CN106291736A - Pilotless automobile track dynamic disorder object detecting method - Google Patents

Abstract

This application discloses a kind of pilotless automobile track dynamic disorder object detecting method, this method carries out detecting and tracking by analyzing and processing 64 line laser data to the dynamic barrier of pilotless automobile surrounding, for the sector region that pilotless automobile front accuracy requirement is higher, Confidence distance theory is used to merge 64 line laser data processing of information and the movement state information of 4 line laser sensors output, improve the Detection accuracy to barrier kinestate significantly, then the result obtained according to fusion carries out time delay correction to the position of moving obstacle, finally take at barrier and on grid map, position difference occupied by position occupied by dynamic barrier and static-obstacle thing is indicated.This method is possible not only in outdoor environment detect barrier movable information exactly, and can eliminate sensing data and process the dynamic barrier position deviation that time delay is brought.

Description

Pilotless automobile track dynamic disorder object detecting method

Technical field

The application relates to a kind of pilotless automobile track dynamic disorder object detecting method, belongs to automatic Pilot and machine regards Feel field.

Background technology

Along with the economic and development of society, automobile quantity increase thus some social problems of causing are the most prominent Go out, the traffic in such as city, the safety of vehicle traveling, the supply of the energy, environmental pollution etc..The social problem Dou Yuan of these reality Contradiction between existing traffic infrastructure and this carrier of automobile, this is not only embodied on traffic jam issue, simultaneously It is also embodied in the environmental pollution freely not caused due to traffic and the condition of road surface relatively fallen behind and advanced vehicle technology to people Life, potential safety hazard produced by property.Now due to the personnel that vehicle accident causes get in society with property loss Coming the most serious, relate generally to the collision of vehicle in vehicle accident, therefore improving the safety of vehicle during traffic travels is to hand over now Logical development there is problem to be solved.The unpiloted research of automobile can be alleviated the generation of this problem to a great extent and carry Gone out new thinking, in Intelligent Vehicle System for track early warning and monitoring, the fatigue detecting of driver, automatic cruise, The research of unmanned grade can alleviate driver's workload under steam, has for improving the safety of existing traffic system The biggest help.

Along with the continuous progress of science and technology, also obtain based on the intelligent control technology of computer technology and automatic technology Obtained development at full speed, thus as this field unmanned intelligent vehicle research oneself through become the problem paid close attention to the most it One.But the progress of pilotless automobile depends on the development of sensor technology, current single sensor is at outdoor complex environment In there is the phenomenon that lane obstructions analyte detection accuracy rate is the highest, and the inaccuracy of lane obstructions analyte detection will directly influence The driving safety of pilotless automobile.

Summary of the invention

It is an object of the invention to provide a kind of pilotless automobile track dynamic disorder object detecting method, to overcome laser Sensor detects the data that dynamic barrier run in outdoor environment and processes that to there is time delay, testing result accuracy rate the most high Problem.

For achieving the above object, the present invention provides following technical scheme:

The embodiment of the present application open a kind of pilotless automobile track dynamic disorder object detecting method, including:

S1, by 64 line laser sensors the barrier of motor vehicle environment carried out detecting and tracking, obtain barrier the most in the same time Hinder thing grid map, and from this barrier grid map, obtain static-obstacle thing grid map and dynamic barrier list；

S2, by the dynamic barrier information in 4 line laser sensor acquisition vehicle front regions；

S3, the dynamic barrier information in 64 line laser sensor dynamic barrier lists and 4 line laser sensors are obtained The dynamic barrier information taken carries out simultaneously match；

S4, the theoretical data to the 64 line laser sensors that the match is successful and 4 line laser sensors of employing Confidence distance are entered Row merges；

S5, the position of moving obstacle is carried out time delay correction according to merging the result that obtains, finally take at barrier On grid map, position difference occupied by position occupied by dynamic barrier and static-obstacle thing is indicated.

Preferably, in above-mentioned pilotless automobile track dynamic disorder object detecting method, in described step s1, use Maximin height map method carries out rasterizing process to the data of 64 line laser sensors.

Preferably, in above-mentioned pilotless automobile track dynamic disorder object detecting method, in described step s1, use The grid that takies in map is clustered by region growing clustering algorithm, after cluster is tracked barrier.

Preferably, in above-mentioned pilotless automobile track dynamic disorder object detecting method, cluster and tracking bag Include: stored by dynamic barrier list and cluster the barrier block message obtained, and these barrier blocks of real-time update Follow the tracks of result.

Preferably, in above-mentioned pilotless automobile track dynamic disorder object detecting method, it is stored in described dynamic barrier Hinder each barrier block in thing list including at least following information: numbering, up-to-date time time once cluster obtains, occupy Position, velocity magnitude direction and acceleration magnitude direction, velocity covariance, acceleration covariance and there is confidence level and motion Confidence level.

Preferably, in above-mentioned pilotless automobile track dynamic disorder object detecting method, in described step s3, synchronize Matching process includes: the dynamic barrier information of 4 line laser sensor outputs is to be indicated with the form of box one by one, The parameter of each box includes center O, and (x, y) and the size direction of speed v, by the center O of box, (x, y) towards its speed The opposite direction translation distance s in degree direction, the size of s meets following formula:

S=λ (t₆₄-t₄)·v

Wherein t₆₄And t₄Being the time-consuming of acquisition process 64 laser sensor data and 4 line laser sensing datas respectively, λ is Parameter.

Preferably, in above-mentioned pilotless automobile track dynamic disorder object detecting method, in described step s3, synchronize Matched processing method includes:

There is no the box of the barrier block successful match therewith that 64 line laser sensors detect, do not make any process；

The barrier block not having 64 line laser sensors of box successful match therewith to detect, does not make any process, still Use the testing result of 64 line laser sensors as final result；

The barrier block having 64 line laser sensors of box successful match therewith to detect, merges 64 line laser sensors Final kinestate is obtained with 4 line laser sensing datas.

Preferably, in above-mentioned pilotless automobile track dynamic disorder object detecting method, in described step s4, merge Method includes:

Assume x₁And x₂Be respectively the 64 calculated kinestates of line laser sensing data and 4 line laser sensors defeated The kinestate gone out, they all Gaussian distributed, certain measures the data that they obtain is x respectively₁And x₂, then theirs is general Rate density function is shown in formula:

$p_1\left(x\right|x_1)=\frac{1}{\sqrt{2\mathrm{\π}}{\mathrm{\σ}}_1}\exp \[-\frac{1}{2}{\left(\frac{x-x_1}{{\mathrm{\σ}}_1}\right)}^2\]$

$p_1\left(x\right|x_2)=\frac{1}{\sqrt{2\mathrm{\π}}{\mathrm{\σ}}_2}\exp \[-\frac{1}{2}{\left(\frac{x-x_2}{{\mathrm{\σ}}_2}\right)}^2\]$

x_iTo x_jConfidence distance d_ijMeet following formula:

$d_{ij}=2{\∫}_{x_i}^{x_j}{p}_i\left(x\right|x_i)dx$

Obtain the Confidence distance matrix on 2 rank:

$D_2=\left[\begin{array}{cc}{d}_{11}& d_{12}\\ d_{21}& d_{22}\end{array}\right]$

Assume that Confidence distance Critical Matrices is

$T_2=\left[\begin{array}{cc}{t}_{11}& t_{12}\\ t_{21}& t_{22}\end{array}\right]$

Obtain the relational matrix on 2 rank:

$\begin{array}{c}{r}_{ij}=\left\{\begin{array}{c}1,d_{ij}\≤t_{ij}\\ 0,d_{ij}>t_{ij}\end{array}\right.\\ R_2=\left[\begin{array}{cc}{r}_{11}& r_{12}\\ r_{21}& r_{22}\end{array}\right]\end{array}$

The sensing data supporting number of probes to be 2 satisfied output merges according to the following formula, final barrier The kinestate X of block, wherein l is to meet output to support that number of probes is the number of probes of 2:

$X=\frac{\underset{k=1}{\overset{l}{\Σ}}\frac{x_k}{{\mathrm{\σ}}_k^2}}{\underset{k=1}{\overset{l}{\Σ}}\frac{1}{{\mathrm{\σ}}_k^2}}.$

Compared with prior art, it is an advantage of the current invention that: the inventive method is possible not only in outdoor environment exactly Detect barrier movable information, and can eliminating sensing data, to process the dynamic disorder object location that time delay brought inclined Difference, takies grid map by the sound state obstacle information barrier in environment more accurately and is described.

Accompanying drawing explanation

In order to be illustrated more clearly that the embodiment of the present application or technical scheme of the prior art, below will be to embodiment or existing In having technology to describe, the required accompanying drawing used is briefly described, it should be apparent that, the accompanying drawing in describing below is only this Some embodiments described in application, for those of ordinary skill in the art, on the premise of not paying creative work, Other accompanying drawing can also be obtained according to these accompanying drawings.

Fig. 1 show the flow chart of dynamic disorder object detecting method in the specific embodiment of the invention；

Fig. 2 show the detection range of 64 line laser sensors and 4 line laser sensors in the specific embodiment of the invention and shows It is intended to；

Fig. 3 show in the specific embodiment of the invention and swashs 64 lines of a crossroad by maximin height map method Barrier obtained by after optical sensor data rasterizing takies grid map；

Fig. 4 show 4 line laser sensor output dynamic barrier information schematic diagram in the specific embodiment of the invention；

Fig. 5 show 64 line laser sensing datas and four line laser sensing datas in the specific embodiment of the invention and synchronizes Schematic diagram；

Fig. 6 show 64 line laser sensing datas and 4 line laser sensing datas coupling in the specific embodiment of the invention Schematic diagram；

Fig. 7 show dynamic barrier position correction schematic diagram in the specific embodiment of the invention.

Detailed description of the invention

The embodiment of the present application discloses a kind of pilotless automobile track dynamic disorder object detecting method, including: 64 laser Sensor, 4 line laser sensors, message processing module, actuator.Its method includes: by 64 laser sensors and 4 line lasers The obstacle information that the information fusion that two kinds of laser sensors of sensor detect obtains carries out coupling checking, if error is setting In fixed accuracy rating, then send control instruction to pilotless automobile actuator, perform thermoacoustic prime engine subsystem, for root Generate control instruction according to the auxiliary information collected, control described unmanned automobile and perform corresponding operation.

Further, the obstacle information that 64 line laser sensors will be merged and four line laser sensors detect, including The information such as barrier shape, position, speed, improve detection of obstacles precision.

The application comprehensively uses two kinds of sensors to improve the degree of accuracy of lane obstructions analyte detection, uses 64 line laser sensings Device carries out detecting and tracking to the barrier around pilotless automobile, utilizes Kalman filter to enter the kinestate of barrier Line trace and prediction, for the sector region that pilotless automobile front accuracy requirement is higher, use Confidence distance theory to melt Conjunction and four line laser sensing datas determine the information of barrier, finally, the information that two kinds of laser sensors detect are melted Close the obstacle information obtained.

Below in conjunction with the accompanying drawing in the embodiment of the present invention, the technical scheme in the embodiment of the present invention is carried out detailed retouching State, it is clear that described embodiment is only a part of embodiment of the present invention rather than whole embodiments.Based on the present invention In embodiment, the every other enforcement that those of ordinary skill in the art are obtained on the premise of not making creative work Example, broadly falls into the scope of protection of the invention.

Shown in Fig. 1 and Fig. 2,64 line laser sensing datas are carried out rasterizing and processes and obtain a barrier and take Grid map, carries out cluster and follows the tracks of the multidate information that can obtain barrier, by dynamic barrier grid map the most in the same time Deleting from grid map and be stored in dynamic barrier list, this deletes dynamic barrier and occupies the grid map of information also It is exactly a static-obstacle thing grid map, then the dynamic barrier information in dynamic barrier list and 4 line lasers is sensed Dynamic barrier information in the pilotless automobile front region that device obtains carries out synchronous fusion and obtains a new dynamic barrier Hinder thing list, finally the dynamic barrier in this new list is merged in static-obstacle thing grid map and obtains a sound The grid map that the difference of state barrier indicates.

Use maximin height map method that 64 line laser sensing datas are carried out rasterizing process.At maximin In height map, ground is modeled as a series of grid, and these grids only comprise two values: all project in same grid The maximum of laser sensor return value and minima.Then by the difference of maximum and minima more than threshold value set in advance.D Grid tag be barrier state；It is non-barrier state by the difference grid tag less than D.For a grid X, if its For barrier state, then occupation value T (X)=1 of this grid is set；If non-barrier state, then it is set to T (X)=0. Fig. 3 is to barrier obtained after 64 line laser sensing data rasterizings of a crossroad by maximin height map method Thing is hindered to take grid map.

It is tracked barrier needing the grid that takies in map to cluster before, uses a kind of region growing to gather Class algorithm, is tracked barrier after cluster, first creates a dynamic barrier list and stores what cluster above obtained Barrier block message, and the tracking result of these barrier blocks of real-time update.It is stored in this dynamic barrier list Each barrier block comprises following information: numbering, up-to-date once cluster time when obtaining, plant oneself, velocity magnitude side To with acceleration magnitude direction, velocity covariance, acceleration covariance and there is confidence level and motion confidence level.

Sensing data simultaneously match, because the dynamic barrier letter in four line laser sensors energy directly output environments Breath, and 64 line laser sensing datas just can obtain dynamic barrier information in environment through a series of process above, two The data acquisition and processing (DAP) of person is the most different, so firstly the need of the obstacle information of both simultaneously matchs.4 line laser sensors The dynamic barrier information of output is to be indicated with the form of box one by one, and the parameter of each box includes length of side a, b, in The size direction of heart position and speed v, as shown in Figure 4, by the center O of box, (x, y) opposite direction towards its velocity attitude is put down Move distance s, as it is shown in figure 5, the size of s meets following formula:

S=λ (t₆₄-t₄)·v

Wherein t₆₄And t₄Being the time-consuming of acquisition process 64 laser sensor data and 4 line laser sensing datas respectively, λ is Parameter.

After synchronously completing, it is possible to the barrier block message of two sensors of coupling, there is the barrier that region is overlapping with certain box Hindering thing block, i.e. with this box successful match, so result of coupling also has following 3 kinds:

1, there is no the box of the barrier block successful match therewith that 64 line laser sensors detect, do not make any process

2, there is no the barrier block that 64 line laser sensors of box successful match therewith detect, such as the OB3 in Fig. 5, yet Do not make any process, still use the testing result of 64 line laser sensors as final result

3, there is the barrier block that 64 line laser sensors of box successful match therewith detect, such as the OB in Fig. 6₁And OB₂, They are required for merging 64 line laser sensors and 4 line laser sensing datas obtain final kinestate.

Data Fusion of Sensor, after simultaneously match completes, uses Confidence distance theoretical to two sensors that the match is successful Data merge, it is assumed that x₁And x₂It is the 64 calculated kinestates of line laser sensing data and 4 line lasers sensing respectively The kinestate of device output, they all Gaussian distributed, certain measures the data that they obtain is x respectively₁And x₂, then they Probability density function see formula:

$\begin{array}{c}{p}_1\left(x\right|x_1)=\frac{1}{\sqrt{2\mathrm{\π}}{\mathrm{\σ}}_1}\exp \[-\frac{1}{2}{\left(\frac{x-x_1}{{\mathrm{\σ}}_1}\right)}^2\]\\ p_1\left(x\right|x_2)=\frac{1}{\sqrt{2\mathrm{\π}}{\mathrm{\σ}}_2}\exp \[-\frac{1}{2}{\left(\frac{x-x_2}{{\mathrm{\σ}}_2}\right)}^2\]\end{array}$

x_iTo x_jConfidence distance d_ijMeet following formula:

$d_{ij}=2{\∫}_{x_i}^{x_j}{p}_i\left(x\right|x_i)dx$

Then the Confidence distance matrix on 2 rank can be obtained:

$D_2=\left[\begin{array}{cc}{d}_{11}& d_{12}\\ d_{21}& d_{22}\end{array}\right]$

Assume that Confidence distance Critical Matrices is

$T_2=\left[\begin{array}{cc}{t}_{11}& t_{12}\\ t_{21}& t_{22}\end{array}\right]$

The relational matrix on 2 rank then can be obtained by above formula:

$\begin{array}{c}{r}_{ij}=\left\{\begin{array}{c}1,d_{ij}\≤t_{ij}\\ 0,d_{ij}>t_{ij}\end{array}\right.\\ R_2=\left[\begin{array}{cc}{r}_{11}& r_{12}\\ r_{21}& r_{22}\end{array}\right]\end{array}$

The sensing data finally supporting number of probes to be 2 satisfied output merges according to the following formula, final must hinder Hindering the kinestate X of thing block, wherein l is to meet output to support that number of probes is the number of probes of 2.

$X=\frac{\underset{k=1}{\overset{l}{\Σ}}\frac{x_k}{{\mathrm{\σ}}_k^2}}{\underset{k=1}{\overset{l}{\Σ}}\frac{1}{{\mathrm{\σ}}_k^2}}$

After having merged, it is contemplated that 64 line laser sensor data acquisitions process time-consuming, planting oneself of dynamic barrier During this period of time having actually occurred change, needing plants oneself to these is modified.Correcting mode is, by dynamic barrier In list those motion confidence levels more than barrier block all that value is set above plant oneself towards the translation of its velocity attitude away from From S ', as it is shown in fig. 7, wherein v ' is the barrier block movement velocity after merging.These are through fusion, revised obstacle Thing block message is the most more newly stored in dynamic barrier list.

S'=λ ' t_v·v'

It should be noted that in this article, term " includes ", " comprising " or its any other variant are intended to non-row Comprising of his property, so that include that the process of a series of key element, method, article or equipment not only include those key elements, and And also include other key elements being not expressly set out, or also include intrinsic for this process, method, article or equipment Key element.In the case of there is no more restriction, statement " including ... " key element limited, it is not excluded that including State and the process of key element, method, article or equipment there is also other identical element.

The above is only the detailed description of the invention of the application, it is noted that for the ordinary skill people of the art For Yuan, on the premise of without departing from the application principle, it is also possible to make some improvements and modifications, these improvements and modifications also should It is considered as the protection domain of the application.

Claims (8)

1. a pilotless automobile track dynamic disorder object detecting method, it is characterised in that including:

S1, by 64 line laser sensors the barrier of motor vehicle environment carried out detecting and tracking, obtain barrier the most in the same time Grid map, and from this barrier grid map, obtain static-obstacle thing grid map and dynamic barrier list；

S2, by the dynamic barrier information in 4 line laser sensor acquisition vehicle front regions；

S3, the dynamic barrier information in 64 line laser sensor dynamic barrier lists and 4 line laser sensors are obtained Dynamic barrier information carries out simultaneously match；

S4, the theoretical data to the 64 line laser sensors that the match is successful and 4 line laser sensors of employing Confidence distance are melted Close；

S5, the position of moving obstacle is carried out time delay correction according to merging the result that obtains, finally take grid at barrier On Tu, position difference occupied by position occupied by dynamic barrier and static-obstacle thing is indicated.

Pilotless automobile track the most according to claim 1 dynamic disorder object detecting method, it is characterised in that: described step In rapid s1, use maximin height map method that the data of 64 line laser sensors are carried out rasterizing process.

Pilotless automobile track the most according to claim 1 dynamic disorder object detecting method, it is characterised in that: described step In rapid s1, use region growing clustering algorithm that the grid that takies in map is clustered, after cluster, barrier is tracked.

Pilotless automobile track the most according to claim 3 dynamic disorder object detecting method, it is characterised in that: cluster and Tracking includes: store the barrier block message that obtains of cluster by dynamic barrier list, and real-time update these The tracking result of barrier block.

Pilotless automobile track the most according to claim 4 dynamic disorder object detecting method, it is characterised in that: it is stored in Each barrier block in described dynamic barrier list includes at least following information: numbering, and up-to-date once cluster is when obtaining Time, plant oneself, velocity magnitude direction and acceleration magnitude direction, velocity covariance, acceleration covariance and existence Confidence level and motion confidence level.

Pilotless automobile track the most according to claim 1 dynamic disorder object detecting method, it is characterised in that: described step In rapid s3, simultaneously match method includes: the dynamic barrier information of 4 line laser sensor outputs is the form with box one by one Being indicated, the parameter of each box includes center O, and (x, y) and the size direction of speed v, by the center O of box (x, y) towards the opposite direction translation distance s of its velocity attitude, the size of s meets following formula:

S=λ (t₆₄-t₄)·v

Wherein t₆₄And t₄Being the time-consuming of acquisition process 64 laser sensor data and 4 line laser sensing datas respectively, λ is ginseng Number.

Pilotless automobile track the most according to claim 6 dynamic disorder object detecting method, it is characterised in that: described step In rapid s3, simultaneously match processing method includes:

There is no the box of the barrier block successful match therewith that 64 line laser sensors detect, do not make any process；

The barrier block not having 64 line laser sensors of box successful match therewith to detect, does not make any process, still uses The testing result of 64 line laser sensors is as final result；

The barrier block having 64 line laser sensors of box successful match therewith to detect, merges 64 line laser sensors and 4 lines Laser sensor data obtain final kinestate.

Pilotless automobile track the most according to claim 1 dynamic disorder object detecting method, it is characterised in that: described step In rapid s4, fusion method includes:

Assume x₁And x₂It is the 64 calculated kinestates of line laser sensing data and the output of 4 line laser sensors respectively Kinestate, they all Gaussian distributed, certain measures the data that they obtain is x respectively₁And x₂, then their probability is close Degree function is shown in formula:

$p_1\left(x\right|x_1)=\frac{1}{\sqrt{2\mathrm{\π}}{\mathrm{\σ}}_1}\exp \[-\frac{1}{2}{\left(\frac{x-x_1}{{\mathrm{\σ}}_1}\right)}^2\]$

$p_1\left(x\right|x_2)=\frac{1}{\sqrt{2\mathrm{\π}}{\mathrm{\σ}}_2}\exp \[-\frac{1}{2}{\left(\frac{x-x_2}{{\mathrm{\σ}}_2}\right)}^2\]$

x_iTo x_jConfidence distance d_ijMeet following formula:

$d_{ij}=2{\∫}_{x_i}^{x_j}{p}_i\left(x\right|x_i)dx$

Obtain the Confidence distance matrix on 2 rank:

$D_2=\left[\begin{array}{cc}{d}_{11}& d_{12}\\ d_{21}& d_{22}\end{array}\right]$

Assume that Confidence distance Critical Matrices is

$T_2=\left[\begin{array}{cc}{t}_{11}& t_{12}\\ t_{21}& t_{22}\end{array}\right]$

Obtain the relational matrix on 2 rank:

$\begin{array}{c}{r}_{ij}=\left\{\begin{array}{cc}1,& d_{ij}\≤t_{ij}\\ 0,& d_{ij}>t_{ij}\end{array}\right.\\ R_2=\left[\begin{array}{cc}{r}_{11}& r_{12}\\ r_{21}& r_{22}\end{array}\right]\end{array}$

The sensing data supporting number of probes to be 2 satisfied output merges according to the following formula, final barrier block Kinestate X, wherein l is to meet output to support that number of probes is the number of probes of 2:

$X=\frac{\underset{k=1}{\overset{l}{\Σ}}\frac{x_k}{{\mathrm{\σ}}_k^2}}{\underset{k=1}{\overset{l}{\Σ}}\frac{1}{{\mathrm{\σ}}_k^2}}.$