CN110175344A - A kind of laser radar harness distribution adjusting and optimizing method for automatic Pilot scene - Google Patents
A kind of laser radar harness distribution adjusting and optimizing method for automatic Pilot scene Download PDFInfo
- Publication number
- CN110175344A CN110175344A CN201910218582.3A CN201910218582A CN110175344A CN 110175344 A CN110175344 A CN 110175344A CN 201910218582 A CN201910218582 A CN 201910218582A CN 110175344 A CN110175344 A CN 110175344A
- Authority
- CN
- China
- Prior art keywords
- harness
- laser radar
- laser
- distribution
- optimization
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000009826 distribution Methods 0.000 title claims abstract description 66
- 238000000034 method Methods 0.000 title claims abstract description 35
- 238000005457 optimization Methods 0.000 claims abstract description 48
- 238000001514 detection method Methods 0.000 claims abstract description 21
- 238000012360 testing method Methods 0.000 claims abstract description 10
- 230000006870 function Effects 0.000 claims description 68
- 238000011156 evaluation Methods 0.000 claims description 35
- 230000008447 perception Effects 0.000 claims description 7
- 230000008569 process Effects 0.000 claims description 7
- 238000009434 installation Methods 0.000 claims description 6
- 238000004364 calculation method Methods 0.000 claims description 5
- 230000015572 biosynthetic process Effects 0.000 claims description 4
- 230000005540 biological transmission Effects 0.000 claims description 3
- 238000001914 filtration Methods 0.000 claims description 3
- 239000002699 waste material Substances 0.000 claims description 3
- 239000000571 coke Substances 0.000 claims 1
- 239000004744 fabric Substances 0.000 claims 1
- 230000000694 effects Effects 0.000 abstract description 10
- 230000001737 promoting effect Effects 0.000 abstract description 2
- 230000003287 optical effect Effects 0.000 description 6
- 230000000007 visual effect Effects 0.000 description 4
- 230000008859 change Effects 0.000 description 3
- 230000004927 fusion Effects 0.000 description 3
- 239000000203 mixture Substances 0.000 description 3
- 230000006872 improvement Effects 0.000 description 2
- 238000004519 manufacturing process Methods 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 239000012141 concentrate Substances 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000003760 hair shine Effects 0.000 description 1
- 238000012804 iterative process Methods 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 238000007500 overflow downdraw method Methods 0.000 description 1
- 238000010408 sweeping Methods 0.000 description 1
- 238000003786 synthesis reaction Methods 0.000 description 1
- 238000010200 validation analysis Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2111/00—Details relating to CAD techniques
- G06F2111/06—Multi-objective optimisation, e.g. Pareto optimisation using simulated annealing [SA], ant colony algorithms or genetic algorithms [GA]
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computer Hardware Design (AREA)
- Evolutionary Computation (AREA)
- Geometry (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Optical Radar Systems And Details Thereof (AREA)
- Traffic Control Systems (AREA)
Abstract
The present invention relates to laser radar sensor and automatic Pilot fields, are distributed adjusting and optimizing method more particularly, to a kind of laser radar harness for automatic Pilot scene.The present invention according to specified optimization harness quantity, establishes laser radar harness model first.Later, according to the requirement of environment sensing task, Optimal Parameters is set, are slightly optimized.Finally, thin optimization tuning is carried out to thick optimum results, the laser harness distribution after being optimized.Obtained optimization laser harness distribution can test and verify it in the virtual environments such as driving simulator in the effect promoting of environment sensing task.Compared with the existing common uniform harness distribution of multi-line laser radar, this method has specific aim to specific object detection task, so that the sensor after optimization has in target detection, Detection accuracy is high, the big feature of range.
Description
Technical field
The present invention relates to laser radar sensor and automatic Pilot fields, are directed to automatic Pilot more particularly, to one kind
The laser radar harness of scene is distributed adjusting and optimizing method.
Background technique
In recent years, unmanned technology is increasingly becoming the research hotspot of domestic and international colleges and universities and enterprise, also because it is commercially being answered
Cause public concern with exposing in test again and again.Unmanned vehicle is during realizing automatic Pilot, the letter of autonomous driving system
Breath perception subsystem is the key foundation of unmanned vehicle automatic Pilot, is to guarantee unmanned vehicle safety and stability, collisionless normally travel
Premise.
As vehicle advises the raising of the precision and integrated level of detecting sensor, more and more sensor deployment applications in recent years
On pilotless automobile platform, wherein laser radar sensor is because of its high-precision distance measurement, information it is rich, and
The high reliability insensitive to light change that vehicle-mounted camera has is compared, is increasingly becoming on unmanned automatic driving automobile main
Sensor.
There are many researchs to concentrate on the point cloud data using laser radar sensor, realizes and complete automatic Pilot dependent field
Environment sensing task in scape, the tasks such as detection, classification, tracking including object target.Due to sweeping for laser radar sensor
It retouches with distance into divergent trend, the point cloud scanned at medium and long distance can be mutually sparse, further influences target detection etc.
The effect of other algorithms, therefore how synthesis improvement laser radar is still one in the scanning of different distance and algorithm perceived effect
Open major issue.
In the prior art, the patent of 102837658 B of Patent No. CN proposes more laser thunders on a kind of intelligent vehicle
Up to data fusion system and method, this method is scanned based on the method for multilasered optical radar data fusion, by increasing laser
The number of sensors of radar improves perceived effect, but the economic cost of laser radar is very high, and merges multilasered optical radar number
According to the data volume and complexity for increasing laser radar data processing, processing difficulty is increased.
In the prior art, the patent of 109375238 A of Patent No. CN proposes a kind of omnidirectional changeable rotation stepping angle
Laser radar and its scan method, this method using it is variable rotation stepping angle laser radar scanning method be scanned, can
To increase the laser point quantity being incident upon on target object in the horizontal direction, but it is incident upon target object in vertical direction
On laser point quantity do not get a promotion, fail the vertical distribution information for further enriching target object, therefore appoint in perception
Promotion effect in business is limited.
Summary of the invention
The present invention in order to overcome at least one of the drawbacks of the prior art described above, provides a kind of for automatic Pilot scene
Laser radar harness is distributed adjusting and optimizing method, has specific aim to specific object detection task, so that the sensing after optimization
Device has Detection accuracy height, the big feature of range in target detection.
In order to solve the above technical problems, the technical solution adopted by the present invention is that: a kind of laser for automatic Pilot scene
Radar harness is distributed adjusting and optimizing method, comprising the following steps:
S1. construct laser radar sensor harness model: according to the rotational structure of multi-line laser radar sensor, definition swashs
The horizontal scan corner variable of optical radar rotation;According to the structure of multi-line laser radar vertical direction multi-emitter, definition transmitting
The variable of channel vertical elevation distribution;
S2. for the harness distribution objectives under different distance, optimization aim evaluation function is constructed: first against short distance,
Vertical visual field as big as possible in order to obtain establishes the specific item scalar functions of harness diverging, to promote the target in the short distance visual field
Detection effect.The specific item offer of tender of harness focusing is established for the detection under medium and long distance for focusing effect as big as possible
Number, to promote the target detection effect in the remote visual field.For the specific installation site of laser radar, the son for avoiding blind area is established
Objective function shines to avoid laser harness and falls in blind area;
S3. according to perception task, adjusting and optimizing parameter is slightly optimized: according to the perception of practical automatic Pilot scene environment
Object detection task sets the weight of the focus and each specific item scalar functions in interest region, is slightly optimized;It is thick excellent
Change first through the thick optimization resolution ratio of setting, generates candidate angle value.For each candidate angular distribution combination, angle is calculated
The evaluation of estimate that Degree distributions are answered, and then iteration chooses the optimal angle combinations of overall merit, as thick optimum results;
S4. according to slightly optimizing as a result, carrying out thin optimization tuning, the radar harness optimized is distributed;Thin optimization is first
The step-length of optimization is set, then for the result slightly optimized as initial estimation, carries out subtle adjusting and optimizing carefully to optimize step-length, obtains
Optimal result is evaluated to the end;
S5. emulation testing and verifying are carried out to the laser radar after optimization.Emulation survey is carried out to the laser radar after optimization
Examination and verifying.Since existing volume production laser radar sensor is unable to adjust vertical direction light launch angle, customization line in the market
The laser radar period of beam angular distribution is long, at high cost, therefore this method proposes in automatic Pilot simulated environment, after optimization
Laser radar carry out emulation testing and verifying.Laser radar sensor mould is realized in emulation first in automatic Pilot simulated environment
Type, and virtual radar is customized according to the harness distribution results of optimization, common Driving Scene is then set, in Same Scene
Under to after optimization and the laser radar of control carries out data acquisition and test of heuristics, to be swashed according to arithmetic result comparison optimization
The performance boost of optical radar.
In the present invention, start with from laser radar sensor, it is rich to improve the point cloud data information that laser radar scanning obtains
Fu Xing, the specific perception task in automatic Pilot scene obtain effect promoting.The present invention is by adjusting the single laser thunder of optimization
Vertical laser harness up to sensor is distributed, and is avoided the problem that cost is too high in the scan method of more radar fusions, is improved
The utilization rate of single laser radar laser harness, and for the scan method of variable rotation stepping angle, it can be one
The laser point number being incident upon on target object is effectively improved within the scope of set a distance, is further had in subsequent detection algorithm
Preferably performance.
Further, according to the horizontal rotation structure and vertical direction of multi-line laser radar sensor in the S1 step
The structure of multi-emitter defines the horizontal scan corner variable of laser radar rotation and the change of transmission channel vertical elevation distribution
Amount, detailed process include:
Laser radar sensor is equipped with motor and rotates in horizontal plane to obtain 360 degree of horizontal view angle;In equal gap
In rotation, laser radar emits laser and carries out single pass, the rotation angle series of scanning is defined as:
Rotation={ θi, i=0,1 ..., m }
Wherein m is to rotate horizontally total step number, θiFor i-th of feathering angle, determined by the angular resolution of level acquisition:
Due to the limitation of motion structure, laser radar configures different laser emission channels, the angle of departure in vertical direction
Degree distribution are as follows:
Wherein, n is vertical port number, βjFor j-th of vertical laser beam channel angle, different angle combinations compositions
The harness of laser radar is distributed c.
Further, optimization object function is constructed in the S2 step to specifically include:
S21. overall assessment function is made of a multiple objective function, is defined as:
WhereinFor total optimization aim evaluation function, FkIt (c) is sub-goal evaluation function, γkFor weight shared by it, c*
To optimize obtained harness distribution results;Overall merit is added up by the sub-goal evaluation function of several Weights and is obtained;
S22. it is as follows to define specific item scalar functions for the diversity that harness is distributed in considering short-range target detection:
Wherein, h is the mounting height of laser radar sensor,C is distributed to the harness of ROI in interest regional scope
Diversity evaluated, and different weighted value δ (μ) is assigned in different distance μ, the weight value function definition of the diversity
Are as follows:
μfocus_iFor the focus of setting, siFor the corresponding weighting ratio coefficient of focus, μ of the δ (μ) in settingfocus_iIt closes
Weight is minimum at focus, maximizes the diversity for paying close attention to the distribution of the harness in certain specified regions to control;
Dis (μ, c, h) is the evaluation function for evaluating laser harness distribution c diversity, is defined as:
Wherein I (μ, c, h) function statistics and filtering are at distance to a declared goal μ, the laser harness being in contact with μ vertical plane,
Dis (μ, c, h) function then calculates the very poor Range of these harness, and according to maximum value maxRange (I (umin, c, h)) into
Row normalization obtains diversity evaluation of the harness distribution c at μ;
S23. the focusing that harness is distributed in the target detection of long range under consideration defines second sub- objective function such as
Under:
Wherein, h is the mounting height of laser radar sensor,C is distributed to the harness of ROI in interest regional scope
Focusing evaluated, and different weighted value ρ (μ) is assigned in different distance μ, the weight value function definition of the focusing
Are as follows:
It is similar with for the weighted value function of harness diversity, μ of the ρ (μ) in settingfocus_iWeighted value is most at focus
It is small, the focusing for paying close attention to the distribution of the harness in certain specified regions is maximized to control;
Con (μ, c, h) is the evaluation function for evaluating laser harness distribution c focusing, is defined as:
The function counts the number of beams Count contacted with vertical plane at distance μ, and according to maximum value
maxCount(I(umax, c, h)) it is normalized, obtain focusing evaluation of the harness distribution c at μ;
S24. in addition to considering that harness is distributed in the diversity and focusing of some regions, near installation site, due to flat
The limitation of platform, the laser harness of transmitting may be fallen on the blind area of mounting platform formation, at roof, therefore to avoid laser
The considerations of harness, which is fallen in, to be caused to waste in blind area, and the limitation of laser harness blind area is added in this method, specific item scalar functions are defined as follows:
The objective functionCumulative laser beam falls in the punishment weight in blind zone BZ, wherein weighting function
Blind (μ, c, h (μ)) is defined as follows:
The function is formed by falling in the imparting of the light beam in blind zone additivity penalty value, and penalty value is by as lower aprons limit
Polynomial equation calculates:
Wherein, S is the proportionality coefficient of the weight of setting, and n is the polynomial order of approximate limitation.
Further, the thick Optimum distribution passes through the thick optimization resolution ratio of setting first, using resolution ratio as step-length, from
The angular distribution upper bound of setting generates candidate angle value set to angular distribution lower bound;During thick Optimized Iterative, in candidate
In angle value set, to specify the quantity of laser radar line number to choose angle value as a candidate combinations;For each time
The angular distribution of choosing combines, and calculates the corresponding evaluation of estimate of angular distributionAnd then iteration chooses the optimal angle group of overall merit
Close c*, as thick optimum results.
Further, the S4 step specifically includes: being distributed c for laser beam angle obtained in the previous step*, according to
Specified thin optimization resolution ratio, to c*In angle value be adjusted, substitute intoCalculation Estimation value changes according to evaluation of estimate adjustment
For direction, optimum results c is finally obtained*。
Compared with prior art, beneficial effect is:
1. the present invention is based on the distributions of its laser harness of single laser radar sensor adjusting and optimizing, compared to multilasered optical radar
Sensor fusion method has many advantages, such as that at low cost, structure is simple, point cloud data amount is small without merging point cloud data;
2. the present invention, can be in a spacing by adjusting the laser harness distribution of optimization laser radar sensor vertical direction
It from the laser point quantity for being incident upon target object is improved in range, is more conducive to further extracting its feature, promotes target
Detect the performance of scheduling algorithm;
3. the present invention is distributed by the laser harness of Multipurpose Optimal Method adjusting and optimizing laser radar sensor, to short distance
Vertical visual field as big as possible is obtained from the sub-goal function for establishing harness diverging, the specific item of harness focusing is established to medium and long distance
Scalar functions obtain focusing effect as well as possible, and the specific item offer of tender for avoiding blind area is established to the specific installation site of laser radar
Number, avoids laser harness from falling in blind area.
Detailed description of the invention
Fig. 1 is laser radar harness distribution adjusting and optimizing method integrated stand composition of the present invention for automatic Pilot scene.
Fig. 2 is thick optimization process basic flow chart of the invention.
Fig. 3 is thin optimization process basic flow chart of the invention.
Specific embodiment
Attached drawing only for illustration, is not considered as limiting the invention;In order to better illustrate this embodiment, attached
Scheme certain components to have omission, zoom in or out, does not represent the size of actual product;To those skilled in the art,
The omitting of some known structures and their instructions in the attached drawings are understandable.Being given for example only property of positional relationship is described in attached drawing
Illustrate, is not considered as limiting the invention.
Embodiment 1:
As shown in Figure 1, a kind of laser radar harness for automatic Pilot scene is distributed adjusting and optimizing method, including following
Step:
Step 1. constructs laser radar sensor harness model;
Its detailed process includes:
Laser radar sensor is equipped with motor and rotates in horizontal plane to obtain 360 degree of horizontal view angle;In equal gap
In rotation, laser radar emits laser and carries out single pass, the rotation angle series of scanning is defined as:
Rotation={ θi, i=0,1 ..., m }
Wherein m is to rotate horizontally total step number, θiFor i-th of feathering angle;It is determined by the angular resolution of level acquisition:
Due to the limitation of motion structure, laser radar configures different laser emission channels, the angle of departure in vertical direction
Degree distribution are as follows:
Wherein, n is vertical port number, βjFor j-th of vertical laser beam channel angle, different angle combinations compositions
The harness of laser radar is distributed c.
Step 2. is directed to the harness distribution objectives under different distance, constructs optimization aim evaluation function;
S21. overall assessment function is made of a multiple objective function, is defined as:
WhereinFor total optimization aim evaluation function, FkIt (c) is sub-goal evaluation function, γkFor weight shared by it, c*
To optimize obtained harness distribution results;Overall merit is added up by the sub-goal evaluation function of several Weights and is obtained;
S22. it is as follows to define specific item scalar functions for the diversity that harness is distributed in considering short-range target detection:
Wherein, h is the mounting height of laser radar sensor,C is distributed to the harness of ROI in interest regional scope
Diversity evaluated, and different weighted value δ (μ) is assigned in different distance μ, the weight value function definition of the diversity
Are as follows:
μfocus_iFor the focus of setting, siFor the corresponding weighting ratio coefficient of focus, μ of the δ (μ) in settingfocus_iIt closes
Weight is minimum at focus, maximizes the diversity for paying close attention to the distribution of the harness in certain specified regions to control;
Dis (μ, c, h) is the evaluation function for evaluating laser harness distribution c diversity, is defined as:
Wherein I (μ, c, h) function statistics and filtering are at distance to a declared goal μ, the laser harness being in contact with μ vertical plane,
Dis (μ, c, h) function then calculates the very poor Range of these harness, and according to maximum value maxRange (I (umin, c, h)) simultaneously
It is normalized, obtains diversity evaluation of the harness distribution c at μ;
S23. the focusing that harness is distributed in the target detection of long range under consideration defines second sub- objective function such as
Under:
Wherein, h is the mounting height of laser radar sensor,C is distributed to the harness of ROI in interest regional scope
Focusing evaluated, and different weighted value ρ (μ) is assigned in different distance μ, the weight value function definition of the focusing
Are as follows:
It is similar with for the weighted value function of harness diversity, μ of the ρ (μ) in settingfocus_iWeighted value is most at focus
It is small, the focusing for paying close attention to the distribution of the harness in certain specified regions is maximized to control;
Con (μ, c, h) is the evaluation function for evaluating laser harness distribution c focusing, is defined as:
The function counts the number of beams Count contacted with vertical plane at distance μ, and according to maximum value
maxCount(I(umax, c, h)) it is normalized, obtain focusing evaluation of the harness distribution c at μ;
S24. in addition to considering that harness is distributed in the diversity and focusing of some regions, near installation site, due to flat
The limitation of platform, the laser harness of transmitting may be fallen on the blind area of mounting platform formation, at roof, therefore to avoid laser
The considerations of harness, which is fallen in, to be caused to waste in blind area, and the limitation of laser harness blind area is added in this method, specific item scalar functions are defined as follows:
The objective functionCumulative laser beam falls in the punishment weight in blind zone BZ, wherein weighting function
Blind (μ, c, h (μ)) is defined as follows:
The function is formed by falling in the imparting of the light beam in blind zone additivity penalty value, and penalty value is by as lower aprons limit
Polynomial equation calculates:
Wherein, S is the proportionality coefficient of the weight of setting, and n is the polynomial order of approximate limitation.
Step 3. is slightly optimized according to perception task, adjusting and optimizing parameter: optimize resolution ratio by the way that setting is thick first, with
Resolution ratio is as step-length, from the angular distribution upper bound of setting to angular distribution lower bound, generates candidate angle value set;Slightly optimizing
In iterative process, in candidate angle value set, to specify the quantity of laser radar line number to choose angle value as a candidate
Combination;For each candidate angular distribution combination, the corresponding evaluation of estimate of angular distribution is calculatedAnd then iteration is chosen always
Evaluate optimal angle combinations c*, as thick optimum results;
As shown in Fig. 2, for optimization process basic flow chart thick in the present invention, specific steps are as follows:
1. setting thick Optimal Parameters, wherein the candidate angle threshold value upper bound is 0 °, and lower bound is 24 °, and the thick resolution ratio that optimizes is 1 °;
2. generating candidate angular distribution set according to specified radar harness quantity 16, set sizes are about
7.4x105;
3. setting optimizing evaluation parameter, wherein setting radar mounting height h=2.0, interest region is ROI=[5,50],
Closely focus the weight γ of sub-goaldisA focus is set as μ in=1.0, δ (μ)focus_1=5.0, coefficient of correspondence s1
=20.0.The weight γ of medium and long distance diverging sub-goalconA focus is set as μ in=5.0, ρ (μ)focus_1=5.0,
Coefficient of correspondence is s1=20.0;
4. iteration chooses beam angulation distribution, Calculation Estimation value;
5. exporting highest evaluation of estimate and its corresponding beam angulation distribution being used as thick optimum results.
According to slightly optimizing as a result, carrying out thin optimization tuning, the radar harness optimized is distributed step 4.;Thin optimization is first
The step-length of optimization is first set, then for the result slightly optimized as initial estimation, carries out subtle adjusting and optimizing carefully to optimize step-length,
It obtains and evaluates optimal result to the end;C is distributed for laser beam angle obtained in the previous step*, differentiated according to specified thin optimization
Rate, to c*In angle value be adjusted, substitute intoCalculation Estimation value adjusts iteration direction according to evaluation of estimate, finally obtains
Optimum results c*;
As shown in figure 3, for the thin optimization process basic flow chart of the present invention, concrete operation step are as follows:
1. thin Optimal Parameters are set, wherein carefully optimization resolution ratio is 0.01 °;
2. being passed to thick optimum results as initial estimation;
3. adjusting the angle the angle value in distribution, Calculation Estimation value carefully to optimize resolution ratio as step-length;
4. changing according to evaluation of estimate in step 3, adjusting and optimizing direction, angular distribution is readjusted, carries out step 3, until
Evaluation of estimate converges on optimal;
5. repeat step 3 and step 4 for each of angular distribution angle value, finally obtains and optimal swash
The distribution of optical radar beam angulation.
Laser radar after step 5. pair optimization carries out emulation testing and verifying.Laser radar after optimization is emulated
Test and validation.Since existing volume production laser radar sensor is unable to adjust vertical direction light launch angle in the market, customization
The laser radar period of beam angulation distribution is long, at high cost, therefore this method proposes in automatic Pilot simulated environment, to optimization
Laser radar afterwards carries out emulation testing and verifying.Emulation realizes laser radar sensor in automatic Pilot simulated environment first
Model, and virtual radar is customized according to the harness distribution results of optimization, common Driving Scene is then set, in same field
Data acquisition and test of heuristics are carried out with the laser radar of control to after optimization under scape, is optimized to be compared according to arithmetic result
The performance boost of laser radar.
Obviously, the above embodiment of the present invention be only to clearly illustrate example of the present invention, and not be pair
The restriction of embodiments of the present invention.For those of ordinary skill in the art, may be used also on the basis of the above description
To make other variations or changes in different ways.There is no necessity and possibility to exhaust all the enbodiments.It is all this
Made any modifications, equivalent replacements, and improvements etc., should be included in the claims in the present invention within the spirit and principle of invention
Protection scope within.
Claims (5)
1. a kind of laser radar harness for automatic Pilot scene is distributed adjusting and optimizing method, which is characterized in that including following
Step:
S1. it constructs laser radar sensor harness model: according to the rotational structure of multi-line laser radar sensor, defining laser thunder
Up to the horizontal scan corner variable of rotation;According to the structure of multi-line laser radar vertical direction multi-emitter, transmission channel is defined
The variable of vertical elevation distribution;
S2. it for the harness distribution objectives under different distance, constructs optimization aim evaluation function: being directed to short distance, establish harness
The specific item scalar functions of diverging;For the detection under medium and long distance, the specific item scalar functions of harness focusing are established;For laser radar spy
Fixed installation site establishes the specific item scalar functions for avoiding blind area;
S3. according to perception task, adjusting and optimizing parameter is slightly optimized: the target perceived according to practical automatic Pilot scene environment
Detection task sets the weight of the focus and each specific item scalar functions in interest region, is slightly optimized;
S4. according to slightly optimizing as a result, carrying out thin optimization tuning, the radar harness optimized is distributed;Thin optimization is set first
The step-length of optimization, then for the result slightly optimized as initial estimation, carry out subtle adjusting and optimizing carefully to optimize step-length, obtain most
After evaluate optimal result;
S5. emulation testing and verifying are carried out to the laser radar after optimization.
2. a kind of laser radar harness for automatic Pilot scene according to claim 1 is distributed adjusting and optimizing method,
It is characterized in that, according to the horizontal rotation structure of multi-line laser radar sensor and vertical direction multi-emitting in the S1 step
The structure of device defines the horizontal scan corner variable of laser radar rotation and the variable of transmission channel vertical elevation distribution, tool
Body process includes:
Laser radar sensor is equipped with motor and rotates in horizontal plane to obtain 360 degree of horizontal view angle;In the rotation of equal gap
In, laser radar emits laser and carries out single pass, the rotation angle series of scanning is defined as:
Rotation={ θi, i=0,1 ..., m }
Wherein m is to rotate horizontally total step number, θiFor i-th of feathering angle;It is determined by the angular resolution of level acquisition:
Due to the limitation of motion structure, laser radar configures different laser emission channels, launch angle point in vertical direction
Cloth are as follows:
Wherein, n is vertical port number, βjFor j-th of vertical laser beam channel angle, different angle combinations form laser
The harness of radar is distributed c.
3. a kind of laser radar harness for automatic Pilot scene according to claim 2 is distributed adjusting and optimizing method,
It is specifically included it is characterized in that, constructing optimization object function in the S2 step:
S21. overall assessment function is made of a multiple objective function, is defined as:
WhereinFor total optimization aim evaluation function, FkIt (c) is sub-goal evaluation function, γkFor weight shared by it, c*For optimization
Obtained harness distribution results;Overall merit is added up by the sub-goal evaluation function of several Weights and is obtained;
S22. it is as follows to define specific item scalar functions for the diversity that harness is distributed in considering short-range target detection:
Wherein, h is the mounting height of laser radar sensor,To the hair of the harness distribution c of ROI in interest regional scope
Scattered property is evaluated, and different weighted value δ (μ), the weight value function of the diversity are assigned in different distance μ is defined as:
μfocus_iFor the focus of setting, siFor the corresponding weighting ratio coefficient of focus, μ of the δ (μ) in settingfocus_iConcern
Weight is minimum at point, maximizes the diversity for paying close attention to the distribution of the harness in certain specified regions to control;
Dis (μ, c, h) is the evaluation function for evaluating laser harness distribution c diversity, is defined as:
Wherein I (μ, c, h) function statistics and filtering are at distance to a declared goal μ, the laser harness being in contact with μ vertical plane, Dis
(μ, c, h) function then calculates the very poor Range of these harness, and according to maximum value maxRange (I (umin, c, h)) returned
One changes, and obtains diversity evaluation of the harness distribution c at μ;
S23. it is as follows to define second sub- objective function for the focusing that harness is distributed in the target detection of long range under consideration:
Wherein, h is the mounting height of laser radar sensor,The poly- of c is distributed to the harness of ROI in interest regional scope
Coke is evaluated, and different weighted value ρ (μ), the weight value function of the focusing are assigned in different distance μ is defined as:
It is similar with for the weighted value function of harness diversity, μ of the ρ (μ) in settingfocus_iWeighted value is minimum at focus, with control
System maximizes the focusing for paying close attention to the distribution of the harness in certain specified regions;
Con (μ, c, h) is the evaluation function for evaluating laser harness distribution c focusing, is defined as:
The function counts the number of beams Count contacted with vertical plane at distance μ, and according to maximum value maxCount (I
(uman, c, h)), and be normalized, obtain focusing evaluation of the harness distribution c at μ;
S24. in addition to considering that harness is distributed in the diversity and focusing of some regions, near installation site, due to platform
Limitation, the laser harness of transmitting may fall on the blind area of mounting platform formation, make to avoid laser harness from falling in blind area
At waste, the considerations of laser harness blind area limits is added, specific item scalar functions are defined as follows:
The objective functionCumulative laser beam falls in the punishment weight in blind zone BZ, wherein weighting function Blind
(μ, c, h (μ)) is defined as follows:
The function is formed by falling in the imparting of the light beam in blind zone additivity penalty value, and penalty value is limited multinomial by such as lower aprons
Formula formula calculates:
Wherein, S is the proportionality coefficient of the weight of setting, and n is the polynomial order of approximate limitation.
4. a kind of laser radar harness for automatic Pilot scene according to claim 3 is distributed adjusting and optimizing method,
It is characterized in that, the thick Optimum distribution passes through the thick optimization resolution ratio of setting first, using resolution ratio as step-length, from setting
The angular distribution upper bound generates candidate angle value set to angular distribution lower bound;During thick Optimized Iterative, in candidate angle value
In set, to specify the quantity of laser radar line number to choose angle value as a candidate combinations;For each candidate angle
Distributed combination is spent, the corresponding evaluation of estimate of angular distribution is calculatedAnd then iteration chooses the optimal angle combinations c of overall merit*, make
For thick optimum results.
5. a kind of laser radar harness for automatic Pilot scene according to claim 4 is distributed adjusting and optimizing method,
It is characterized in that, the S4 step specifically includes: being distributed c for laser beam angle obtained in the previous step*, according to specified
Thin optimization resolution ratio, to c*In angle value be adjusted, substitute intoCalculation Estimation value adjusts iteration direction according to evaluation of estimate,
Finally obtain optimum results c*。
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910218582.3A CN110175344B (en) | 2019-03-21 | 2019-03-21 | Laser radar wire harness distribution adjustment optimization method for automatic driving scene |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910218582.3A CN110175344B (en) | 2019-03-21 | 2019-03-21 | Laser radar wire harness distribution adjustment optimization method for automatic driving scene |
Publications (2)
Publication Number | Publication Date |
---|---|
CN110175344A true CN110175344A (en) | 2019-08-27 |
CN110175344B CN110175344B (en) | 2020-11-03 |
Family
ID=67689221
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910218582.3A Active CN110175344B (en) | 2019-03-21 | 2019-03-21 | Laser radar wire harness distribution adjustment optimization method for automatic driving scene |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110175344B (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113468735A (en) * | 2021-06-24 | 2021-10-01 | 国汽(北京)智能网联汽车研究院有限公司 | Laser radar simulation method, device and system and storage medium |
Citations (13)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20110190972A1 (en) * | 2010-02-02 | 2011-08-04 | Gm Global Technology Operations, Inc. | Grid unlock |
US20140300758A1 (en) * | 2013-04-04 | 2014-10-09 | Bao Tran | Video processing systems and methods |
CN104615821A (en) * | 2015-01-30 | 2015-05-13 | 长春工业大学 | Automobile wire harness crosstalk pre-estimation method |
CN105158756A (en) * | 2015-08-27 | 2015-12-16 | 电子科技大学 | Centralized MIMO radar radio frequency stealth multi-target tracking wave beam pointing method |
CN106093934A (en) * | 2016-08-26 | 2016-11-09 | 电子科技大学 | Multiple target location estimation method after through-wall radar imaging based on improvement dynamic programming |
US20170124781A1 (en) * | 2015-11-04 | 2017-05-04 | Zoox, Inc. | Calibration for autonomous vehicle operation |
CN107942310A (en) * | 2017-10-25 | 2018-04-20 | 中国人民解放军信息工程大学 | The resource joint optimization method of distributed MIMO radar system multiple target location estimation |
CN108304605A (en) * | 2017-11-09 | 2018-07-20 | 清华大学 | Car steering auxiliary system sensor preferred disposition method |
CN108415032A (en) * | 2018-03-05 | 2018-08-17 | 中山大学 | A kind of point cloud semanteme map constructing method based on deep learning and laser radar |
CN108490419A (en) * | 2018-06-04 | 2018-09-04 | 电子科技大学 | A kind of vehicle-mounted multi-line laser radar system of automatic Pilot |
CN108562894A (en) * | 2018-04-19 | 2018-09-21 | 电子科技大学 | Radar beam is directed toward the distribution method with transmission power |
CN108899941A (en) * | 2018-07-13 | 2018-11-27 | 中国电力科学研究院有限公司 | A kind of offshore wind farm multi-objective reactive optimization configuration method and system that soft direct sending goes out |
CN108995537A (en) * | 2018-06-04 | 2018-12-14 | 浙江吉利新能源商用车有限公司 | A kind of full-vehicle control device driven for Vehicular intelligent |
-
2019
- 2019-03-21 CN CN201910218582.3A patent/CN110175344B/en active Active
Patent Citations (13)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20110190972A1 (en) * | 2010-02-02 | 2011-08-04 | Gm Global Technology Operations, Inc. | Grid unlock |
US20140300758A1 (en) * | 2013-04-04 | 2014-10-09 | Bao Tran | Video processing systems and methods |
CN104615821A (en) * | 2015-01-30 | 2015-05-13 | 长春工业大学 | Automobile wire harness crosstalk pre-estimation method |
CN105158756A (en) * | 2015-08-27 | 2015-12-16 | 电子科技大学 | Centralized MIMO radar radio frequency stealth multi-target tracking wave beam pointing method |
US20170124781A1 (en) * | 2015-11-04 | 2017-05-04 | Zoox, Inc. | Calibration for autonomous vehicle operation |
CN106093934A (en) * | 2016-08-26 | 2016-11-09 | 电子科技大学 | Multiple target location estimation method after through-wall radar imaging based on improvement dynamic programming |
CN107942310A (en) * | 2017-10-25 | 2018-04-20 | 中国人民解放军信息工程大学 | The resource joint optimization method of distributed MIMO radar system multiple target location estimation |
CN108304605A (en) * | 2017-11-09 | 2018-07-20 | 清华大学 | Car steering auxiliary system sensor preferred disposition method |
CN108415032A (en) * | 2018-03-05 | 2018-08-17 | 中山大学 | A kind of point cloud semanteme map constructing method based on deep learning and laser radar |
CN108562894A (en) * | 2018-04-19 | 2018-09-21 | 电子科技大学 | Radar beam is directed toward the distribution method with transmission power |
CN108490419A (en) * | 2018-06-04 | 2018-09-04 | 电子科技大学 | A kind of vehicle-mounted multi-line laser radar system of automatic Pilot |
CN108995537A (en) * | 2018-06-04 | 2018-12-14 | 浙江吉利新能源商用车有限公司 | A kind of full-vehicle control device driven for Vehicular intelligent |
CN108899941A (en) * | 2018-07-13 | 2018-11-27 | 中国电力科学研究院有限公司 | A kind of offshore wind farm multi-objective reactive optimization configuration method and system that soft direct sending goes out |
Non-Patent Citations (3)
Title |
---|
RUI WEN等: "A SMART CABLE MANAGEMENT SYSTEM IN SUPPORT OF THE SMART CITY", 《CIRED WORKSHOP》 * |
YOUNG-JIN CHA等: "Multi-objective genetic algorithms for cost-effective distributions of actuators and sensors in large structures", 《EXPERT SYSTEMS WITH APPLICATIONS》 * |
董士军: "车载激光雷达系统设计与实验研究", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113468735A (en) * | 2021-06-24 | 2021-10-01 | 国汽(北京)智能网联汽车研究院有限公司 | Laser radar simulation method, device and system and storage medium |
CN113468735B (en) * | 2021-06-24 | 2024-03-22 | 国汽(北京)智能网联汽车研究院有限公司 | Laser radar simulation method, device, system and storage medium |
Also Published As
Publication number | Publication date |
---|---|
CN110175344B (en) | 2020-11-03 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN111671352B (en) | Automatic door induction triggering area adjusting method | |
CN110888125B (en) | Automatic driving vehicle positioning method based on millimeter wave radar | |
CN104541181B (en) | Radar device | |
CN112433934B (en) | Simulation test method, simulation test device, computer equipment and storage medium | |
CN109557522A (en) | Multi-beam laser scanner | |
CN108008408B (en) | Search and track imaging method, apparatus and system | |
CN105842698B (en) | A kind of rapid simulation method of High Resolution SAR image | |
CN107533127A (en) | The apparatus and method of improved laser scanning | |
CN102667495A (en) | Methods, devices and systems for remote wind sensing | |
CN105115862A (en) | Cloud particle detection method and cloud particle detector | |
CN109343032A (en) | A kind of laser thunder emission system based on diffraction beam splitting element | |
CN107561554A (en) | Inversion method with multi-wavelength laser radar data is counted based on solar luminosity | |
CN103917892A (en) | Method for measuring spatial points | |
CN109785632A (en) | A kind of magnitude of traffic flow statistical method and device | |
CN110175344A (en) | A kind of laser radar harness distribution adjusting and optimizing method for automatic Pilot scene | |
CN110083915A (en) | A kind of high frequency region Radar Target Scatter central forward automation modeling method | |
CN110537018A (en) | AI device, laser radar apparatus and wind power plant's control system | |
CN111624623B (en) | Wind field inversion method based on laser radar non-uniform scanning | |
WO2016187405A1 (en) | Systems and methods for predicting arrival of wind event | |
CN116563466A (en) | Deep learning-based three-dimensional Shan Mudian cloud completion method | |
CN110502978A (en) | A kind of laser radar waveform Modulation recognition method based on BP neural network model | |
Manivasagam et al. | Towards zero domain gap: A comprehensive study of realistic lidar simulation for autonomy testing | |
CN117571056B (en) | Environment protection monitoring method and system based on Internet of things | |
DE102014216368B4 (en) | MULTI-ARM LIDAR SYSTEM | |
CN109543553A (en) | The photoelectricity recognition and tracking method of low small slow target based on machine learning |
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 |