CN107301654A - A kind of positioning immediately of the high accuracy of multisensor is with building drawing method - Google Patents
A kind of positioning immediately of the high accuracy of multisensor is with building drawing method Download PDFInfo
- Publication number
- CN107301654A CN107301654A CN201710437709.1A CN201710437709A CN107301654A CN 107301654 A CN107301654 A CN 107301654A CN 201710437709 A CN201710437709 A CN 201710437709A CN 107301654 A CN107301654 A CN 107301654A
- Authority
- CN
- China
- Prior art keywords
- point
- laser radar
- estimation
- camera
- frame
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/20—Analysis of motion
- G06T7/246—Analysis of motion using feature-based methods, e.g. the tracking of corners or segments
- G06T7/248—Analysis of motion using feature-based methods, e.g. the tracking of corners or segments involving reference images or patches
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T17/00—Three dimensional [3D] modelling, e.g. data description of 3D objects
- G06T17/10—Constructive solid geometry [CSG] using solid primitives, e.g. cylinders, cubes
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/70—Determining position or orientation of objects or cameras
- G06T7/73—Determining position or orientation of objects or cameras using feature-based methods
- G06T7/74—Determining position or orientation of objects or cameras using feature-based methods involving reference images or patches
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2200/00—Indexing scheme for image data processing or generation, in general
- G06T2200/08—Indexing scheme for image data processing or generation, in general involving all processing steps from image acquisition to 3D model generation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10016—Video; Image sequence
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10024—Color image
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Geometry (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Computer Graphics (AREA)
- Software Systems (AREA)
- Multimedia (AREA)
- Length Measuring Devices By Optical Means (AREA)
- Image Processing (AREA)
- Image Analysis (AREA)
Abstract
The invention provides a kind of positioning immediately of the high accuracy of multisensor with building drawing method, i.e., high-precision real-time three-dimensional structure is carried out to surrounding environment with laser radar using color camera, and calculate position and the posture of camera in real time.On the basis of quick vision SLAM, introduce the information of laser radar, and selectively use laser radar information, modified result is carried out in time, again three-dimensional map is built by revised result, make positioning that there is correct yardstick and higher precision with building figure, while making algorithm that there is relatively low complexity.
Description
Technical field
The invention belongs to technical field of machine vision, and in particular to a kind of positioning immediately of the high accuracy of multisensor is with building figure side
Method.
Background technology
The movement of robot at this stage plans that the independent navigation ability of robot is dependent on instant by manual path mostly
Position and build diagram technology (Simultaneous Localization and Mapping, hereinafter referred to as SLAM).Its core is appointed
Business is, when robot enters unknown working environment, using sensor information, high efficiency and exactly structure to be carried out to surrounding environment
(Mapping) is built, while obtaining the position of equipment in space and posture (Localization).Except that can apply in machine
Outside people field, the spatial pursuit and automatic Pilot of virtual reality and augmented reality equipment can similarly use SLAM technologies.
SLAM problems have been pass by 30 years since the proposition, sensor and computational methods during this used in SLAM problems
It all there occurs huge change.Using vision sensor the SLAM technologies of main flow more, including monocular sensor, many mesh sensors and
Sensor of coloured image and depth information (RGB-D) etc..In recent years, with the development of laser radar technique, laser radar is set
The standby accurate estimation ability by it to environmental metrics, and related for characteristic the allows laser radar such as illumination variation is insensitive
SLAM schemes have wide development space in actual applications.
The SLAM schemes of the view-based access control model of main flow can be divided into two classes according to optimization method at this stage, and a class is to use wave filter
SLAM methods, another kind of is using figure optimization SLAM methods.
It is fairly simple using the SLAM scheme model constructions of wave filter, but error can gradually add up unrepairable.Davison
Et al. in document " A.Davison, I.Reid, and N.Molton.MonoSLAM:Real-Time Single Camera
SLAM.IEEE Transactions on Pattern Analysis and Machine Intelligence,pp.1052-
MonoSLAM schemes are proposed in 1067,2007. ", are the monocular sensor based on extended Kalman filter (abbreviation EKF)
SLAM schemes, are solved by building road sign point in the picture and carrying out model construction to SLAM using EKF.
Using figure optimization method SLAM schemes due to posture figure to be built often operand than larger.Artal et al. exists
Document " R.Artal, J.Montiel, and J.Tardos.ORB-SLAM:AVersatile and Accurate
Monocular SLAM System.IEEE Transactions on Robotics,vol.31,no.5,pp 1147-1163,
The ORB-SLAM proposed in 2015. " is the monocular vision SLAM schemes for being currently based on figure optimization method, and ORB-SLAM passes through to figure
As extracting ORB feature descriptors (feature descriptor that a kind of extraction rate is exceedingly fast), very high speed can be reached, and pass through
Perfect figure optimization operation, obtained map also has very high accuracy.
The problem of SLAM algorithms of monocular vision sensor exist in actual applications has：Monocular vision sensor is difficult to estimate
Count the scale of environment, i.e. yardstick；The depth of environment is tried to achieve by mathematical computations such as trigonometric ratios, often with error；Visual sensing
Device is difficult to processing illumination acutely, the scene such as quick mobile and shortage texture, can cause Attitude estimation precise decreasing；In order to eliminate
Above-mentioned error is, it is necessary to introduce large-scale optimizatoin, and solution is difficult and time-consuming.
And other sensor such as laser radar sensors are used alone have that calculating is slow, information is not enough enriched etc. and ask
Topic, SLAM tasks can not be equally solved well.
The content of the invention
In order to overcome the deficiencies in the prior art, the present invention provides a kind of high accuracy positioning immediately of multisensor with building figure side
Method, that is, merge vision sensor and complete high accuracy method of the positioning with building figure immediately with laser radar data.In quick vision
On the basis of SLAM, by introducing the information of laser radar, with higher positioning with building figure precision, and by laser radar
Data are selectively used, and in the case where ensureing accuracy, make the complexity of algorithm relatively low.
A kind of positioning immediately of the high accuracy of multisensor is with building drawing method, it is characterised in that step is as follows：
Step 1：Input color image is simultaneously pre-processed：The color image sequence that camera is shot is inputted, by each frame cromogram
As being converted to gray level image；
Step 2：Feature extraction：Feature extraction is carried out to each frame gray level image using ORB feature extraction algorithms, obtains every
The ORB characteristic points of one two field picture；
Step 3：Characteristic matching and estimation：Characteristic point progress using approximate KNN storehouse to adjacent two field pictures
Match somebody with somebody, estimation is carried out to being solved using the PnP based on random sampling consensus method to the point matched, the motion for obtaining camera is estimated
Count sequenceWherein,The i-th frame and the are being shot for camera
Estimation between i-1 two field pictures, i=2 ..., N, N is number of image frames, RCamRepresent camera spin matrix, tCamRepresent camera
Transposed matrix；
Step 4：Input correspondence laser radar data is simultaneously pre-processed：Laser radar data is inputted, will be apart from more than laser thunder
0.8 times of the point up to range removes, and laser radar point cloud data is constituted by left point；It is assumed here that described laser radar with
The position relationship of the camera of shoot coloured image is fixed, has completed combined calibrating in step 1, and the two visual field is identical, internal ginseng
Number is known；
Step 5：The characteristic matching and estimation of laser radar point cloud data：To laser radar point cloud data every x frames
A characteristic matching and estimation are carried out, i.e., each frame laser radar point cloud data is divided into four etc. in the horizontal direction
Big region, 4 marginal points and 4 planar point constitutive characteristic points are extracted in each region according to smoothness, according to smoothness to jth frame
Matched with the characteristic point of jth-x frame data, if the point to matching is obtained to carrying out estimation using ICP algorithm
The estimation sequence of laser radar
Wherein,For estimation of the laser radar between scanning jth frame and jth-x frame data, j=x+
1,2x+1 ..., M, M are laser radar data totalframes, and x spans are [5,20], RLidarLaser radar spin matrix is represented,
tLidarRepresent laser radar transposed matrix；
Described marginal point is the minimum point of smoothness in region, and described planar point is smoothness maximum in region
Point, smoothness isWherein, S represents the set that points all in kth time scanning are constituted,
Lk,pWith Lk,qP and q points space coordinate are represented, | | | | represent Euclidean distance；Described refers to according to smoothness progress matching
If the difference of smoothness is less than the 5% of the smoothness of current point, match, otherwise, mismatch；
Step 6：Error correction and recording track：The estimation of the laser radar obtained with step 5 replaces the correspondence time
The estimation sum of camera, obtains the camera motion estimated sequence S ' after error correction, the i.e. movement locus of camera in interval；
Step 7：Set up three-dimensional point cloud map：According toCalculating obtains three-dimensional point cloud map pclmap, its
In,For n-thjTwo field picture transforms to the three-dimensional point cloud under the first two field picture angular view coordinate system
Map, if the picture frame serial number n that the synchronization camera of m frames laser radar point cloud scanning is shotj, by njThe sequence of composition
For N, m=x+1,2x+1 ..., M；For n-thjThe point cloud of two field picture
Map, L be step 2 in extract n-thjThe number of the characteristic point of two field picture, (xl,yl,zl) represent point coordinate, by following public affairs
Formula is calculated：
xl=(ul-cx)·dl/fx
yl=(vl-cy)·dl/fy
zl=dl
Wherein, l=1 ..., L, (ul,vl) for the image pixel coordinates of l-th characteristic point, (fx,fy,cx,cy) for what is given
Camera parameter, dlThe depth value retrieved for l-th of characteristic point in the corresponding laser radar point cloud of image capture moment；
Such as fruit dot cloud density is excessive, then down-sampled to a cloud map datum progress, that is, obtains final three-dimensional point cloud map；
The described point excessive number for referring to one cubic metre of space midpoint of cloud density is more than 10000.
The beneficial effects of the invention are as follows：On the basis of quick vision SLAM, with reference to the information of laser radar, make SLAM
Positioning in task has higher precision and correct yardstick with figure of founding a capital, at the same overcome single laser radar SLAM speed compared with
Slowly not the shortcomings of, not having colouring information；On the premise of accuracy is ensured, algorithm selectively uses laser radar information, it is ensured that
The real-time and low complex degree of algorithm；This algorithm can output campaign estimated result, and correct the result in time in real time, simultaneously
Three-dimensional map is built by revised result, algorithm practicality had so both been ensure that, and in turn ensure that algorithm in a long time
There is higher precision.
Brief description of the drawings
Fig. 1 is a kind of high accuracy positioning immediately of multisensor of the present invention with building drawing method flow chart
Embodiment
The present invention is further described with reference to the accompanying drawings and examples, and the present invention includes but are not limited to following implementations
Example.
The invention provides a kind of high accuracy of multisensor, positioning, with building drawing method, uses color camera and laser immediately
Radar carries out high-precision real-time three-dimensional structure to surrounding environment, and calculates position and the appearance of the camera of shooting image in real time
State.The inventive method realize premise be：Be for the camera of shooting and the position relationship of laser radar it is fixed, i.e., the two
There are identical displacement and deflection in shooting process.Also, laser radar has been completed in the combined calibrating with camera, i.e. image
Each pixel its depth value can be obtained in laser radar three-dimensional point cloud, and the two visual field is identical, known to inner parameter.
As shown in figure 1, positioning is with building drawing method immediately for a kind of high accuracy of multisensor of the present invention, its implementation process is such as
Under：
Step 1：Input color image is simultaneously pre-processed
The present embodiment uses PointGray Flea2 color camera shoot coloured images.Image original resolution is 1392
× 1040, frame per second is 17fps (frame is per second), i.e., the image sequence of one second 17 frames of input.First, image is pre-processed,
Coloured image is converted into gray level image to carry out subsequent arithmetic.
Step 2：Feature extraction
The gray level image inputted for previous step, extracts ORB (Oriented FAST and Rotated frame by frame first
BRIEF) characteristic point.Characteristic point herein can also be replaced by other kinds of characteristic point, such as SIFT, SURF according to demand,
But because the extraction of these characteristic points more takes, the overall speed of service of algorithm may be affected.
ORB (Oriented FAST and Rotated BRIEF) is the algorithm that a kind of rapid characteristic points are extracted and described.
This algorithm is to be proposed by scholars such as Rublee in 2011, " Rublee Ethan, et al. " ORB:An efficient
alternative to SIFT or SURF."IEEE International Conference on Computer Vision
IEEE,2012:2564-2571.”.ORB algorithms are divided into two parts, are feature point extraction and feature point description respectively.Feature extraction
It is by FAST (Features from Accelerated Segment Test) algorithm development is Lai feature point description is basis
BRIEF (Binary Robust Independent Elementary Features) feature describes algorithm improvement.ORB is special
It is to combine the detection method of FAST characteristic points with BRIEF Feature Descriptors to levy, and is done on the basis of they are original
Improve and optimization.The extraction rate that our experiments show that ORB features is about 100 times of SIFT, is 10 times of SURF.
The present embodiment directly extracts each frame figure of function pair using the ORB feature descriptors in computer vision storehouse OPENCV
As carrying out feature extraction.
Step 3：Characteristic matching and estimation
To extracting two field pictures adjacent after characteristic point, approximate KNN storehouse (Fast library for are used
Approximate nearest neighbors, FLANN) carry out characteristic matching.FLANN storehouses be it is most complete at present it is approximate most
Neighborhood matching is increased income storehouse, and it not only realizes a series of lookup algorithms, further comprises a kind of mechanism of the most fast algorithm of automatic selection.
FLANN basic algorithm is by Muja et al. in proposition in 2009, " M.Muja and D.Lowe, " Fast Approximate
Nearest Neighbors with Automatic Algorithm Configuration",in International
Conference on Computer Vision Theory and Applications(VISAPP'09),2009”。
The each two field picture newly inputted is extracted after characteristic point, using in the OPENCV of computer vision storehouse
FLANNbasedMatcher functions carry out characteristic matching with previous frame image, obtain some to corresponding feature in two field pictures
Point；Then the PnP that random sampling consensus method (RANdom SAmple Consensus, RANSAC) is based on to matching result is solved
The attitudes vibration between two field pictures is shot to camera to estimate.RANSAC is according to one group of sample number for including abnormal data
According to collection, the mathematical model parameter of data is calculated, the algorithm of effective sample data is obtained.By Fischler and Bolles in 1981
Propose at first in year, " Fischler, M.A.and Bolles, R.C.Random Sample Consensus:A Paradigm
for Model Fitting with Applications to Image Analysis and Automated
Cartography.Communications of the ACM,24(6):381–395,1981.”.The purpose that PnP is solved is profit
Determine that camera, relative to the displacement and rotation of world coordinate system, is existed by Moreno-Noguer with the known planar point pair matched
Propose within 2007, " F.Moreno-Noguer, V.Lepetit and P.Fua, Accurate Non-Iterative O (n)
Solution to the PnP Problem,IEEE International Conference on Computer Vision,
Rio de Janeiro,Brazil,October 2007.”.The present embodiment is directly used in the OPENCV of computer vision storehouse
Solve PnPRansac functions carry out computing.
By this step can obtain between arbitrary neighborhood two field pictures camera on world coordinate system (the first two field picture
Visual angle be principal direction coordinate system) estimation (RCam,tCam) and camera estimation sequenceWherein,The i-th frame and the i-th -1 frame are being shot for camera
Estimation between image, i=2 ..., N, N is number of image frames, RCamRepresent camera spin matrix, tCamRepresent camera displacement
Matrix.
Because, with the presence of much noise, the attitudes vibration that this step is obtained still has necessarily in feature point extraction and matching process
Error, will be modified using step 6.
Step 4：Input correspondence laser radar data is simultaneously pre-processed
The present embodiment uses the HDL-64E laser radars that Velodyne companies of the U.S. produce, and is a kind of small-sized multi-thread panorama
Laser radar, possesses the visual field of 26.9 ° of longitudinal direction, laterally 360 °.It can be evenly distributed each second on 64 environmental scanning lines
220000 points, to represent distance and position relationship of the surrounding space point apart from laser radar central point.Effective range
For 120m.
In actual use, for reduce error, will remove beyond the 0.8 of effective range times of point, that is, remove measurement away from
From the spatial point more than 80m, it is subsequently introduced and remaining constitutes cloud data a little.
Step 5：The characteristic matching and estimation of laser radar point cloud data
In this step, the present invention is by the laser radar point cloud data according to input, every x frames, extracts feature and carries out
Estimation, to reduce algorithm complex, improve computational efficiency, so, can obtain the estimation sequence of laser radar
RowWherein,It is that laser radar is scanning jth
Estimation between frame and jth-x frame data, j=x+1,2x+1 ..., M, M is laser radar data frame number, x spans
For [5,20], x values are 5, R in the present embodimentLidarRepresent laser radar spin matrix, tLidarRepresent laser radar displacement square
Battle array.Specially：
Because a cloud is discrete three-dimensional point in space, there is no other information in addition to coordinate, and same spatial point
Coordinate is continually changing in mobile data acquisition, so the traceable feature defined in a cloud, determines for this
A kind of feature extraction mode based on smoothness of justice, that is, defining smoothness c is：
Wherein, S represents the set that points all in kth time scanning are constituted, Lk,pWith Lk,qP and q points space coordinate are represented,
| | | | represent Euclidean distance.
The calculating of smoothness is carried out for point all in S, the maximum point of smoothness is defined as planar point, and smoothness is minimum
Point be then defined as marginal point.It is excessive in order to avoid the characteristic point of generation, the point cloud scanned every time is divided into four etc. first
Big region, such as takes 0 °, 90 °, 180 ° and 270 ° four principal directions, each direction or so respectively takes ± 45 ° to be divided into a region.Often
Individual region is up to 4 marginal points and 4 planar point constitutive characteristic points.
Obtain after characteristic point, the characteristic point of jth frame and jth-x frame data is matched according to smoothness, smoothness
Difference is less than the 5% characteristic point point of the smoothness of current point to being considered expression of the same spatial point in twice sweep,
The point pair as matched.Moved after the characteristic point matched using ICP algorithm (Iterative Closest Point)
Estimation, i.e., obtain the estimation (R of laser radar by solving two groups of three dimensions pointsLidar,tLidar).ICP algorithm by
Zhang et al. propositions, " Zhang, Zhengyou (1994) " Iterative point matching for
registration of free-form curves and surfaces".International Journal of
Computer Vision.Springer.13(12):119–152.”
Step 6：Error correction and recording track
Because laser radar point cloud data precision is high, data are reliable, and the error in the result of positioning is minimum, but computing
Speed is slower；And the algorithm based on coloured image is very fast in arithmetic speed, but precision is poor.The present invention is for two kinds of sensors
The different qualities followed the trail of of posture, carry out error by the way of laser radar estimation replaces camera motion estimation sum
Amendment, lifts the accuracy of total algorithm, reduces the overall time complexity of algorithm.
Estimation (the once fortune between scanning jth frame and jth-x frame data of the laser radar obtained with step 5
Dynamic estimation) replace correspondence time interval (time intervals of scanning jth frame and jth-x frame data) interior camera estimation it
With, obtain the camera motion estimated sequence after error correction, that is, camera movement locus.
Step 7：Set up three-dimensional point cloud map
When finally setting up high-precision three-dimensional point cloud map, only using the estimation of the less laser radar of error
(RLidar,tLidar).Specially：
According toCalculating obtains three-dimensional point cloud map pclmap, wherein
N-th shot for the synchronization camera that m frames laser radar point cloud is scannedjTwo field picture (sets Laser Radar Scanning m frame points
The picture frame serial number n that cloud synchronization camera is shotj, by njThe sequence of composition is N) transform to the first two field picture angular view coordinate
Three-dimensional point cloud map under system,For n-thjThe point cloud of two field picture
Figure, L is n-th in step 2jTwo field picture extracts obtained feature point number, (xl,yl,zl) represent point coordinate, as follows
Calculate：
xl=(ul-cx)·dl/fx
yl=(vl-cy)·dl/fy
zl=dl
Wherein, l=1 ..., L, (ul,vl) for the image coordinate of l-th characteristic point, (fx,fy,cx,cy) it is given camera
Parameter, dlThe depth value retrieved for l-th of characteristic point in the corresponding laser radar point cloud of image capture moment.
As can be seen that the point cloud map so built is superimposed by Multiple-Scan, it is overall excessively dense.It is aobvious in order to improve
Show speed, a reduction point cloud dense degree, the installation space dot density threshold value in the storage of a cloud, i.e., when in one cubic metre of space
When spatial point number is more than 10000, the down-sampled number to ensure spatial point is carried out to cloud data in rational scope,
Obtain final point cloud map.Certainly, when scanning narrow space, spatial point density threshold can be correspondingly improved.
The present embodiment is to be in central processing unitI5-4590 3.4GHz CPU, internal memory 16G, Ubuntu14.04
The experiment carried out on the computer of operating system, experiment using third party's data set in urban environment camera motion track to prove
Levels of precision.
Test the sequence 13 used and come from KITTI data sets, the database is by A.Geiger et al. in document
“A.Geiger and Philip Lenz,Vision meets Robotics:The KITTI Dataset,
International Journal of Robotics Research, 2013 " middle propositions, by PointGray Flea2
Color cameras camera shoot coloured images, cloud data is gathered by Velodyne HDL-64E laser radars.
The ratio (%) and algorithm of the offset distance occupation of land figure size of camera motion track and actual value is used to take as property
Energy parameter, the ORB-SLAM algorithms of the inventive method and main flow in monocular vision is contrasted, as a result as listed in table 1.Can be with
Find out, the inventive method not only increases overall arithmetic speed, and surmount in precision after laser radar information is introduced
Main flow vision SLAM methods.The present invention is also applied in terms of automatic Pilot, robot autonomous navigation.
The algorithm performance of table 1 compares
| Method | Time-consuming (s) | Offset distance (%) |
| The inventive method | 120.3 | 1.8 |
| ORB-SLAM | 126.0 | 3.1 |
Claims (1)
1. a kind of positioning immediately of the high accuracy of multisensor is with building drawing method, it is characterised in that step is as follows：
Step 1：Input color image is simultaneously pre-processed：The color image sequence that camera is shot is inputted, each color image frame is turned
It is changed to gray level image；
Step 2：Feature extraction：Feature extraction is carried out to each frame gray level image using ORB feature extraction algorithms, each frame is obtained
The ORB characteristic points of image；
Step 3：Characteristic matching and estimation：The characteristic point of adjacent two field pictures is matched using approximate KNN storehouse,
Estimation is carried out to being solved using the PnP based on random sampling consensus method to the point matched, the estimation of camera is obtained
SequenceWherein,For camera shoot the i-th frame and i-th-
Estimation between 1 two field picture, i=2 ..., N, N is number of image frames, RCamRepresent camera spin matrix, tCamRepresent phase seat in the plane
Move matrix；
Step 4：Input correspondence laser radar data is simultaneously pre-processed：Laser radar data is inputted, will be apart from more than laser radar amount
0.8 times of point of journey removes, and laser radar point cloud data is constituted by left point；It is assumed here that described laser radar and step 1
The position relationship of the camera of middle shoot coloured image fixes, completed combined calibrating, and the two visual field is identical, inner parameter
Know；
Step 5：The characteristic matching and estimation of laser radar point cloud data：Laser radar point cloud data is carried out every x frames
Characteristic matching and estimation, i.e., be divided into four Ge Deng great areas in the horizontal direction by each frame laser radar point cloud data
4 marginal points and 4 planar point constitutive characteristic points are extracted in domain, each region according to smoothness, according to smoothness to jth frame and the
The characteristic point of j-x frame data is matched, if the point to matching obtains laser to carrying out estimation using ICP algorithm
The estimation sequence of radar
Wherein,For estimation of the laser radar between scanning jth frame and jth-x frame data, j=x+1,2x+
1 ..., M, M are laser radar data totalframes, and x spans are [5,20], RLidarRepresent laser radar spin matrix, tLidar
Represent laser radar transposed matrix；
Described marginal point is the minimum point of smoothness in region, and described planar point is the maximum point of smoothness in region, light
Slippery isWherein, S represents the set that points all in kth time scanning are constituted, Lk,pWith
Lk,qP and q points space coordinate are represented, | | | | represent Euclidean distance；Described refers to if light according to smoothness progress matching
The difference of slippery is less than the 5% of the smoothness of current point, then matches, otherwise, mismatches；
Step 6：Error correction and recording track：The estimation of the laser radar obtained with step 5 replaces correspondence time interval
The estimation sum of interior camera, obtains the camera motion estimated sequence S ' after error correction, the i.e. movement locus of camera；
Step 7：Set up three-dimensional point cloud map：According toCalculating obtains three-dimensional point cloud map pclmap, wherein,For n-thjTwo field picture is with transforming to the three-dimensional point cloud under the first two field picture angular view coordinate system
Figure, if the picture frame serial number n that the synchronization camera of m frames laser radar point cloud scanning is shotj, by njThe sequence of composition is
N, m=x+1,2x+1 ..., M；For n-thjThe point cloud of two field picture
Figure, L be step 2 in extract n-thjThe number of the characteristic point of two field picture, (xl,yl,zl) represent point coordinate, as follows
Calculate：
xl=(ul-cx)·dl/fx
yl=(vl-cy)·dl/fy
zl=dl
Wherein, l=1 ..., L, (ul,vl) for the image coordinate of l-th characteristic point, (fx,fy,cx,cy) join for given camera
Number, dlThe depth value retrieved for l-th of characteristic point in the corresponding laser radar point cloud of image capture moment；
Such as fruit dot cloud density is excessive, then down-sampled to a cloud map datum progress, that is, obtains final three-dimensional point cloud map；It is described
Point the excessive number for referring to one cubic metre of space midpoint of cloud density more than 10000.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710437709.1A CN107301654B (en) | 2017-06-12 | 2017-06-12 | Multi-sensor high-precision instant positioning and mapping method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710437709.1A CN107301654B (en) | 2017-06-12 | 2017-06-12 | Multi-sensor high-precision instant positioning and mapping method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN107301654A true CN107301654A (en) | 2017-10-27 |
| CN107301654B CN107301654B (en) | 2020-04-03 |
Family
ID=60135365
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710437709.1A Active CN107301654B (en) | 2017-06-12 | 2017-06-12 | Multi-sensor high-precision instant positioning and mapping method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107301654B (en) |
Cited By (36)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108319976A (en) * | 2018-01-25 | 2018-07-24 | 北京三快在线科技有限公司 | Build drawing method and device |
| CN108399643A (en) * | 2018-03-15 | 2018-08-14 | 南京大学 | A kind of outer ginseng calibration system between laser radar and camera and method |
| CN108489496A (en) * | 2018-04-28 | 2018-09-04 | 北京空间飞行器总体设计部 | Noncooperative target Relative Navigation method for estimating based on Multi-source Information Fusion and system |
| CN108647646A (en) * | 2018-05-11 | 2018-10-12 | 北京理工大学 | The optimizing detection method and device of low obstructions based on low harness radar |
| CN108709560A (en) * | 2018-08-15 | 2018-10-26 | 苏州中研讯科智能科技有限公司 | Carrying robot high accuracy positioning air navigation aid based on straightway feature |
| CN108734654A (en) * | 2018-05-28 | 2018-11-02 | 深圳市易成自动驾驶技术有限公司 | It draws and localization method, system and computer readable storage medium |
| CN108759823A (en) * | 2018-05-28 | 2018-11-06 | 浙江大学 | The positioning of low speed automatic driving vehicle and method for correcting error in particular link based on images match |
| CN108896994A (en) * | 2018-05-11 | 2018-11-27 | 武汉环宇智行科技有限公司 | A kind of automatic driving vehicle localization method and equipment |
| CN109031304A (en) * | 2018-06-06 | 2018-12-18 | 上海国际汽车城（集团）有限公司 | Vehicle positioning method in view-based access control model and the tunnel of millimetre-wave radar map feature |
| CN109073398A (en) * | 2018-07-20 | 2018-12-21 | 深圳前海达闼云端智能科技有限公司 | Map establishing method, positioning method, device, terminal and storage medium |
| CN109166149A (en) * | 2018-08-13 | 2019-01-08 | 武汉大学 | A kind of positioning and three-dimensional wire-frame method for reconstructing and system of fusion binocular camera and IMU |
| CN109326006A (en) * | 2018-09-30 | 2019-02-12 | 百度在线网络技术（北京）有限公司 | Map amalgamation method and device |
| CN109341706A (en) * | 2018-10-17 | 2019-02-15 | 张亮 | A kind of production method of the multiple features fusion map towards pilotless automobile |
| CN109407115A (en) * | 2018-12-25 | 2019-03-01 | 中山大学 | A kind of road surface extraction system and its extracting method based on laser radar |
| CN109579843A (en) * | 2018-11-29 | 2019-04-05 | 浙江工业大学 | Multirobot co-located and fusion under a kind of vacant lot multi-angle of view build drawing method |
| CN109634279A (en) * | 2018-12-17 | 2019-04-16 | 武汉科技大学 | Object positioning method based on laser radar and monocular vision |
| CN109974742A (en) * | 2017-12-28 | 2019-07-05 | 沈阳新松机器人自动化股份有限公司 | A kind of laser Method for Calculate Mileage and map constructing method |
| CN110068824A (en) * | 2019-04-17 | 2019-07-30 | 北京地平线机器人技术研发有限公司 | A kind of sensor pose determines method and apparatus |
| CN110163968A (en) * | 2019-05-28 | 2019-08-23 | 山东大学 | RGBD camera large-scale three dimensional scenario building method and system |
| CN110398745A (en) * | 2019-08-05 | 2019-11-01 | 湖南海森格诺信息技术有限公司 | Fork truck localization method based on laser radar and vision |
| CN110786585A (en) * | 2019-11-21 | 2020-02-14 | 朱利 | Multifunctional intelligent helmet |
| CN110823211A (en) * | 2019-10-29 | 2020-02-21 | 珠海市一微半导体有限公司 | Multi-sensor map construction method, device and chip based on visual SLAM |
| WO2020097840A1 (en) * | 2018-11-15 | 2020-05-22 | Beijing Didi Infinity Technology And Development Co., Ltd. | Systems and methods for correcting a high-definition map based on detection of obstructing objects |
| CN111207670A (en) * | 2020-02-27 | 2020-05-29 | 河海大学常州校区 | Line structured light calibration device and method |
| CN111428578A (en) * | 2020-03-03 | 2020-07-17 | 深圳市镭神智能***有限公司 | Self-body and positioning method and device thereof |
| CN111445530A (en) * | 2020-03-24 | 2020-07-24 | 广东博智林机器人有限公司 | Wheelchair bed and wheelchair homing path determination method |
| CN111481109A (en) * | 2019-01-28 | 2020-08-04 | 北京奇虎科技有限公司 | Map noise elimination method and device based on sweeper |
| WO2020154966A1 (en) * | 2019-01-30 | 2020-08-06 | Baidu.Com Times Technology (Beijing) Co., Ltd. | A rgb point clouds based map generation system for autonomous vehicles |
| CN111694903A (en) * | 2019-03-11 | 2020-09-22 | 北京地平线机器人技术研发有限公司 | Map construction method, map construction device, map construction equipment and readable storage medium |
| CN111912431A (en) * | 2020-03-19 | 2020-11-10 | 中山大学 | Positioning precision test scheme for mobile robot navigation system |
| CN112132754A (en) * | 2020-11-25 | 2020-12-25 | 蘑菇车联信息科技有限公司 | Vehicle movement track correction method and related device |
| CN112150507A (en) * | 2020-09-29 | 2020-12-29 | 厦门汇利伟业科技有限公司 | Method and system for synchronously reproducing 3D model of object posture and displacement |
| CN112230211A (en) * | 2020-10-15 | 2021-01-15 | 长城汽车股份有限公司 | Vehicle positioning method and device, storage medium and vehicle |
| CN112484746A (en) * | 2020-11-26 | 2021-03-12 | 上海电力大学 | Monocular vision-assisted laser radar odometer method based on ground plane |
| TWI772177B (en) * | 2021-09-10 | 2022-07-21 | 迪伸電子股份有限公司 | Movement control method of automatic guided device and automatic guided device |
| CN114863075A (en) * | 2022-07-05 | 2022-08-05 | 深圳市新天泽消防工程有限公司 | Fire-fighting evacuation path planning method, device and equipment based on multiple sensors |
Citations (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20040167716A1 (en) * | 2002-12-17 | 2004-08-26 | Goncalves Luis Filipe Domingues | Systems and methods for controlling a density of visual landmarks in a visual simultaneous localization and mapping system |
| CN104359464A (en) * | 2014-11-02 | 2015-02-18 | 天津理工大学 | Mobile robot positioning method based on stereoscopic vision |
| CN104764457A (en) * | 2015-04-21 | 2015-07-08 | 北京理工大学 | Urban environment composition method for unmanned vehicles |
| CN104933708A (en) * | 2015-06-07 | 2015-09-23 | 浙江大学 | Barrier detection method in vegetation environment based on multispectral and 3D feature fusion |
| US20150304634A1 (en) * | 2011-08-04 | 2015-10-22 | John George Karvounis | Mapping and tracking system |
| CN105856230A (en) * | 2016-05-06 | 2016-08-17 | 简燕梅 | ORB key frame closed-loop detection SLAM method capable of improving consistency of position and pose of robot |
| CN106052674A (en) * | 2016-05-20 | 2016-10-26 | 青岛克路德机器人有限公司 | Indoor robot SLAM method and system |
| CN106127739A (en) * | 2016-06-16 | 2016-11-16 | 华东交通大学 | A kind of RGB D SLAM method of combination monocular vision |
| CN106153048A (en) * | 2016-08-11 | 2016-11-23 | 广东技术师范学院 | A kind of robot chamber inner position based on multisensor and Mapping System |
| CN106296812A (en) * | 2016-08-18 | 2017-01-04 | 宁波傲视智绘光电科技有限公司 | Synchronize location and build drawing method |
| CN106443687A (en) * | 2016-08-31 | 2017-02-22 | 欧思徕（北京）智能科技有限公司 | Piggyback mobile surveying and mapping system based on laser radar and panorama camera |
| CN106446815A (en) * | 2016-09-14 | 2017-02-22 | 浙江大学 | Simultaneous positioning and map building method |
| CN106595659A (en) * | 2016-11-03 | 2017-04-26 | 南京航空航天大学 | Map merging method of unmanned aerial vehicle visual SLAM under city complex environment |
| CN106681330A (en) * | 2017-01-25 | 2017-05-17 | 北京航空航天大学 | Robot navigation method and device based on multi-sensor data fusion |
-
2017
- 2017-06-12 CN CN201710437709.1A patent/CN107301654B/en active Active
Patent Citations (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20040167716A1 (en) * | 2002-12-17 | 2004-08-26 | Goncalves Luis Filipe Domingues | Systems and methods for controlling a density of visual landmarks in a visual simultaneous localization and mapping system |
| US20150304634A1 (en) * | 2011-08-04 | 2015-10-22 | John George Karvounis | Mapping and tracking system |
| CN104359464A (en) * | 2014-11-02 | 2015-02-18 | 天津理工大学 | Mobile robot positioning method based on stereoscopic vision |
| CN104764457A (en) * | 2015-04-21 | 2015-07-08 | 北京理工大学 | Urban environment composition method for unmanned vehicles |
| CN104933708A (en) * | 2015-06-07 | 2015-09-23 | 浙江大学 | Barrier detection method in vegetation environment based on multispectral and 3D feature fusion |
| CN105856230A (en) * | 2016-05-06 | 2016-08-17 | 简燕梅 | ORB key frame closed-loop detection SLAM method capable of improving consistency of position and pose of robot |
| CN106052674A (en) * | 2016-05-20 | 2016-10-26 | 青岛克路德机器人有限公司 | Indoor robot SLAM method and system |
| CN106127739A (en) * | 2016-06-16 | 2016-11-16 | 华东交通大学 | A kind of RGB D SLAM method of combination monocular vision |
| CN106153048A (en) * | 2016-08-11 | 2016-11-23 | 广东技术师范学院 | A kind of robot chamber inner position based on multisensor and Mapping System |
| CN106296812A (en) * | 2016-08-18 | 2017-01-04 | 宁波傲视智绘光电科技有限公司 | Synchronize location and build drawing method |
| CN106443687A (en) * | 2016-08-31 | 2017-02-22 | 欧思徕（北京）智能科技有限公司 | Piggyback mobile surveying and mapping system based on laser radar and panorama camera |
| CN106446815A (en) * | 2016-09-14 | 2017-02-22 | 浙江大学 | Simultaneous positioning and map building method |
| CN106595659A (en) * | 2016-11-03 | 2017-04-26 | 南京航空航天大学 | Map merging method of unmanned aerial vehicle visual SLAM under city complex environment |
| CN106681330A (en) * | 2017-01-25 | 2017-05-17 | 北京航空航天大学 | Robot navigation method and device based on multi-sensor data fusion |
Non-Patent Citations (5)
| Title |
|---|
| LV QIANG 等: "Absolute Scale Estimation of ORB-SLAM Algorithm based on Laser Ranging", 《PROCEEDINGS OF THE 35TH CHINESE CONTROL CONFERENCE》 * |
| RA´UL MUR-ARTAL 等: "ORB-SLAM: a Versatile and Accurate Monocular SLAM System", 《IEEE TRANSACTIONS ON ROBOTICS》 * |
| SATOSHIFUJIMOTO 等: "ORBSLAMMAPINITIALIZATIONIMPROVEMENTUSINGDEPTH", 《ICIP 2016》 * |
| 侯荣波 等: "基于ORB-SLAM的室内机器人定位和三维稠密地图构建", 《计算机应用》 * |
| 林连秀 等: "基于ORB-SLAM的移动机器人嵌入式实现与优化", 《微型机与应用》 * |
Cited By (55)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109974742A (en) * | 2017-12-28 | 2019-07-05 | 沈阳新松机器人自动化股份有限公司 | A kind of laser Method for Calculate Mileage and map constructing method |
| CN108319976A (en) * | 2018-01-25 | 2018-07-24 | 北京三快在线科技有限公司 | Build drawing method and device |
| CN108399643A (en) * | 2018-03-15 | 2018-08-14 | 南京大学 | A kind of outer ginseng calibration system between laser radar and camera and method |
| CN108489496A (en) * | 2018-04-28 | 2018-09-04 | 北京空间飞行器总体设计部 | Noncooperative target Relative Navigation method for estimating based on Multi-source Information Fusion and system |
| CN108896994A (en) * | 2018-05-11 | 2018-11-27 | 武汉环宇智行科技有限公司 | A kind of automatic driving vehicle localization method and equipment |
| CN108647646A (en) * | 2018-05-11 | 2018-10-12 | 北京理工大学 | The optimizing detection method and device of low obstructions based on low harness radar |
| CN108647646B (en) * | 2018-05-11 | 2019-12-13 | 北京理工大学 | Low-beam radar-based short obstacle optimized detection method and device |
| CN108759823A (en) * | 2018-05-28 | 2018-11-06 | 浙江大学 | The positioning of low speed automatic driving vehicle and method for correcting error in particular link based on images match |
| CN108759823B (en) * | 2018-05-28 | 2020-06-30 | 浙江大学 | Low-speed automatic driving vehicle positioning and deviation rectifying method on designated road based on image matching |
| CN108734654A (en) * | 2018-05-28 | 2018-11-02 | 深圳市易成自动驾驶技术有限公司 | It draws and localization method, system and computer readable storage medium |
| CN109031304A (en) * | 2018-06-06 | 2018-12-18 | 上海国际汽车城（集团）有限公司 | Vehicle positioning method in view-based access control model and the tunnel of millimetre-wave radar map feature |
| CN109073398A (en) * | 2018-07-20 | 2018-12-21 | 深圳前海达闼云端智能科技有限公司 | Map establishing method, positioning method, device, terminal and storage medium |
| CN109073398B (en) * | 2018-07-20 | 2022-04-08 | 达闼机器人有限公司 | Map establishing method, positioning method, device, terminal and storage medium |
| CN109166149A (en) * | 2018-08-13 | 2019-01-08 | 武汉大学 | A kind of positioning and three-dimensional wire-frame method for reconstructing and system of fusion binocular camera and IMU |
| CN109166149B (en) * | 2018-08-13 | 2021-04-02 | 武汉大学 | Positioning and three-dimensional line frame structure reconstruction method and system integrating binocular camera and IMU |
| CN108709560A (en) * | 2018-08-15 | 2018-10-26 | 苏州中研讯科智能科技有限公司 | Carrying robot high accuracy positioning air navigation aid based on straightway feature |
| CN109326006A (en) * | 2018-09-30 | 2019-02-12 | 百度在线网络技术（北京）有限公司 | Map amalgamation method and device |
| CN109341706A (en) * | 2018-10-17 | 2019-02-15 | 张亮 | A kind of production method of the multiple features fusion map towards pilotless automobile |
| CN109341706B (en) * | 2018-10-17 | 2020-07-03 | 张亮 | Method for manufacturing multi-feature fusion map for unmanned vehicle |
| US11035958B2 (en) | 2018-11-15 | 2021-06-15 | Bejing Didi Infinity Technology And Development Co., Ltd. | Systems and methods for correcting a high-definition map based on detection of obstructing objects |
| WO2020097840A1 (en) * | 2018-11-15 | 2020-05-22 | Beijing Didi Infinity Technology And Development Co., Ltd. | Systems and methods for correcting a high-definition map based on detection of obstructing objects |
| CN109579843A (en) * | 2018-11-29 | 2019-04-05 | 浙江工业大学 | Multirobot co-located and fusion under a kind of vacant lot multi-angle of view build drawing method |
| CN109579843B (en) * | 2018-11-29 | 2020-10-27 | 浙江工业大学 | Multi-robot cooperative positioning and fusion image building method under air-ground multi-view angles |
| CN109634279A (en) * | 2018-12-17 | 2019-04-16 | 武汉科技大学 | Object positioning method based on laser radar and monocular vision |
| CN109634279B (en) * | 2018-12-17 | 2022-08-12 | 瞿卫新 | Object positioning method based on laser radar and monocular vision |
| CN109407115A (en) * | 2018-12-25 | 2019-03-01 | 中山大学 | A kind of road surface extraction system and its extracting method based on laser radar |
| CN109407115B (en) * | 2018-12-25 | 2022-12-27 | 中山大学 | Laser radar-based pavement extraction system and extraction method thereof |
| CN111481109B (en) * | 2019-01-28 | 2022-08-26 | 北京奇虎科技有限公司 | Map noise elimination method and device based on sweeper |
| CN111481109A (en) * | 2019-01-28 | 2020-08-04 | 北京奇虎科技有限公司 | Map noise elimination method and device based on sweeper |
| CN112105890B (en) * | 2019-01-30 | 2023-11-17 | 百度时代网络技术（北京）有限公司 | Map generation system based on RGB point cloud for automatic driving vehicle |
| CN112105890A (en) * | 2019-01-30 | 2020-12-18 | 百度时代网络技术（北京）有限公司 | RGB point cloud based map generation system for autonomous vehicles |
| WO2020154966A1 (en) * | 2019-01-30 | 2020-08-06 | Baidu.Com Times Technology (Beijing) Co., Ltd. | A rgb point clouds based map generation system for autonomous vehicles |
| CN111694903A (en) * | 2019-03-11 | 2020-09-22 | 北京地平线机器人技术研发有限公司 | Map construction method, map construction device, map construction equipment and readable storage medium |
| CN111694903B (en) * | 2019-03-11 | 2023-09-12 | 北京地平线机器人技术研发有限公司 | Map construction method, device, equipment and readable storage medium |
| CN110068824B (en) * | 2019-04-17 | 2021-07-23 | 北京地平线机器人技术研发有限公司 | Sensor pose determining method and device |
| CN110068824A (en) * | 2019-04-17 | 2019-07-30 | 北京地平线机器人技术研发有限公司 | A kind of sensor pose determines method and apparatus |
| CN110163968B (en) * | 2019-05-28 | 2020-08-25 | 山东大学 | RGBD camera large three-dimensional scene construction method and system |
| CN110163968A (en) * | 2019-05-28 | 2019-08-23 | 山东大学 | RGBD camera large-scale three dimensional scenario building method and system |
| CN110398745A (en) * | 2019-08-05 | 2019-11-01 | 湖南海森格诺信息技术有限公司 | Fork truck localization method based on laser radar and vision |
| CN110823211A (en) * | 2019-10-29 | 2020-02-21 | 珠海市一微半导体有限公司 | Multi-sensor map construction method, device and chip based on visual SLAM |
| CN110786585A (en) * | 2019-11-21 | 2020-02-14 | 朱利 | Multifunctional intelligent helmet |
| CN111207670A (en) * | 2020-02-27 | 2020-05-29 | 河海大学常州校区 | Line structured light calibration device and method |
| CN111428578A (en) * | 2020-03-03 | 2020-07-17 | 深圳市镭神智能***有限公司 | Self-body and positioning method and device thereof |
| CN111912431A (en) * | 2020-03-19 | 2020-11-10 | 中山大学 | Positioning precision test scheme for mobile robot navigation system |
| CN111445530A (en) * | 2020-03-24 | 2020-07-24 | 广东博智林机器人有限公司 | Wheelchair bed and wheelchair homing path determination method |
| CN112150507A (en) * | 2020-09-29 | 2020-12-29 | 厦门汇利伟业科技有限公司 | Method and system for synchronously reproducing 3D model of object posture and displacement |
| CN112150507B (en) * | 2020-09-29 | 2024-02-02 | 厦门汇利伟业科技有限公司 | 3D model synchronous reproduction method and system for object posture and displacement |
| CN112230211A (en) * | 2020-10-15 | 2021-01-15 | 长城汽车股份有限公司 | Vehicle positioning method and device, storage medium and vehicle |
| CN112132754B (en) * | 2020-11-25 | 2021-06-04 | 蘑菇车联信息科技有限公司 | Vehicle movement track correction method and related device |
| CN112132754A (en) * | 2020-11-25 | 2020-12-25 | 蘑菇车联信息科技有限公司 | Vehicle movement track correction method and related device |
| CN112484746B (en) * | 2020-11-26 | 2023-04-28 | 上海电力大学 | Monocular vision auxiliary laser radar odometer method based on ground plane |
| CN112484746A (en) * | 2020-11-26 | 2021-03-12 | 上海电力大学 | Monocular vision-assisted laser radar odometer method based on ground plane |
| TWI772177B (en) * | 2021-09-10 | 2022-07-21 | 迪伸電子股份有限公司 | Movement control method of automatic guided device and automatic guided device |
| CN114863075A (en) * | 2022-07-05 | 2022-08-05 | 深圳市新天泽消防工程有限公司 | Fire-fighting evacuation path planning method, device and equipment based on multiple sensors |
| CN114863075B (en) * | 2022-07-05 | 2022-10-14 | 深圳市新天泽消防工程有限公司 | Fire-fighting evacuation path planning method, device and equipment based on multiple sensors |
Also Published As
| Publication number | Publication date |
|---|---|
| CN107301654B (en) | 2020-04-03 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107301654A (en) | A kind of positioning immediately of the high accuracy of multisensor is with building drawing method | |
| CN110853075B (en) | Visual tracking positioning method based on dense point cloud and synthetic view | |
| CN112001926B (en) | RGBD multi-camera calibration method, system and application based on multi-dimensional semantic mapping | |
| CN106296812B (en) | It is synchronous to position and build drawing method | |
| CN114399554B (en) | Calibration method and system of multi-camera system | |
| CN106960442A (en) | Based on the infrared night robot vision wide view-field three-D construction method of monocular | |
| CN109615698A (en) | Multiple no-manned plane SLAM map blending algorithm based on the detection of mutual winding | |
| CN111899280B (en) | Monocular vision odometer method adopting deep learning and mixed pose estimation | |
| CN114419147A (en) | Rescue robot intelligent remote human-computer interaction control method and system | |
| Qian et al. | Robust visual-lidar simultaneous localization and mapping system for UAV | |
| CN113012208B (en) | Multi-view remote sensing image registration method and system | |
| CN103544710A (en) | Image registration method | |
| Ceriani et al. | Pose interpolation slam for large maps using moving 3d sensors | |
| CN115272596A (en) | Multi-sensor fusion SLAM method oriented to monotonous texture-free large scene | |
| CN111860651A (en) | Monocular vision-based semi-dense map construction method for mobile robot | |
| CN104751451B (en) | Point off density cloud extracting method based on unmanned plane low latitude high resolution image | |
| CN117197333A (en) | Space target reconstruction and pose estimation method and system based on multi-view vision | |
| Dong et al. | Probability driven approach for point cloud registration of indoor scene | |
| CN116740488B (en) | Training method and device for feature extraction model for visual positioning | |
| CN108921896A (en) | A kind of lower view vision compass merging dotted line feature | |
| CN117029870A (en) | Laser odometer based on road surface point cloud | |
| CN116681844A (en) | Building white film construction method based on sub-meter stereopair satellite images | |
| CN116402904A (en) | Combined calibration method based on laser radar inter-camera and monocular camera | |
| CN115830116A (en) | Robust visual odometer method | |
| Chen et al. | Binocular vision localization based on vision SLAM system with multi-sensor fusion |
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 |