09 青岛慧拓智能机器有限公司 A kind of automatic driving vehicle evaluation system and method
CN108255151A (en) * 2017-12-15 2018-07-06 工业和信息化部计算机与微电子发展研究中心（中国软件评测中心） A kind of evaluation system of automatic driving vehicle
CN108304986A (en) * 2017-12-15 2018-07-20 工业和信息化部计算机与微电子发展研究中心（中国软件评测中心） A kind of assessment method of automatic driving vehicle behavior safety
CN109814533A (en) * 2019-01-30 2019-05-28 同济大学 A kind of intelligent network connection vehicle of multimode fusion is in ring test system

Cited By (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110853393A (en) * 2019-11-26 2020-02-28 清华大学 Intelligent network vehicle test field data acquisition and fusion method and system
CN110853393B (en) * 2019-11-26 2020-12-11 清华大学 Intelligent network vehicle test field data acquisition and fusion method and system
CN111611746A (en) * 2020-05-20 2020-09-01 中国公路工程咨询集团有限公司 Intelligent network networking test oriented database management system
CN113960996A (en) * 2020-07-20 2022-01-21 华为技术有限公司 Planning method and device for obstacle avoidance path of driving device
CN113793074A (en) * 2021-02-20 2021-12-14 同济大学 Social cooperation evaluation method for automatic driving vehicle
CN113793074B (en) * 2021-02-20 2023-08-29 同济大学 Social cooperation evaluation method for automatic driving vehicle
CN113887037A (en) * 2021-09-28 2022-01-04 中国科学技术大学先进技术研究院 Automatic driving system evaluation method under network connection environment with different permeability
CN113887037B (en) * 2021-09-28 2024-04-26 中国科学技术大学先进技术研究院 Automatic driving system evaluation method under network environment with different permeability
CN114298518A (en) * 2021-12-22 2022-04-08 北京工业大学 Road risk evaluation index system under networked vehicle environment

Also Published As

Publication number Publication date
CN110414803B (en) 2022-01-07

Similar Documents

Publication Publication Date Title
CN110414803A (en) The assessment method and device of automated driving system level of intelligence under different net connection degree
CN110531740B (en) Intelligent degree quantitative evaluation method for intelligent vehicle
US10192008B2 (en) Detecting static and dynamic objects
CN107272683A (en) Parallel intelligent vehicle control based on ACP methods
CN104569911B (en) OBU positioning method, RSU and ETC system
CN112800157B (en) Dynamic occupying grid model construction method and application architecture design method thereof
CN113029151B (en) Intelligent vehicle path planning method
CN103646557B (en) Based on the three-dimensional microform intelligent-traffic management system of Internet of Things
CN109686092B (en) Traffic network accessibility assessment method
CN110182217A (en) A kind of traveling task complexity quantitative estimation method towards complicated scene of overtaking other vehicles
CN109283843A (en) A kind of lane-change method for planning track merged based on multinomial with particle swarm algorithm
CN115662166B (en) Automatic driving data processing method and automatic driving traffic system
CN111880191A (en) Map generation method based on multi-agent laser radar and visual information fusion
CN116129066A (en) Automatic driving high-precision map model based on digital twin and high-precision static map manufacturing method
CN112258850A (en) Edge side multi-sensor data fusion system of vehicle-road cooperative system
CN113312733A (en) Method, device, equipment and storage medium for calibrating simulation model parameters of signal-controlled roundabout
CN117521389A (en) Vehicle perception test method based on vehicle-road collaborative perception simulation platform
Geng et al. Human-driver speed profile modeling for autonomous vehicle's velocity strategy on curvy paths
Sabikan et al. Modelling of time-to collision for unmanned aerial vehicle using particles swarm optimization
WO2021175821A1 (en) Computer-implemented method for route calculation for an autonomously driving motor vehicle, method for driving an autonomously driving motor vehicle, computer program product and motor vehicle
CN111208820B (en) Particle unmanned vehicle set under artificial intelligence big data, control method and medium
Prokudin et al. Development of vehicle speed forecasting method for intelligent highway transport system
Schultz et al. Calibration of distributions of commercial motor vehicles in CORSIM
CN110069894A (en) A kind of objective mapping test method for intelligent automobile traffic coordinating
Gao et al. A micro-simulation framework for studying CAVs behavior and control utilizing a traffic simulator, chassis simulation, and a shared roadway friction database

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