CN106644503B - A kind of intelligent vehicle planning aptitude tests platform - Google Patents
A kind of intelligent vehicle planning aptitude tests platform Download PDFInfo
- Publication number
- CN106644503B CN106644503B CN201610939638.0A CN201610939638A CN106644503B CN 106644503 B CN106644503 B CN 106644503B CN 201610939638 A CN201610939638 A CN 201610939638A CN 106644503 B CN106644503 B CN 106644503B
- Authority
- CN
- China
- Prior art keywords
- planning
- intelligent vehicle
- test
- measured
- vehicle
- 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.)
- Active
Links
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
- G01M17/00—Testing of vehicles
- G01M17/007—Wheeled or endless-tracked vehicles
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01C—MEASURING DISTANCES, LEVELS OR BEARINGS; SURVEYING; NAVIGATION; GYROSCOPIC INSTRUMENTS; PHOTOGRAMMETRY OR VIDEOGRAMMETRY
- G01C25/00—Manufacturing, calibrating, cleaning, or repairing instruments or devices referred to in the other groups of this subclass
Abstract
The invention discloses a kind of intelligent vehicles to plan aptitude tests platform, including disk, planning capacity test system, intelligent vehicle to be measured.The disk is used for stored samples data, assignment file, Key for Reference, code of points and test result;The planning capacity test system is the software systems tested for the planning ability of intelligent vehicle, is mainly made of three parts: test examination question exam pool, test process visualization and test result and method for quantitatively evaluating.The intelligent vehicle to be measured includes programmed decision-making unit, control unit and vehicle-mounted executing agency, is communicated between intelligent vehicle to be measured and planning capacity test system by network interface.Semi-physical simulation test platform provided by the invention based on " vehicle-environment " closed-loop system, enable intelligent vehicle one can controlled, experiment condition be easy to repeat it is variable, ability, which is safely and effectively tested, to be planned to it with a relatively appropriate cost and space.
Description
Technical field
The present invention relates to intelligent vehicle analysis test evaluation technical field more particularly to a kind of planning aptitude tests of intelligent vehicle are flat
Platform.
Background technique
Unmanned intelligent vehicle integrates the functions such as environment sensing, decision rule and control, is the heat studied recently
Point.Decision rule system is most important component in intelligent vehicle, and algorithm performance directly determines intelligent vehicle when driving
Safety and reliability.In order to construct the reliable planning system of performance, need adequately to test it in development process,
With various design defect present in exposure system.Conventional real train test is difficult to meet intelligent vehicle autonomous driving proficiency testing
Demand, firstly, the safety problem of intelligent vehicle is effectively verified not yet;Secondly, road needed for intelligent vehicle test
Condition is higher, needs to put into higher cost.Find a kind of repeatable, efficient and safe test method be at present there is an urgent need to
It solves the problems, such as.
Summary of the invention
The object of the invention is to remedy the disadvantages of known techniques, provides a kind of survey for intelligent vehicle planning ability
Platform and its evaluation method are tried, provides a kind of important idea and method with verifying for the analysis of intelligent vehicle planning system, design.
The present invention is achieved by the following technical solutions:
A kind of intelligent vehicle planning aptitude tests platform, including disk, planning capacity test system and intelligent vehicle to be measured.
The disk is used for stored samples data, assignment file, Key for Reference, code of points, test result.
The planning capacity test system is the software systems tested for the planning ability of intelligent vehicle, mainly
It is made of three parts: test examination question exam pool, test process visual Simulation, test result and method for quantitatively evaluating;
The intelligent vehicle to be measured includes programmed decision-making unit, control unit, vehicle body hanger and vehicle-mounted executing agency.To
Surveying intelligent vehicle is the test object for planning capacity test system, is communicated between the two by network interface.
The sample data being stored in disk is the grating map for planning aptitude tests, can pass through two kinds of sides
Formula obtains.First, passing through sensory perceptual system, such as camera, radar the sensor contextual data obtained and the planning being converted in advance
Map, i.e. grating map;Second, in a test system, tester can voluntarily draw grating map according to testing requirement.
The grating map is test macro for planning path test be used to simulated scenario environment it is a kind of two-dimensionally
Figure.For grating map using the lower right corner as coordinate origin, the coordinate of each grid is expressed as (xn,yn), attribute value -2, dynamic disorder
Object;- 1, static-obstacle thing;0, it can traffic areas;1, the planning path not driven to;2, the planning path run over;3, it is quick
Sensillary area domain.Wherein the planning in path carries out in non-barrier region.It is identified not in grating map with different marks simultaneously
Same object, such as starting point, terminal, task point and the intelligent vehicle model of planning path.
The planning map is divided into three kinds: sector planning map, Global motion planning map and Dynamic Programming map.Part rule
Drawing map size is 500px × 500px, correspond to actual scene environment with certain proportion, for example, actual scene for 100m ×
100m, then each 1px × 1px cell represents the scene in the region 1m × 1m.Global motion planning map is multiple sector plannings
Figure is spliced, i.e., larger range of scene environment, according to the real-time Regeneration planning map of the traveling-position of intelligent vehicle.Dynamically
Planning map is that test macro dynamically adds barrier on the basis of sector planning map or Global motion planning map, specifically
Ground, when intelligent vehicle model driving is to a certain sensitizing range, test macro is on the basis of initial planning path in specified position
Set the corresponding barrier of addition.The sensitizing range is that grating map has been set in initialization, with addition barrier
Position be it is one-to-one, that is, a certain sensitizing range HotArea has been determinedn: { (xa1,ya1),(xa2,ya2) ..., (xan,
yan), the region DynamicObstacle of corresponding addition barriern: { (xb1, yb1), (xb2, yb2) ..., (xbn, ybn)}
It also determines that simultaneously.
The intelligent vehicle model is the true dummy model of intelligent vehicle to be measured in a test system, dynamically instead
The driving status of intelligent vehicle to be measured is reflected, attribute has: position, speed, the information of course and barrier all around.
The test examination question exam pool can be divided into part, global and dynamic three according to the three of grating map kinds of different types
The different types of planning aptitude tests of kind.
The sector planning aptitude tests are based on single frames local grid map M, necessary point NESS={ (xs,ys),
(xg,yg), (xs,ys) and (xg,yg) it is respectively Origin And Destination, sector planning returns to path P={ (x of plannings,ys),(x1,
y1),(x2,y2),...,(xn,yn),(xg,yg), (xn,yn) it is the coordinate of cell successively passed through.
The Global motion planning aptitude tests are global grating map M, necessary point NESS={ (xs,ys),(xt1,yt1),
(xt2,yt2),...,(xtn,ytn),(xg,yg), (xs,ys) and (xg,yg) it is respectively Origin And Destination, (xt1,yt1),(xt2,
yt2),...,(xtn,ytn) it is necessary task point, it is desirable that path P={ (x that Global motion planning returnss,ys),(x1,y1),(x2,
y2),...,(xn,yn),(xg,yg) it must sequentially pass through these task points, i.e.,(xn,yn) it is the cell successively passed through
Coordinate.
The Dynamic Programming aptitude tests are intelligent vehicles during the locally or globally planning ability of progress, work as intelligence
Can vehicle driving to a certain sensitizing range of grating map when, test macro according to initial planning path grating map specific bit
It sets and dynamically adds barrier, according to test assignment, by starting point S to terminal E may there are two types of planning path A and B, when test is
When system detects intelligent vehicle model driving to sensitizing range A, the addition dynamic barrier A on grating map immediately, by unit
Lattice attribute value is changed to (xn,yn)=- 2 cancels the setting of sensitizing range A and sensitizing range B associated with it, by cell later
Attribute value is changed to (xn,yn)=0.The incidence relation of so-called sensitizing range for its corresponding dynamic barrier for, be exactly
Say that the corresponding dynamic barrier B of sensitizing range B cannot be added by being added to the corresponding dynamic barrier A in sensitizing range, be added to
Dynamic barrier B cannot add dynamic barrier A, otherwise just can path from starting point S to terminal E.At the same time, it surveys
Grating map change information by network signals intelligent vehicle to be measured, is made it change planning path in time, to reach by test system
The weight-normality to scene transition for detecting intelligent vehicle to be measured in real time draws ability.
The test process visualization formulation, it is characterised in that test macro is fed back according to intelligent vehicle to be measured
Information utilizes its motion state of intelligent vehicle model Dynamically Announce and track on grating map.Test macro will test examination question letter
Breath, including original grating map and test assignment etc. are sent to intelligent vehicle to be measured by network, and vehicle to be measured cooks up traveling
Path further controls the vehicles executing agency output actions such as steering column, throttle, braking, gear.Meanwhile these being executed
The action message of mechanism periodically feeds back to test macro, and test macro is according to the information such as speed, the course of feedback, control intelligence
Auto model carries out corresponding synchronization action.During Dynamic Programming aptitude tests, test macro is being continuously updated intelligent vehicle
While the position of model and motion profile, the obstacle information of addition is continuously updated, is dynamically shown to be measured
The path planning of intelligent vehicle is with weight-normality stroke as a result, reaching the effect of visualization of motion process.
The test result and evaluation method is test macro when intelligent vehicle to be measured completes test assignment,
It is analyzed and evaluated for the planning path and other test datas of submission, is carried out for its reasonability and optimality
Scoring.Some specific quantitatively evaluating indexs are shown below:
A) path length
B) task situation is completed
Whether the motion planning path by testing vehicle passes through starting point, terminal and task point in assignment file etc., comes
Judge whether to complete planning tasks success,
C) risk of collision (Risk, R) in path
It is less than most by the way that whether the minimum distance of every in the motion planning path of calculating vehicle and peripheral obstacle meet
Small collision distance, to judge that path whether there is risk,
D) time-consuming T is planned
T=TE-TS (4)
TSIt is denoted as the test assignment data time that intelligent vehicle to be measured receives test macro, TEIt is denoted as intelligent vehicle to be measured
Test assignment is completed, sends the time that task completes label to test macro.
Several evaluation indexes, path (weight) plan that every topic total score calculation method is as follows:
wiIndicate the weight of every kind of test index, SiIndicate the score of every kind of test index.
The intelligent vehicle to be measured is to collect network communication interface, decision making function module, control function module to hold with vehicle-mounted
Row structure is in the unmanned intelligent vehicle of one.After intelligent vehicle receives test assignment, held by decision and control module control
Row mechanism response action, while the motion change result that will test feeds back to test macro, test macro is by its dynamic change
Process reaction is on the intelligent vehicle model in virtual scene.Intelligent vehicle to be measured is placed on test-bed during the test,
It rolls wheel on rotary drum, carrys out simulated automotive and travelled on road, in the process of movement, the position of car body itself is not
Change.It thus meets and completes various testing experiments in relatively narrow lab space.
The invention has the advantages that the semi-physical simulation test platform provided by the invention based on " vehicle-environment " closed-loop system,
That has abandoned that unmanned vehicle tested in actual environment is various limited, enables intelligent vehicle can be controlled, real at one
The condition of testing be easy to repeat it is variable, ability, which is safely and effectively surveyed, to be planned to it with a relatively appropriate cost and space
Examination;
The probabilistic interference for taking environmental data is separately won present invention eliminates vehicle-mounted hardware sense part, by means of real ring
True test scene and sensing data, so more pointedly only examine the planning ability of intelligent vehicle in border
It surveys.Test macro dynamically becomes scene environment directly using having converted the grating map data of storage in disk
More, it is satisfied with the repeatability of test environment and the requirement of variability in this way;
The present invention visualizes the motion state of intelligent vehicle in test process in test macro, in virtual simulation environment
In, real-time dynamicly more intuitively monitor the motion change situation of intelligent vehicle;
The present invention proposes the method for a kind of analysis and assessment for the planning ability of intelligent vehicle, from overall performance,
The comprehensive more efficient planning ability for accurately assessing intelligent vehicle.
Detailed description of the invention
Fig. 1 is that intelligent vehicle plans capacity test system general frame figure.
Fig. 2 is cell attribute information in grating map.
Fig. 3 is the set-up mode of dynamic barrier and sensitizing range in grating map.
Fig. 4 (a) is the design flow diagram of Dynamic Programming aptitude tests;The determination flow chart of Fig. 4 (b) addition barrier.
Fig. 5 is the driving path of T moment intelligent vehicle preliminary planning to be measured.
Fig. 6 is the driving path for add intelligent vehicle weight-normality stroke to be measured after dynamic barrier the T+n moment.
Specific embodiment
A kind of intelligent vehicle planning aptitude tests platform, including disk, planning capacity test system and intelligent vehicle to be measured.
The disk is used for stored samples data, assignment file, Key for Reference, code of points, test result.
The planning capacity test system is the software systems tested for the planning ability of intelligent vehicle, mainly
It is made of three parts: test examination question exam pool, test process visual Simulation, test result and method for quantitatively evaluating;
As shown in Figure 1, the intelligent vehicle to be measured includes programmed decision-making unit, control unit, vehicle body hanger and vehicle-mounted
Executing agency.Intelligent vehicle to be measured is the test object for planning capacity test system, is communicated between the two by network interface.
The sample data being stored in disk is the grating map for planning aptitude tests, can pass through two kinds of sides
Formula obtains.First, passing through sensory perceptual system, such as camera, radar the sensor contextual data obtained and the planning being converted in advance
Map, i.e. grating map;Second, in a test system, tester can voluntarily draw grating map according to testing requirement.
The grating map is test macro for planning path test be used to simulated scenario environment it is a kind of two-dimensionally
Figure.For grating map using the lower right corner as coordinate origin, the coordinate of each grid is expressed as (xn,yn), attribute value is as shown in Figure 2 :-
2, dynamic barrier;- 1, static-obstacle thing;0, it can traffic areas;1, the planning path not driven to;2, the planning run over
Path;3, sensitizing range.Wherein the planning in path carries out in non-barrier region.Simultaneously with different marks in grating map
To identify different objects, such as starting point, terminal, task point and the intelligent vehicle model of planning path.
The planning map is divided into three kinds: sector planning map, Global motion planning map and Dynamic Programming map.Part rule
Drawing map size is 500px × 500px, correspond to actual scene environment with certain proportion, for example, actual scene for 100m ×
100m, then each 1px × 1px cell represents the scene in the region 1m × 1m.Global motion planning map is multiple sector plannings
Figure is spliced, i.e., larger range of scene environment, according to the real-time Regeneration planning map of the traveling-position of intelligent vehicle.Dynamically
Planning map is that test macro dynamically adds barrier on the basis of sector planning map or Global motion planning map, specifically
Ground, when intelligent vehicle model driving is to a certain sensitizing range, test macro is on the basis of initial planning path in specified position
Set the corresponding barrier of addition.The sensitizing range is that grating map has been set in initialization, with addition barrier
Position be it is one-to-one, as shown in Fig. 4 (b), that is, a certain sensitizing range HotArea has been determinedn: { (xa1,ya1),(xa2,
ya2) ..., (xan,yan), the region DynamicObstacle of corresponding addition barriern: { (xb1, yb1), (xb2,
yb2) ..., (xbn, ybn) also determine that simultaneously.
The intelligent vehicle model is the true dummy model of intelligent vehicle to be measured in a test system, dynamically instead
Reflect the driving status of intelligent vehicle to be measured, attribute has: position, speed, the information of course and barrier all around are such as schemed
Shown in 2.
The test examination question exam pool can be divided into part, global and dynamic three according to the three of grating map kinds of different types
The different types of planning aptitude tests of kind.
The sector planning aptitude tests are based on single frames local grid map M, necessary point NESS={ (xs,ys),
(xg,yg), (xs,ys) and (xg,yg) it is respectively Origin And Destination, sector planning returns to path P={ (x of plannings,ys),(x1,
y1),(x2,y2),...,(xn,yn),(xg,yg), (xn,yn) it is the coordinate of cell successively passed through.
The Global motion planning aptitude tests are global grating map M, necessary point NESS={ (xs,ys),(xt1,yt1),
(xt2,yt2),...,(xtn,ytn),(xg,yg), (xs,ys) and (xg,yg) it is respectively Origin And Destination, (xt1,yt1),(xt2,
yt2),...,(xtn,ytn) it is necessary task point, it is desirable that path P={ (x that Global motion planning returnss,ys),(x1,y1),(x2,
y2),...,(xn,yn),(xg,yg) it must sequentially pass through these task points, i.e.,(xn,yn) it is the cell successively passed through
Coordinate.
The Dynamic Programming aptitude tests are intelligent vehicles during the locally or globally planning ability of progress, work as intelligence
Can vehicle driving to a certain sensitizing range of grating map when, test macro according to initial planning path grating map specific bit
It sets and dynamically adds barrier.As shown in figure 3, according to test assignment, by starting point S to terminal E may there are two types of planning path A and
B adds dynamic disorder when test macro detects intelligent vehicle model driving to sensitizing range A on grating map immediately
Cell attribute value is changed to (x by object An,yn)=- 2 cancels setting for sensitizing range A and sensitizing range B associated with it later
It sets, cell attribute value is changed to (xn,yn)=0.The incidence relation of so-called sensitizing range is for its corresponding dynamic barrier
For, that is the corresponding dynamic disorder of sensitizing range B cannot be added by being added to the corresponding dynamic barrier A in sensitizing range
Object B, dynamic barrier A cannot be added by being added to dynamic barrier B, otherwise just can path from starting point S to terminal E.
At the same time, test macro makes it change planning in time by grating map change information by network signals intelligent vehicle to be measured
Path, to reach the weight-normality to the scene transition stroke ability for detecting intelligent vehicle to be measured in real time.Fig. 4 (a) is Dynamic Programming energy
The design flow diagram of power test.
The test process visualization formulation, it is characterised in that test macro is fed back according to intelligent vehicle to be measured
Information utilizes its motion state of intelligent vehicle model Dynamically Announce and track on grating map.Test macro will test examination question letter
Breath, including original grating map and test assignment etc. are sent to intelligent vehicle to be measured by network, and vehicle to be measured cooks up traveling
Path further controls the vehicles executing agency output actions such as steering column, throttle, braking, gear.Meanwhile these being executed
The action message of mechanism periodically feeds back to test macro, and test macro is according to the information such as speed, the course of feedback, control intelligence
Auto model carries out corresponding synchronization action.During Dynamic Programming aptitude tests, test macro is being continuously updated intelligent vehicle
While the position of model and motion profile, the obstacle information of addition is continuously updated, is dynamically shown to be measured
The path planning of intelligent vehicle is with weight-normality stroke as a result, reaching the effect of visualization of motion process.
The test result and evaluation method is test macro when intelligent vehicle to be measured completes test assignment, for
The planning path of submission and other test datas are analyzed and evaluated, and score for its reasonability and optimality.It gives below
Go out some specific quantitatively evaluating indexs:
E) path length
F) task situation is completed
Whether the motion planning path by testing vehicle passes through starting point, terminal and task point in assignment file etc., comes
Judge whether to complete planning tasks success,
G) risk of collision (Risk, R) in path
It is less than most by the way that whether the minimum distance of every in the motion planning path of calculating vehicle and peripheral obstacle meet
Small collision distance, to judge that path whether there is risk,
H) time-consuming T is planned
T=TE-TS (9)
TSIt is denoted as the test assignment data time that intelligent vehicle to be measured receives test macro, TEIt is denoted as intelligent vehicle to be measured
Test assignment is completed, sends the time that task completes label to test macro.
Several evaluation indexes, path (weight) plan that every topic total score calculation method is as follows:
wiIndicate the weight of every kind of test index, SiIndicate the score of every kind of test index.
The intelligent vehicle to be measured is to collect network communication interface, decision making function module, control function module to hold with vehicle-mounted
Row structure is in the unmanned intelligent vehicle of one.After intelligent vehicle receives test assignment, held by decision and control module control
Row mechanism response action, while the motion change result that will test feeds back to test macro, test macro is by its dynamic change
Process reaction is on the intelligent vehicle model in virtual scene.Intelligent vehicle to be measured is placed on test-bed during the test,
It rolls wheel on rotary drum, carrys out simulated automotive and travelled on road, in the process of movement, the position of car body itself is not
Change.It thus meets and completes various testing experiments in relatively narrow lab space.
Step 1, intelligent vehicle to be measured is travelled to test-bed.To plan capacity test system as server, with to be measured
Intelligent vehicle is client, takes ICP/IP protocol to be communicated between the two in local area network.
Step 2, selection test examination question, test macro read the grating map of the test question according to examination question from disk, and
Carried out in visual Simulation module display starting point, task point, terminal location information.
Step 3, virtual test auto model is constructed, as shown in Fig. 2, its attribute has: position (xn,yn), speed v, course ω
And cell attribute information all around, it adds it in the grating map in previous step, initial position is rising for task
Point, remains static.Test macro opens more new thread, receives the status information from true intelligent vehicle output to be measured,
Dynamically reflect its driving path and motion state.
Step 4, test macro, which is opened, sends thread, and grating map and mission bit stream are sent to intelligence to be measured by network
Vehicle.Timing at this time is TSMoment.
Step 5, intelligent vehicle to be measured is by itself programmed decision-making unit, the test assignment that test macro is sended over into
Row analysis, obtains the path from initial position S to final position E, if if having task point, path must be in order successively by appointing
Business point.
Step 6, planning path result is sent to test macro by network by intelligent vehicle to be measured.
Step 7, the program results that test macro is sended over according to intelligent vehicle to be measured, mark one in grating map
Planning path, i.e. (xn,yn)=1, as shown in Figure 5.During next intelligent vehicle is advanced, position is travelled according to it
It sets, updates the position of intelligent vehicle model, and by planning path labeled as the path driven to, i.e. (xn,yn)=2, such as Fig. 6
It is shown.
Step 8, intelligent vehicle to be measured by own control systems according to planning path to steering, throttle, braking and shift
Control command is sent etc. vehicle-mounted executing agency.For car body while rotary drum stage motion, the transmission thread of intelligent vehicle to be measured is every
Every the t time by the status information feedback of executing agency to test macro.
Step 9, test macro is constantly updated according to the information for returning to the executing agency come according to S and orientation ω is displaced
The position of auto model in simulated environment.Wherein displacement calculates:
Wherein VnowFor the velocity amplitude that current time return comes, VpreFor the velocity amplitude that previous moment return comes, the two is taken
Average speed, t are the time interval for receiving the two speed.
Wherein orientation ω is 360 ° of directions centered on auto model, as shown in Figure 2.
Step 10, test macro creates a queue, during intelligent vehicle model sport, records its driving process warp
The point crossed, i.e. (xn,ynThe coordinate of)=2.
Step 11, during Dynamic Programming aptitude tests, test macro during the traveling of intelligent vehicle model,
Constantly judge whether paths traversed is sensitizing range, i.e. (xn,yn)=3.Once entering sensitizing range, test macro
Deceleration or halt instruction are sent to intelligent vehicle to be measured immediately, while dynamic disorder is added in the designated position in grating map
Object, i.e., by its cell attribute value (xn,yn)=1 is changed to (xn,yn)=-, 2 and is updated in visualization view.
Step 12, after test macro completes addition dynamic barrier, cancel the sensitizing range and sensitizing range associated with it
The setting in domain, i.e. cell attribute value (xn,yn)=3 are changed to (xn,yn)=0 pays attention to the driving path recorded in step 10 queue
Cell attribute it is constant.Meanwhile updated grating map information is sent to intelligent vehicle to be measured by test macro.
Step 13, after intelligent vehicle to be measured receives the information that grating map updates, confirm that the dynamic barrier of addition is
Whether the no driving path to initial planning has an impact, if having an impact, decision package cooks up one using current location as starting point again
Item and sends back to test macro to the optimal path of terminal.
Step 14, test macro Regeneration planning path, as shown in Figure 6.Step 7 is returned to until reaching terminal.
Step 15, after test vehicle is reached home, task completion signal is sent to test macro.
Step 16, test macro receives task completion signal, and timing at this time is TEMoment.
Step 17, until previous step, intelligent vehicle to be measured complete a test examination question.Test knot of the test macro to this
Fruit is analysed and evaluated, and steps are as follows:
Step 1001, vehicle movement is obtained by the coordinate of the passing point saved in queue in step 10 according to formula (1)
Geometric locus finds out the total length of planning path.
Step 1002, according to formula (2), whether all pass through starting point, terminal by test vehicle path curves and appoint
Business point, to judge the performance of test assignment.
Step 1003, according to formula (3), by calculating and the minimum of periphery barrier at every of vehicle movement geometric locus
Whether distance is less than minimum collision distance, to judge the risk of collision coefficient of planning path.
Step 1004, according to formula (4), the time value recorded by step 4 and step 16 calculates this subtask and completes institute
The time of consumption.
Step 1005, by above-mentioned 4 steps obtain as a result, in conjunction with each distribute different weighted values, according to formula (5)
Calculate the overall assessment score of this test examination question.
Step 18, the above-mentioned intelligent detecting and evaluating algorithms to test result are to be automatically performed in test macro by software,
Score is shown on test macro interface after completing to test examination question, is referred to for tester.
Claims (6)
1. a kind of intelligent vehicle plans aptitude tests platform, it is characterised in that: include disk, planning capacity test system and to
Survey intelligent vehicle;
The disk is used for stored samples data, assignment file, Key for Reference, code of points and test result;
The planning capacity test system is the software systems tested for the planning ability of intelligent vehicle, by three portions
Divide and constitute: test examination question exam pool, test process visualization and test result and method for quantitatively evaluating;
The intelligent vehicle to be measured includes programmed decision-making unit, control unit and vehicle-mounted executing agency, and intelligent vehicle to be measured is
It plans the test object of capacity test system, is led between intelligent vehicle to be measured and planning capacity test system by network interface
Letter;The sample data being stored in disk is the grating map for planning aptitude tests, is obtained by two ways: first,
The contextual data obtained in advance by sensory perceptual system and the grating map being converted to;Second, being surveyed in planning capacity test system
Examination personnel voluntarily draw grating map according to testing requirement;
The grating map is to plan that capacity test system is used to one kind two of simulated scenario environment for planning path test
Map is tieed up, for grating map using the lower right corner as coordinate origin, the coordinate of each grid is expressed as (xn,yn), attribute value are as follows: -2, it moves
State barrier;- 1, static-obstacle thing;0, it can traffic areas;1, the planning path not driven to;2, the planning road run over
Diameter;3, sensitizing range, wherein the planning in path carries out in non-barrier region, at the same in grating map with different marks come
Identify different objects;The grating map is divided into three kinds: sector planning map, Global motion planning map and Dynamic Programming
Figure;Sector planning map size is 500px × 500px, corresponds to actual scene environment with certain proportion;Global motion planning map is
Multiple sector planning maps are spliced, i.e., larger range of scene environment, in real time more according to the traveling-position of intelligent vehicle
New Global motion planning map;Dynamic Programming map is planning capacity test system in sector planning map or the base of Global motion planning map
Barrier is dynamically added on plinth, when intelligent vehicle model driving is to a certain sensitizing range, planning capacity test system exists
Corresponding barrier is added in specified position on the basis of initial planning path, the sensitizing range is grating map initial
It has been set when change, the position with addition barrier is one-to-one, it is determined that a certain sensitizing range HotArean:
{(xa1,ya1),(xa2,ya2) ..., (xan,yan), the region DynamicObstacle of corresponding addition barriern: { (xb1,
yb1),(xb2,yb2) ..., (xbn,ybn) also determine that simultaneously.
2. a kind of intelligent vehicle according to claim 1 plans aptitude tests platform, it is characterised in that: the intelligent vehicle
Model is dummy model of the true intelligent vehicle to be measured in planning capacity test system, dynamically reflects intelligent vehicle to be measured
Driving status, attribute has: position, speed, the information of course and barrier all around.
3. a kind of intelligent vehicle according to claim 1 plans aptitude tests platform, it is characterised in that: the test
Journey visual Simulation, the information that planning capacity test system is fed back according to intelligent vehicle to be measured utilize intelligence on grating map
Energy its motion state of auto model Dynamically Announce and track, planning capacity test system will test test question information, including original grid
Lattice map and test assignment are sent to intelligent vehicle to be measured by network, and intelligent vehicle to be measured cooks up driving path, control vehicle
Steering column, throttle, braking, gear output action, meanwhile, the action message of these executing agencies is periodically fed back to
It plans capacity test system, plans speed, course information of the capacity test system according to feedback, control intelligent vehicle model carries out
Corresponding synchronization action, during Dynamic Programming aptitude tests, planning capacity test system is being continuously updated intelligent vehicle mould
While the position of type and motion profile, the obstacle information of addition is continuously updated, dynamically shows intelligence to be measured
The path planning of vehicle is with weight-normality stroke as a result, reaching the effect of visualization of motion process.
4. a kind of intelligent vehicle according to claim 1 plans aptitude tests platform, it is characterised in that: the test knot
Fruit and method for quantitatively evaluating are planning capacity test systems when intelligent vehicle to be measured completes test assignment, for the rule of submission
It draws path and test data is analyzed and evaluated, score for its reasonability and optimality, specific quantitatively evaluating index
It is as follows:
(1) path length L
(2) task situation M is completed
BSuccess is used to identify whether test vehicle is completed to advise by the movement of starting point, terminal and task point in assignment file
The task of drawing indicates completion task when being 1, indicate not completing task when being 0;
(3) the risk of collision R in path
BRisk be in the motion planning path according to vehicle every be less than minimum with whether the minimum distance of peripheral obstacle meets
Collision distance judges that path indicates there are risk of collision to indicate that risk of collision is not present when being 0 with the presence or absence of risk, when being 1;
(4) time-consuming T is planned
T=TE-TS (4)
TSIt is denoted as the test assignment data time that intelligent vehicle to be measured receives planning capacity test system, TEIt is denoted as intelligence to be measured
Vehicle completes test assignment, sends the time that task completes label to planning capacity test system.
5. a kind of intelligent vehicle according to claim 4 plans aptitude tests platform, it is characterised in that: the quantization is commented
Valence index, the every topic total score calculation method of path planning are as follows:
wiIndicate the weight of every kind of test index, SiIndicate the score of every kind of test index.
6. a kind of intelligent vehicle according to claim 1 plans aptitude tests platform, it is characterised in that: the intelligence to be measured
Energy vehicle is to integrate network communication interface, decision making function module, control function module and the vehicle-mounted unmanned intelligence for executing structure
Energy vehicle after intelligent vehicle receives test assignment, passes through decision making function module and control function module controls vehicle-mounted execution machine
Structure response action, while the motion change result that will test feeds back to planning capacity test system, plans capacity test system
By the reaction of its dynamic changing process on the intelligent vehicle model in virtual scene, intelligent vehicle to be measured is placed in during the test
It on test-bed, rolls wheel on rotary drum, carrys out simulated automotive and travelled on road, in the process of movement, car body itself
Position there is no change, satisfaction various testing experiments are completed in narrow lab space.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610939638.0A CN106644503B (en) | 2016-10-24 | 2016-10-24 | A kind of intelligent vehicle planning aptitude tests platform |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610939638.0A CN106644503B (en) | 2016-10-24 | 2016-10-24 | A kind of intelligent vehicle planning aptitude tests platform |
Publications (2)
Publication Number | Publication Date |
---|---|
CN106644503A CN106644503A (en) | 2017-05-10 |
CN106644503B true CN106644503B (en) | 2019-04-09 |
Family
ID=58821224
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201610939638.0A Active CN106644503B (en) | 2016-10-24 | 2016-10-24 | A kind of intelligent vehicle planning aptitude tests platform |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106644503B (en) |
Families Citing this family (24)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US10474790B2 (en) * | 2017-06-02 | 2019-11-12 | TuSimple | Large scale distributed simulation for realistic multiple-agent interactive environments |
CN109211575B (en) * | 2017-07-05 | 2020-11-20 | 百度在线网络技术(北京)有限公司 | Unmanned vehicle and site testing method, device and readable medium thereof |
CN107678306B (en) * | 2017-10-09 | 2021-04-16 | 驭势(上海)汽车科技有限公司 | Dynamic scene information recording and simulation playback method, device, equipment and medium |
CN107729878A (en) * | 2017-11-14 | 2018-02-23 | 智车优行科技(北京)有限公司 | Obstacle detection method and device, equipment, vehicle, program and storage medium |
CN109839925A (en) * | 2017-11-24 | 2019-06-04 | 东莞产权交易中心 | A kind of dynamic route of vehicle and speed general control strategy |
CN108088686B (en) * | 2017-12-21 | 2020-11-20 | 驭势科技(浙江)有限公司 | Test simulation method and device, computer storage medium and electronic equipment |
CN108829087B (en) * | 2018-07-19 | 2019-09-10 | 山东省科学院自动化研究所 | A kind of intelligent test system of autonomous driving vehicle |
WO2020014951A1 (en) * | 2018-07-20 | 2020-01-23 | 深圳市道通智能航空技术有限公司 | Method and apparatus for building local obstacle map, and unmanned aerial vehicle |
CN109141924A (en) * | 2018-09-12 | 2019-01-04 | 五方智能车科技有限公司 | A kind of examination and test method of intelligent vehicle basic performance |
CN109141929A (en) * | 2018-10-19 | 2019-01-04 | 重庆西部汽车试验场管理有限公司 | Intelligent network joins automobile emulation test system and method |
CN109764881B (en) * | 2018-12-07 | 2021-05-07 | 北京掌行通信息技术有限公司 | Unmanned vehicle testing method and device, electronic equipment and medium |
CN111443702A (en) * | 2018-12-29 | 2020-07-24 | 沈阳新松机器人自动化股份有限公司 | Method for verifying robot path consistency |
TWI684085B (en) * | 2019-04-24 | 2020-02-01 | 揚昇育樂事業股份有限公司 | Self-driving travel path central controlling device of self-driving car |
CN110456767A (en) * | 2019-08-02 | 2019-11-15 | 上海哲奥实业有限公司 | A kind of vehicle operating system framework based on ecological chain and industrial chain |
CN110633905B (en) * | 2019-09-06 | 2023-05-02 | 武汉理工大学 | Intelligent Che Yun platform reliability calculation method and system |
CN110595798B (en) * | 2019-09-19 | 2022-04-05 | 北京百度网讯科技有限公司 | Test method and device |
CN110673590B (en) * | 2019-10-21 | 2021-02-05 | 安徽江淮汽车集团股份有限公司 | Method, device and equipment for judging reliability of automatic driving system and storage medium |
CN112785838B (en) * | 2019-11-11 | 2023-05-30 | 千寻位置网络有限公司 | Test yard supervision system and method thereof |
CN111007834B (en) * | 2019-12-13 | 2021-05-18 | 北京经纬恒润科技股份有限公司 | Laboratory test system and method for ADAS path planning function |
CN111581102A (en) * | 2020-05-11 | 2020-08-25 | 中国人民解放军陆军研究院装甲兵研究所 | Test question bank system based on environmental data |
CN111879319B (en) * | 2020-06-29 | 2023-10-20 | 中国科学院合肥物质科学研究院 | Indoor test method and system for ground unmanned platform and computer equipment |
CN112525542B (en) * | 2020-10-26 | 2022-09-27 | 泉州装备制造研究所 | Intelligent vehicle performance test system and method thereof |
CN113418718B (en) * | 2021-05-24 | 2022-07-01 | 东风柳州汽车有限公司 | Test task monitoring system and method based on test vehicle |
CN116663939B (en) * | 2023-07-31 | 2023-10-17 | 北京理工大学 | Unmanned vehicle path planning scene and task complexity evaluation method and system |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104020674A (en) * | 2014-04-21 | 2014-09-03 | 华南农业大学 | Matlab simulation visualized platform of vehicle Bug obstacle avoidance algorithm |
CN104260722A (en) * | 2014-09-23 | 2015-01-07 | 北京理工大学 | Automatic parking system |
Family Cites Families (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US8577595B2 (en) * | 2008-07-17 | 2013-11-05 | Memsic, Inc. | Location and path-map generation data acquisition and analysis systems |
CN103234763B (en) * | 2013-04-09 | 2015-04-15 | 北京理工大学 | System and method for quantitatively evaluating unmanned vehicles |
CN103207090B (en) * | 2013-04-09 | 2016-02-24 | 北京理工大学 | A kind of automatic driving vehicle environmental simulation test macro and method of testing |
CN103335853B (en) * | 2013-07-18 | 2015-09-16 | 中国科学院自动化研究所 | A kind of automatic driving vehicle Cognitive Aptitude Test system and method |
CN105354356B (en) * | 2015-09-29 | 2019-03-29 | 中国人民解放军63892部队 | A kind of air situation fusion performance evaluation system and method based on radar information simulation |
-
2016
- 2016-10-24 CN CN201610939638.0A patent/CN106644503B/en active Active
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104020674A (en) * | 2014-04-21 | 2014-09-03 | 华南农业大学 | Matlab simulation visualized platform of vehicle Bug obstacle avoidance algorithm |
CN104260722A (en) * | 2014-09-23 | 2015-01-07 | 北京理工大学 | Automatic parking system |
Also Published As
Publication number | Publication date |
---|---|
CN106644503A (en) | 2017-05-10 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN106644503B (en) | A kind of intelligent vehicle planning aptitude tests platform | |
JP6842574B2 (en) | Systems and methods for obtaining passenger feedback in response to autonomous vehicle driving events | |
US11900738B2 (en) | Systems and methods to obtain feedback in response to autonomous vehicle failure events | |
CN109466543A (en) | Plan autokinetic movement | |
US20220048533A1 (en) | Method and system for validating autonomous control software for a self-driving vehicle | |
CN110001658A (en) | It predicts in path for vehicle | |
US11501449B2 (en) | Method for the assessment of possible trajectories | |
CN112286206B (en) | Automatic driving simulation method, system, equipment, readable storage medium and platform | |
CN110047276A (en) | The congestion status of barrier vehicle determines method, apparatus and Related product | |
US20220198107A1 (en) | Simulations for evaluating driving behaviors of autonomous vehicles | |
KR102589587B1 (en) | Dynamic model evaluation package for autonomous driving vehicles | |
EP3869341A1 (en) | Play-forward planning and control system for an autonomous vehicle | |
CN115907250A (en) | Learning-based commenter for adjusting a motion planner of an autonomously driven vehicle | |
CN117130298A (en) | Method, device and storage medium for evaluating an autopilot system | |
CN111323034A (en) | Determining a subsequent test route during a test run of a vehicle | |
EP4198804A1 (en) | Simulation obstacle vehicles with driving styles | |
JP2023548516A (en) | Methods for providing information about road users | |
CN109284889B (en) | Unmanned vehicle reaction agility evaluation method and device | |
EP3454269A1 (en) | Planning autonomous motion | |
Escuín et al. | A Long-Distance Smart Driving Service Based on Floating Car Data and Open Data | |
US20240144748A1 (en) | Systems and Methods to Obtain Feedback in Response to Autonomous Vehicle Failure Events | |
Sharma | Analyzing Crash Potential in Mixed Traffic With Autonomous and Human-Driven Vehicles | |
Detering et al. | Requirements for precise simulation models for traffic flow optimizing adas | |
EP4338053A1 (en) | Tools for performance testing autonomous vehicle planners | |
EP4338058A1 (en) | Tools for performance testing autonomous vehicle planners |
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 |