Car steering auxiliary system sensor preferred disposition method
Technical field
The present invention relates to the preferred disposition methods of car steering auxiliary system sensor, particularly point out one kind and drive for automobile
Sail screening optimizing and the method for optimizing configuration of auxiliary system environment sensing sensor.
Background technique
Environment sensing sensor is the important component of car steering auxiliary system, be mainly used to perceive traffic environment,
Detect and identify surrounding traffic participant and traffic sign etc., including millimetre-wave radar, laser radar, ultrasonic radar and vision
Sensor etc..Continuous improvement with people to car steering assistance capability requirements, more and more sensors are integrated in vapour
Che Shang, context aware systems more sophisticated, using the method that tradition superposition is integrated or relies on experience configuration easily cause structural redundancy,
The wasting of resources and system cost are increased sharply, and bring difficulty to the upgrading and optimization of automobile intelligent driving assistance system.Therefore urgent
A kind of sensor preferred disposition method is needed, automatic screening optimizing and configuration optimization are carried out to sensor needed for system, are allowed to
While meeting environmental information perception demand, optimizes the composition and arrangement of sensor, reduce system cost.
Summary of the invention
It is lengthy and tedious existing for needs and existing configuration method for car steering auxiliary system environment sensing sensor integration
The problem of not optimizing, the object of the present invention is to provide a kind of car steering auxiliary system sensor preferred disposition methods, to realize
To the complex optimum of its Choice of Sensors and arrangement.
To solve the above problems, the present invention takes following technical scheme: a kind of car steering auxiliary system sensor is preferred
Configuration method, which comprises the steps of:
Firstly, typical case scene is chosen, to the ring of perception needed for system according to car steering auxiliary system functional localization
Border information carries out Unify legislation, and range, precision, the information type perceived is needed including car steering auxiliary system;
Secondly, building car steering auxiliary system sensor is matched from the multiple dimensions for including cost, function and reliability
The indicator evaluation system set, and then establish Multidimensional Comprehensive evaluation index model;
Again, the perceptional function model for establishing alternative sensor describes sensing range, the perceived accuracy, detection of sensor
Rate and the onboard sensing range of different installation sites;
Finally, according to screening conditions and comprehensive evaluation index model, using full search algorithm, to alternative sensor and standby
It selects installation site to carry out screening optimizing, realizes the computer solving of optimal sensor location.
It further, further include after carrying out screening optimizing to alternative sensor and alternative installation site, to preferably out
Allocation plan carry out system testing and evaluation, circulation correct the step of, comprising:
Firstly, building typical traffic scene in simulation software;
Secondly, according to preferred sensor allocation plan sensors configured on automobile;
Then, emulation testing and verifying are carried out to car steering auxiliary system function;
Finally, according to verification the verifying results, circulation amendment sensor allocation plan.
Further, it is to the method for establishing sensor configuration Multidimensional Comprehensive evaluation index model:
Firstly, constructing evaluation index respectively from three function of sensor, cost, reliability dimensions;
Secondly, determining the weight coefficient of each performance indicator using Delphi method;
Again, each index is normalized;
Finally, linear weighted function summation is carried out to the evaluation index under each dimension, as final comprehensive performance evaluation index:
In formula, J (Xn) it is comprehensive evaluation index, XnThe set of sensors of representative configuration, shared n sensor, i=1,2,3
Respectively represent three kinds of function, cost, reliability indexs, JiIndex value after respectively representing three kinds of index normalization, δiIt respectively represents
The weight coefficient of three kinds of indexs can be determined according to Delphi method.
Sensor function index, including but not limited to awareness coverage accounting S (Xn) evaluation:
S(Xn)=F (Xn)/SR
Wherein, XnThe set of sensors of configuration is represented, n sensor, F (X are sharedn) represent sensor sense under the configuration
Know range, SVSensing range needed for representing system.
The sensor indicator of costs, cost price and installation cost comprising selected sensor, is defined with following formula:
Wherein, P (Xn) it is the sensor unit sheet configured, miFor i-th kind of number of sensors, m kind, Pr are sharediIt is i-th kind
Sensor unit price;LijFor the installation cost of i-th kind of sensor onboard j-th of position, l installation site is shared.
Sensor reliability index, including but not limited to using the set of sensors X of configurationnMiddle Mean work
Reliability index of the time index minimum value as sensor, is defined with following formula:
T(Xn)=min [MTTF (xi)]
Wherein T (Xn) it is the mean time between failures that sensor configures, MTTF (xi) represent the flat of i-th kind of sensor
Equal time between failures.
Further, it is to the method for establishing sensor perceptional function model:
According to the mechanism of perception and technical parameter of car steering auxiliary system sensors, it is established by different level and perceives function
Energy model, is specifically divided into following three level, therefrom chooses a kind of mode:
True value layer: not considering sensor performance, installation site of the consideration sensor on vehicle, and output transducer is visited
Target range and orientation are surveyed, geometrical model is belonged to;
Performance layer: according to sensor technology parameter, output transducer detects target range, error range and detectivity, belongs to
In parameter model;
Energy layer: consider the factors such as weather attenuation, the detection specific reflective surface area of target, establish energy reflection model, export
The energy that sensor detection target is reflected back, describes its detectivity, belongs to energy model.
Further, the method for carrying out screening optimizing to alternative sensor and its position is:
Firstly, the primary election constraint condition of setting sensor, including but not limited to, from sensor configuration assessment indicator system
It chooses one or more index limits and sets constraint condition;
Secondly, carrying out primary election to common sensor according to constraint condition, screens out and be obviously unsatisfactory for constraint condition
Sensor, and then the sensor for meeting primary election requirement is determined as alternative sensor;
Then, according to the resemblance of alternative sensor and installation requirement, determine that their onboard all are adapted to fit
Position, be alternative installation site;
Again, the alternative sensor and alternative installation site that meet primary election condition are numbered respectively, is calculated using global search
To selection algorithm before the floating of method, including but not limited to sequence, alternative sensor and its installation site are constantly selected into and are rejected, often
It is selected into an installation site on a sensor to vehicle, calculates the comprehensive performance evaluation index value J (X of the configurationn), index value increases
Add and be just selected into, index value, which is reduced, just to be rejected, and is constantly selected into and is rejected in numerical order, until selecting the comprehensive performance of allocation plan
Evaluation index is optimal;
Finally, realizing the computer solving of Optimum sensor placement scheme.
The beneficial effect that the present invention is realized is: the present invention constructs the configuration of car steering auxiliary system sensor first
Multidimensional Comprehensive evaluation model, using it as the preferred objective function of sensor and interpretational criteria;Next establishes variety classes biography
The perceptional function model of sensor, describes the performances such as the sensing range of each sensor;Finally, using full search algorithm, to alternative
Sensor, quantity and installation site carry out screening optimizing.The present invention realizes crucial sensing in car steering auxiliary system
It is auxiliary to optimize car steering while meeting environmental information perception demand for the screening optimizing of device and configuration optimization, required scheme
The composition and arrangement of auxiliary system sensor, while taking into account system reliability, reduce system cost.
Detailed description of the invention
Attached drawing is only used for showing the purpose of specific embodiment, and is not to be construed as limiting the invention, in entire attached drawing
In, identical reference symbol indicates identical component.
Fig. 1 is car steering auxiliary system sensor configuration preferred method schematic diagram;
Fig. 2 is the flow chart that auto-panorama looks around monitoring system sensor screening optimizing;
Fig. 3 is that auto-panorama looks around monitoring system sensor configuration diagram.
Specific embodiment
With reference to the accompanying drawing, and monitoring system is looked around using auto-panorama the present invention is described in detail as embodiment,
But it will be appreciated by those skilled in the art that following embodiment be not to technical solution of the present invention make unique restriction, it is all
Any equivalents done under technical solution of the present invention Spirit Essence or change are regarded as belonging to protection model of the invention
It encloses.
Auto-panorama looks around monitoring system as a kind of driving assistance system, and major function positioning is: for car steering
Member provides more intuitive auxiliary driving image information, allows driver to understand and checks vehicle-surroundings whether there are obstacles and understands
The relative bearing and distance of barrier help driver's parked vehicle or low speed by complex road condition, effectively reduce scratch, touch
The generation of accidents such as hit, subside.
Using the mentioned method of the present invention to the process of panoramic looking-around monitoring system sensor progress preferred disposition, include one by one
Following steps:
(1) Unify legislation is carried out to the environmental information perception demand of driving assistance system;
(2) sensor configuration Multidimensional Comprehensive evaluation model is established, as the preferred objective function of sensor;
(3) sensor perceptional function model is established;
(4) screening optimizing is carried out to alternative sensor and alternative installation site;
(5) test of sensing system functional simulation, evaluation and circulation are corrected.
Specific explanations are as follows:
(1) the environmental information perception demand of driving assistance system is described
It to the panoramic looking-around monitoring system that the present embodiment is mentioned, is positioned according to system function, considers two kinds of applied fields respectively
Scape: the scene that scene of parking and low speed pass through complex road condition.Vehicle needs the perception of surrounding traffic environmental information in scene of parking
It asks are as follows: 360 degree of vehicle periphery have visual pattern, convenient for observation stop line and periphery barrier, can detect at least 3 meters of periphery
Obstacle distance.In the scene that low speed passes through complex road condition, road driveway width is generally 3.5 meters, and driver is desirable to see
The range-azimuth information of surrounding traffic participant, to avoid scratching or colliding;To the perception demand of surrounding traffic environmental information
Are as follows: 360 degree of vehicle periphery have visual pattern, and front can detect the barrier (front truck or pedestrian) at least 5 meters, simultaneously
Arranged on left and right sides can also detect the vehicles or pedestrians close to two sides lane line, and detection range should be greater than vehicle outside to lane line
Distance.
(2) the Multidimensional Comprehensive evaluation model of driving assistance system sensor configuration is established
1) firstly, according to vehicle to the perception demand of environmental information, from sensing system function, cost and reliability three
Dimension establishes evaluation index respectively.
In the configuration of the present embodiment sensor, functional parameter selects vision coverage area accounting S1(Xn) and ranging covering model
Enclose accounting S2(Xn) evaluate respectively, it is defined with lower two formula:
S1(Xn)=F1(Xn)/SV
S2(Xn)=F2(Xn)/SD
Wherein, XnThe set of sensors of configuration is represented, n sensor, F are shared1(Xn) represent configuration visual sensor it is total
Investigative range, F2(Xn) represent the total investigative range of distance measuring sensor configured, SVVisual detection range, S needed for representing systemD
Ranging investigative range needed for representing system.
The indicator of costs includes the cost price and installation cost of selected sensor, and wherein installation cost is equipped with installation position
It closes, is defined with following formula:
Wherein, P (Xn) it is that sensor configures totle drilling cost, miFor i-th kind of number of sensors, m kind, Pr are sharediIt is passed for i-th kind
Sensor unit price;LijFor the installation cost of i-th kind of sensor onboard j-th of position, l installation site is shared.
Reliability index is used to evaluate the mass property of selected sensor, when primary evaluation index has Mean work
Between MTTF, average time between failures MTBF etc., the present embodiment selects the mean time between failures MTTF conduct of sensor
The reliability evaluation index of single-sensor, the set of sensors X of selection and deploymentnIn the minimum value of the index match as sensor
The reliability index set, is defined with following formula:
T(Xn)=min [MTTF (xi)]
Wherein T (Xn) it is the mean time between failures that the sensor configures, MTTF (xi) represent i-th of sensor
Mean time between failures, i=1,2 ... n.
2) secondly, above-mentioned each index is normalized, the expression formula for having dimension is transformed to nondimensional expression
Formula, and data are mapped to processing in (- 1,1) range, establish the evaluation index J under each dimensioni, i=1,2,3, represent function,
Three kinds of indexs of cost and reliability, and summation, as comprehensive evaluation index are weighted to each evaluation index:
In formula, δ1、δ2、δ3The respectively weight coefficient of function, cost and reliability index can be determined according to Delphi method;
J1、J2、J3Index value after respectively representing three system function, cost and reliability index normalization.
(3) sensor perceptional function model is established
According to the common ultrasonic radar of car steering auxiliary system, millimetre-wave radar, laser radar, monocular vision, double
The mechanism of perception and technical parameter of the sensors such as visual feel, fisheye camera are established the geometrical model for describing its investigative range and are retouched
State detection accuracy/resolution ratio performance parameter model.
Wherein, each single-sensor investigative range geometrical model is the function about detection range d and view angle theta:
S=f (d, θ)=π θ d2/360
For the sake of simplification, the performances such as detection accuracy of sensor are directly indicated with sensor technology parameter in the present embodiment.It is more
Perceptional function after a sensor installation onboard contains the performances such as total investigative range, average detection precision, wherein total detection
Range with the investigative range of selected single sensor in addition to having outside the Pass, and also related with sensor mounting location, installation site is adjacent
Sensor, investigative range overlapping region is redundancy section, and total investigative range indicates with following formula:
In formula, StotalIndicate the total investigative range of k sensor, SiIndicate the investigative range of i-th of sensor of configuration,
SoverlapIt indicates that sensor detects overlapping range, can be acquired according to the relative distance geometrical relationship of adjacent sensors.
(4) screening optimizing is carried out to alternative sensor and alternative installation site
1) firstly, the constraint condition of setting sensor primary election.According to market survey and existing similar product is compared, positions institute
The product of design is the panoramic looking-around monitoring system of low cost, thereby determines that system maximum cost price is no more than Y member for constraint
Condition.In the step other than setting system cost is as constraint condition, also setting sensor can be required according to product technology most
The performance constraints such as low detection range, detection accuracy, lowest resolution, to screen out the sensing for obviously not meeting constraint condition
Device.Constraint condition can be selects one or several indexs as constraint from multidimensional evaluation index.
2) secondly, carrying out primary election to sensors according to constraint condition, rejecting is obviously unsatisfactory for constraint condition (cost)
Such as laser radar, millimetre-wave radar, binocular vision, low-resolution cameras sensor;And then the biography of primary election condition will be met
Sensor is determined as alternative sensor;
3) then, according to the resemblance of alternative sensor and installation requirement, the position that they are onboard adapted to fit is determined
It sets and (has multiple), be alternative installation site;
4) again, alternative sensor and alternative installation site are numbered respectively, it is right to selection algorithm before being floated using sequence
Alternative sensor and its installation site are selected into and are rejected.
Sensor is selected into criterion: in the set of sensors X being just selected intokOn the basis of (having k), then it is alternative from other
Some sensor x is selected into sensor (shared n)iOn to vehicle in an alternative installation site, if being selected into xiSensor collection afterwards
Close Xk+1Comprehensive evaluation index value J (Xk+1) increase, then the sensor can be selected into, that is, meet formula:
J(Xk+1)=J (Xk+xi)>J(Xk)
Sensor rejects criterion: from the set of sensors X of primary electionkMiddle some sensor of rejecting xrIf comprehensive after rejecting
Close evaluation index value J (Xk-1) increase, then the sensor is rejected, that is, meet formula:
J(Xk-1)=J (Xk-xr)>J(Xk)
Alternative sensor and its installation site are selected into embodiment and to reject process as shown in Figure 2:
1. being first selected into visual sensor, sensor, one sensor of every increase, meter are selected into alternative installation site
Calculate a J (Xk), while increase with vision coverage area, cost accordingly also be will increase, until discovery J (Xk+1) it is maximum when
Until, it does not continue to be added;As it is possible that there are redundancies, so also needing gradually to reject from the set of sensors selected again
Sensor observes comprehensive evaluation index value, until index value J (X after rejecting some (or certain) sensork-1) be up to
Only, set of sensors X at this timek-1It is exactly allocation optimum.
2. and then equally radar sensor is selected into, radar sensor is selected into alternative installation site, it is every to increase by one
A sensor calculates a J (Xk), while increase with ranging coverage area, cost response also increases, until discovery J
(Xk+1) it is maximum when until, do not continue to be added;Then sensor is rejected in set of sensors from having selected again, make the J after rejecting
(Xk-1) maximum.
The selection course of various kinds of sensors is the same, is here that expression selects two class sensors respectively.
3. realizing the computer of optimal sensor configuration finally, sequence forward floating selection algorithm programs in Matlab
It solves.
(5) system function emulation testing, evaluation and the circulation after screening optimizing are corrected
In order to verify the reasonability of required sensor allocation plan, emulation testing is carried out by PreScan simulation software.It is first
Park scene and congestion road conditions scene are first built in emulation platform;Secondly according to the sensor allocation plan preferably gone out in automobile
Upper sensors configured;Then it selects common control method operation vehicle to be parked or by congestion road conditions, test automobile is complete
Scape looks around monitoring function;Finally according to verification the verifying results, circulation amendment sensor allocation plan advanced optimizes allocation plan.
In the present embodiment, e.g., alternative visual sensor has tetra- kinds of C1, C2, C3, C4, alternative radar sensor have U1,
Tetra- kinds of U2, U3, U4, alternative installation position is equipped with 30 points of vehicle's surroundings.Can be acquired using the mentioned method of the present invention: it is inexpensive about
Under the conditions of beam, the sensor being selected into be 4 numbers be C4, the fisheye camera of position 3,9,18,27 and 12 numbers are U3, position
It is set to 1,2,4,5,6,14,16,17,19,20,22,30 ultrasonic sensor, installation site is as shown in Figure 3.Required result with
Existing product allocation plan is consistent, and cost reduces, it was demonstrated that the feasibility and validity of mentioned method.