CN109764869A - A kind of positioning of autonomous crusing robot and the three-dimensional map construction method of binocular camera and inertial navigation fusion - Google Patents
A kind of positioning of autonomous crusing robot and the three-dimensional map construction method of binocular camera and inertial navigation fusion Download PDFInfo
- Publication number
- CN109764869A CN109764869A CN201910042766.9A CN201910042766A CN109764869A CN 109764869 A CN109764869 A CN 109764869A CN 201910042766 A CN201910042766 A CN 201910042766A CN 109764869 A CN109764869 A CN 109764869A
- Authority
- CN
- China
- Prior art keywords
- positioning
- image
- crusing robot
- binocular camera
- construction method
- 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.)
- Pending
Links
- 238000010276 construction Methods 0.000 title claims abstract description 23
- 230000004927 fusion Effects 0.000 title claims abstract description 22
- 238000004804 winding Methods 0.000 claims abstract description 17
- 238000001514 detection method Methods 0.000 claims abstract description 13
- 238000012545 processing Methods 0.000 claims abstract description 13
- 238000011084 recovery Methods 0.000 claims abstract description 10
- 238000005457 optimization Methods 0.000 claims abstract description 8
- 238000000605 extraction Methods 0.000 claims abstract description 6
- 238000005070 sampling Methods 0.000 claims abstract description 5
- 230000003287 optical effect Effects 0.000 claims abstract description 4
- 238000000034 method Methods 0.000 claims description 19
- 239000003595 mist Substances 0.000 claims description 19
- 238000006116 polymerization reaction Methods 0.000 claims description 6
- 230000008030 elimination Effects 0.000 claims description 3
- 238000003379 elimination reaction Methods 0.000 claims description 3
- 239000004615 ingredient Substances 0.000 claims description 3
- 238000010606 normalization Methods 0.000 claims description 3
- 238000003672 processing method Methods 0.000 claims description 3
- 230000003068 static effect Effects 0.000 claims description 3
- 230000005540 biological transmission Effects 0.000 claims 1
- 230000001815 facial effect Effects 0.000 claims 1
- 238000012216 screening Methods 0.000 abstract description 3
- 230000006870 function Effects 0.000 description 10
- 230000000694 effects Effects 0.000 description 5
- 238000007689 inspection Methods 0.000 description 4
- 238000006243 chemical reaction Methods 0.000 description 2
- 238000004891 communication Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 238000003709 image segmentation Methods 0.000 description 2
- 230000004807 localization Effects 0.000 description 2
- 238000012423 maintenance Methods 0.000 description 2
- 230000001133 acceleration Effects 0.000 description 1
- 230000003321 amplification Effects 0.000 description 1
- 239000003245 coal Substances 0.000 description 1
- 230000007423 decrease Effects 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 239000002184 metal Substances 0.000 description 1
- 238000003199 nucleic acid amplification method Methods 0.000 description 1
- 230000008447 perception Effects 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 239000000126 substance Substances 0.000 description 1
Abstract
The invention discloses a kind of positioning of the autonomous crusing robot of binocular camera and inertial navigation fusion and three-dimensional map construction method, carry out denoising and pre- score to inertial guidance data, carry out carrying out feature extraction and optical flow tracking after defogging processing to left and right camera.Estimate that camera pose, screening key frame carry out winding detection and map structuring using nonlinear optimization.Map structuring step can carry out depth recovery by the left and right camera image of the corresponding binocular camera of key frame, find out the three-dimensional coordinate of known depth pixel.Above ground portion in point cloud is removed using stochastical sampling consistency, constructs three-dimensional map.Autonomous crusing robot can accurately be positioned through the invention, and construct three-dimensional environment map application in further path planning.
Description
Technical field
The present invention relates to crusing robot positioning and map structuring field more particularly to a kind of binocular camera and inertial navigation fusion
Crusing robot positioning with three-dimensional map construction method.
Background technique
As science and technology is constantly progressive, robot is applied in every field, and the intelligent development of robot is also already
As a kind of trend and obtain more and more researcher's concerns.It is numerous dangerous, no from the air to seabed from interior to outdoor
It is suitable in the environment of mankind's activity, on the post mechanical, repeatability is high, we are it can be seen that intelligent robot just gradually substitutes
The mankind complete job requirement.
Different from traditional robot, the intelligence of intelligent mobile robot is mainly reflected in its spy for having sensing capability
Point.Autonomous crusing robot is the common intelligent robot for having inspection ability, and main realize is found properly from a certain position
Path runs to source location and completes the function of particular detection task, and detection content includes equipment routing inspection, Chart recognition, number
According to reading etc..Autonomous crusing robot is currently mainly used in the environment such as substation, oil field, coal mine, warehouse, chemical plant, substitutes
Manpower is to solve the problems such as operating efficiency present in manual inspection mode is low, labour cost is high, maintenance risk is big.Not
Know, DYNAMIC COMPLEX, in the environment of variation, how crusing robot independently determines ambient enviroment, three-dimensional locating for itself pose, perception
Spatial information is still a technological difficulties.
The positioning method that crusing robot uses now mainly has electromagnetic location, inertial positioning, UWB positioning, GNSS satellite
Positioning.Electromagnetic location is laid with metal wire on the path that robot travels, and pilot frequency is loaded in metallicity, passes through identification
Pilot frequency realizes the guiding of robot.It is a major advantage that lead concealment is preferable, and system is stable and reliable, anti-interference
Ability is good.But system position error is big, and will ground be laid with magnetic stripe, it is poor so as to cause robot dexterity, maintenance at
This is also relatively high.Inertial positioning is a kind of reckoning air navigation aid that do not rely on any external information.Robot is mainly sharp
Acceleration and the azimuth that itself is measured with sensors such as electronic compass, accelerometer or gyroscopes, to calculate next point
Position.But data are obtained by integral, constant error can be accumulated by amplification with time change, navigate for a long time
It will lead to positioning accuracy decline.Inertial positioning is a kind of reckoning method and odometer localization method is having the same lacks
Point, when Robot, which focuses on rerail mark, to be worked, error can constantly add up.UWB is a kind of no-load communication techniques, is utilized
The non-sinusoidal waveform burst pulse of nanosecond to picosecond grade transmits data.But one kind as pseudolite positioning method, UWB is in complexity
Locating effect is less desirable under dynamic environment.GNSS satellite positioning (navigation) system can be provided to worldwide navigation communication equipment
The location information services such as real-time position, speed, but positioning accuracy can only achieve several meters, is generally used for outdoor navigation and believes
It number is easy by trees, building etc. to be blocked influence, and interior is not available.
For this purpose, the present invention now proposes the positioning of autonomous crusing robot and the three-dimensional map of a kind of binocular camera and inertial navigation fusion
Construction method.
Summary of the invention
In order to overcome the deficiencies in the prior art, the present invention provides independently patrolling for a kind of binocular camera and inertial navigation fusion
Robot localization and three-dimensional map construction method are examined, high-precision pose estimation and building three-dimensional ring condition are carried out by the method
Figure, can be applied to autonomous crusing robot.
To achieve the above object, the technical solution adopted by the present invention are as follows:
A kind of positioning of autonomous crusing robot and the three-dimensional map construction method of binocular camera and inertial navigation fusion, positioning and three
Dimension map constructing method is as follows: inertial guidance data carries out pre- score processing after denoising, and left and right camera is after defogging processing
Carry out feature extraction and optical flow tracking；Inertial guidance data and binocular camera data enter positioning step after being initialized, and pass through
Nonlinear optimization estimation crusing robot pose and select key frame enter winding detection and map structuring step；In map structuring
Step carries out depth recovery using the method for cut tree to the corresponding binocular camera RGB image of key frame, removes ground to cloud
Construct three-dimensional map；Global B A optimization is carried out, inspection machine is optimized when winding detection is judged as YES in winding detecting step
People's pose and map.
Further, defogging processing method is as follows: judging whether there is mist in crusing robot local environment using dark
Haze, mist graph model indicate are as follows:
I (x)=J (x) t (x)+A (1-t (x)),
Wherein I (x) is to mist elimination image, and J (x) is the fog free images to be restored, and A is global atmosphere light ingredient, and t (x) is
Transmissivity；
The relationship in each channel of image further indicates that are as follows:
Wherein C indicates each channel of image；For arbitrary image J, dark is defined as:
J in formulacIndicate each channel of color image, Ω (x) indicates each window；The estimated value of transmissivity solves are as follows:
W is constant parameter in formula, so that image is had the depth of field to retain a degree of mist, w often takes the value less than 1.
Further, image recovery method is as follows:
T in formula0For threshold value；Probability statistics are carried out to dark t (x) pixel of image, are total picture when pixel is greater than threshold value ζ
When the η ratio of element, image is as the processing of mist figure；Determine that ζ is 192, η 40%.
Further, binocular camera data and inertial guidance data if be able to carry out initialization just directly initialize, if cannot be into
Row initialization is just first initialized using binocular in fusion inertial guidance data.
Further, generating a key frame must satisfy following several conditions:
1) at least through 15 frame images from the generation of a upper key frame；
2) at least 50 characteristic points are traced into；
It 3) has been more than the threshold value set at a distance from previous keyframe.
Further, when crusing robot increases by two compulsory constraints when rotated:
1) it has to generate a key frame when robot short time rotation angle is more than 15 °；
2) it when robot is static, has within every 3 seconds generate a key frame.
Further, regarding camera image I as connected undirected graph G=(V, E) in map structuring step；For connected pixel
The side e weight of s and r indicates are as follows:
we=w (s, r)=| I (s)-I (r) |；
Non local polymerization cost CA d(p) C of weighting is indicateddSum:
Wherein q indicate image I in each pixel, S (p, q) indicate with pixel q in function to pixel to p in contribution
Weighting function；
In tree construction, S (p, q) is defined as:
Wherein σ is the parameter for distance adjustment；
Cost value of the pixel p at the horizontal d of each parallax:
Further, depth is z in map structuring stepcPixel (u, v) corresponding three-dimensional points Pw are as follows:
The corresponding normal vector of the cloud are as follows:
The radius of initial error range are as follows:
R=zc/f；
Initial confidence level are as follows:
C=exp (- γ2/2σ2),
Wherein, γ is the depth data after normalization；σ is generally taken as 0.6.
Further, interior point and exterior point are calculated using stochastical sampling consistency in map structuring step, it is flat to solve ground
Face equation:
Ax+by+cz+d=0,
Wherein, a, b, c are the coordinate of normal vector in Hessian normal form, and d is constant.
Further, defining the relative pose between two frame of image in winding detecting step:
The residual error of i-th frame and jth frame is defined as:
All sequences while and overall goals function when closed loop is defined as:
Wherein S is sequence side, and L is closed loop side, and h is robust kernel function.
Further, the image used is grayscale image in positioning and winding detection, the picture of grayscale image and RGB figure is defined
Element conversion are as follows: grayscale image=0.29*R+0.58*G+0.11*B；RGB image is used in map structuring step.
The invention has the advantages that:
Autonomous crusing robot can accurately be positioned through the invention, and construct three-dimensional environment map application
In further path planning；High-precision pose estimation and building three-dimensional environment map are carried out by the method, can be applied
In autonomous crusing robot.
Detailed description of the invention
Fig. 1 is positioning and three-dimensional map construction method flow chart of the invention；
Fig. 2 is mist primitive figure of the invention；
Fig. 3 is the dark channel diagram of mist figure of the invention；
Fig. 4 is mist figure defog effect figure of the invention；
Fig. 5 is left camera image of the invention；
Fig. 6 is right camera image of the invention；
Fig. 7 is each subtree root node distribution map in image of the invention；
Fig. 8 is image segmentation figure of the invention；
Fig. 9 is that binocular estimating depth of the invention restores figure；
Figure 10 is the three-dimensional point cloud with ground of the invention；
Figure 11 is the three-dimensional point cloud on removal ground of the invention；
Figure 12 is three-dimensional map of the invention.
Specific embodiment
To keep the purposes, technical schemes and advantages of the invention implemented clearer, below in conjunction in the embodiment of the present invention
Attached drawing, technical solution in the embodiment of the present invention is further described in more detail.In the accompanying drawings, identical from beginning to end or class
As label indicate same or similar element or element with the same or similar functions.Described embodiment is the present invention
A part of the embodiment, instead of all the embodiments.The embodiments described below with reference to the accompanying drawings are exemplary, it is intended to use
It is of the invention in explaining, and be not considered as limiting the invention.Based on the embodiments of the present invention, ordinary skill people
Member's every other embodiment obtained without making creative work, shall fall within the protection scope of the present invention.Under
Face is described in detail the embodiment of the present invention in conjunction with attached drawing.
As shown in Figure 1, a kind of positioning of the autonomous crusing robot of binocular camera and inertial navigation fusion and three-dimensional map building side
Method, including the following contents: entire method frame is divided into data processing, initialization, positioning, map structuring and winding and detects five steps
Suddenly.Inertial guidance data carries out pre- score processing after denoising, and left and right camera carries out feature extraction and light after defogging processing
Stream tracking；Inertial guidance data and binocular camera data enter positioning step after being initialized, and are patrolled by nonlinear optimization estimation
It examines robot pose and selects key frame and enter winding detection and map structuring step；It is corresponding to key frame in map structuring step
Binocular camera RGB image depth recovery is carried out using the method for cut tree, a three-dimensional map is constructed to cloud removal ground；It is returning
Ring detecting step carries out Global B A optimization, optimizes crusing robot pose and map when winding detection is judged as YES.
In data processing step: judging whether there is haze in crusing robot local environment using dark, mist figure
Model is expressed as:
I (x)=J (x) t (x)+A (1-t (x)),
Wherein I (x) is to mist elimination image, and J (x) is the fog free images to be restored, and A is global atmosphere light ingredient, and t (x) is
Transmissivity.The relationship in each channel of image can further indicate that are as follows:
Wherein C indicates each channel of image；For arbitrary image J, dark is defined as:
J in formulacIndicate each channel of color image, Ω (x) indicates each window；The estimated value of transmissivity solves are as follows:
W is constant parameter in formula, so that image is had the depth of field to retain a degree of mist, w often takes the value less than 1.Finally
Image recovery can indicate are as follows:
T in formula0For threshold value；Probability statistics are carried out to dark t (x) pixel of image, are total picture when pixel is greater than threshold value ζ
When the η ratio of element, image is as the processing of mist figure；Determine that ζ is 192, η 40%.
It should be noted that Fig. 2 is mist primitive figure of the invention, Fig. 3 is the dark channel diagram of mist figure of the invention, and Fig. 4 is
Mist figure defog effect figure of the invention；It follows that defogging processing method through the invention fully achieves defog effect.
In initialization step: the directly initialization if being able to carry out initialization of binocular camera data and inertial guidance data,
If not can be carried out initialization just first to initialize using binocular in fusion inertial guidance data.
In positioning step: generating a key frame must satisfy following several conditions: 1) from the generation of a upper key frame
At least through 15 frame images；2) at least 50 characteristic points are traced into；It 3) has been more than the threshold set at a distance from previous keyframe
Value；When crusing robot increases by two compulsory constraints when rotated: 1) must when robot short time rotation angle is more than 15 °
A key frame need be generated；2) it when robot is static, has within every 3 seconds generate a key frame.It can guarantee to track in this way
It will not lose, and improve the precision of tracking.Key frame for construct three-dimensional map when will further progress screening, pick
Except the biggish frame of similarity.
In map structuring step: carrying out depth recovery using cut tree, regard camera image I as connected undirected graph G=
(V,E).Wherein V indicates pixel collection in image I, and E is the set on side between pixel and pixel.For connected pixel s
It can be indicated with the side e weight of r are as follows:
we=w (s, r)=| I (s)-I (r) |,
One tree T can be made of a series of side E, and the weight between side can regard the depth cost of tree as.For
Pixel p and q, the existence anduniquess path between them in tree T, in path the summation of side right weight determine the distance between they D (p,
q)。Cd(p) matching cost of the pixel p on depth layer d is indicated.Non local polymerization cost CA d(p) C of weighting is indicateddSum.
Wherein q indicate image I in each pixel, S (p, q) indicate with pixel q in function to pixel to p in contribution
Weighting function；In tree construction, S (p, q) is defined as:
Wherein σ is the parameter for distance adjustment；In the polymerization of first cost, when all child nodes of pixel p are all interviewed
After asking, its cost value is indicated are as follows:
In the final cost polymerization carried out from root node to leaf node, the final cost polymerization cost of pixel p is by his father
Node Pr (p) is determined:
Image cost is recalculated, allows D to indicate left disparity map, then cost of the pixel p at the horizontal d of each parallax
Value:
It should be noted that Fig. 5 is left camera image of the invention；Fig. 6 is right camera image of the invention；Fig. 7
For subtree root node distribution map each in image of the invention；Fig. 8 is image segmentation figure of the invention；Fig. 9 is the present invention
Binocular estimating depth restore figure；It follows that the above method through the invention realizes depth recovery image.
In map structuring step, the pose of current key frame is T, and corresponding Lie algebra is ξ, available known depth
For zcPixel (u, v) corresponding three-dimensional points Pw are as follows:
The corresponding normal vector of the cloud are as follows:
The radius of initial error range are as follows:
R=zc/ f,
Initial confidence level are as follows:
C=exp (- γ2/2σ2),
Wherein, γ is the depth data after normalization；σ is generally taken as 0.6.
Interior point and exterior point are calculated using stochastical sampling consistency, interior point is defined as a certain range of apart from horizontal plane
Three-dimensional point, on the contrary it is exterior point.After filtering out ground point cloud to each key frame, the set P of these clouds is utilizedi(i=1 ...,
N) floor equation is solved:
Ax+by+cz+d=0,
It is wherein a, b, c are the coordinate of normal vector in Hessian normal form, and d is constant；Divide the interior point of extraction from cloud
It is in the corresponding plane of estimation parameter or with plan range in a certain range.
In winding detecting step: when winding detection is judged as YES, being optimized using Global B A；Define two frame of image it
Between relative pose:
When detecting that the frame in present frame and database constitutes closed loop, it is excellent that closed loop is carried out to all frames in database
Change.The residual error of i-th frame and jth frame is defined as:
All sequences while and overall goals function when closed loop is defined as:
Wherein S is sequence side, and L is closed loop side, and h is robust kernel function.
In positioning and winding detection, the image used is grayscale image, defines the pixel conversion of grayscale image and RGB figure
Are as follows: grayscale image=0.29*R+0.58*G+0.11*B；RGB image is used in map structuring step.
It should be noted that Figure 10 is the three-dimensional point cloud with ground of the invention；Figure 11 is removal ground of the invention
Three-dimensional point cloud；Figure 12 is three-dimensional map of the invention.It follows that the above method through the invention realizes dimensionally
Figure.
It can to sum up obtain, the invention discloses a kind of positioning of the autonomous crusing robot of binocular camera and inertial navigation fusion and three-dimensional
Map constructing method carries out denoising and pre- score to inertial guidance data, carries out carrying out feature after defogging processing to left and right camera
Extraction and optical flow tracking.Estimate that camera pose, screening key frame carry out winding detection and map structuring using nonlinear optimization.Ground
Figure construction step can carry out depth recovery by the left and right camera image of the corresponding binocular camera of key frame, find out known depth
Spend the three-dimensional coordinate of pixel.Above ground portion in point cloud is removed using stochastical sampling consistency, constructs three-dimensional map.By this hair
It is bright autonomous crusing robot accurately to be positioned, and construct three-dimensional environment map application and advised in further path
It draws.
Finally it should be noted that: the above embodiments are merely illustrative of the technical scheme of the present invention and are not intended to be limiting thereof；To the greatest extent
The present invention is described in detail with reference to preferred embodiments for pipe, it should be understood by those ordinary skilled in the art that: still
It can modify to a specific embodiment of the invention or some technical features can be equivalently replaced；Without departing from this hair
The spirit of bright technical solution should all cover within the scope of the technical scheme claimed by the invention.
Claims (10)
1. a kind of positioning of the autonomous crusing robot of binocular camera and inertial navigation fusion and three-dimensional map construction method, feature exist
In positioning is as follows with three-dimensional map construction method:
Inertial guidance data carries out pre- score processing after denoising, left and right camera carried out after defogging processing feature extraction and
Optical flow tracking；
Inertial guidance data and binocular camera data enter positioning step after being initialized, and estimate survey monitor by nonlinear optimization
Device people pose and select key frame enter winding detection and map structuring step；
Depth recovery is carried out using the method for cut tree to the corresponding binocular camera RGB image of key frame in map structuring step,
Three-dimensional map is constructed to cloud removal ground；
Global B A optimization is carried out, crusing robot pose and ground are optimized when winding detection is judged as YES in winding detecting step
Figure.
2. the positioning of the autonomous crusing robot of a kind of binocular camera according to claim 1 and inertial navigation fusion and three-dimensional map
Construction method, which is characterized in that defogging processing method is as follows:
Judge whether there is haze in crusing robot local environment using dark, mist graph model indicates are as follows:
I (x)=J (x) t (x)+A (1-t (x)),
Wherein I (x) is to mist elimination image, and J (x) is the fog free images to be restored, and A is global atmosphere light ingredient, and t (x) is transmission
Rate；
The relationship in each channel of image further indicates that are as follows:
Wherein C indicates each channel of image；For arbitrary image J, dark is defined as:
J in formulacIndicate each channel of color image, Ω (x) indicates each window；The estimated value of transmissivity solves are as follows:
W is constant parameter in formula, so that image is had the depth of field to retain a degree of mist, w often takes the value less than 1.
3. the positioning of the autonomous crusing robot of a kind of binocular camera according to claim 2 and inertial navigation fusion and three-dimensional map
Construction method, which is characterized in that image recovery method is as follows:
T in formula0For threshold value；Probability statistics are carried out to dark t (x) pixel of image, when pixel is greater than the η that threshold value ζ is total pixel
When ratio, image is as the processing of mist figure；Determine that ζ is 192, η 40%.
4. the positioning of the autonomous crusing robot of a kind of binocular camera according to claim 1 and inertial navigation fusion and three-dimensional map
Construction method, it is characterised in that: the directly initialization if being able to carry out initialization of binocular camera data and inertial guidance data, if not
It can be carried out initialization just first to initialize using binocular in fusion inertial guidance data.
5. the positioning of the autonomous crusing robot of a kind of binocular camera according to claim 1 and inertial navigation fusion and three-dimensional map
Construction method, which is characterized in that generating a key frame must satisfy following several conditions:
1) at least through 15 frame images from the generation of a upper key frame；
2) at least 50 characteristic points are traced into；
It 3) has been more than the threshold value set at a distance from previous keyframe.
6. the positioning of the autonomous crusing robot of a kind of binocular camera according to claim 5 and inertial navigation fusion and three-dimensional map
Construction method, which is characterized in that when crusing robot increases by two compulsory constraints when rotated:
1) it has to generate a key frame when robot short time rotation angle is more than 15 °；
2) it when robot is static, has within every 3 seconds generate a key frame.
7. the positioning of the autonomous crusing robot of a kind of binocular camera according to claim 1 and inertial navigation fusion and three-dimensional map
Construction method, it is characterised in that:
Regard camera image I as connected undirected graph G=(V, E) in map structuring step；For the side e power of connected pixel s and r
It indicates again are as follows:
we=w (s, r)=| I (s)-I (r) |；
Non local polymerization cost CA d(p) C of weighting is indicateddSum:
Wherein q indicate image I in each pixel, S (p, q) indicate with pixel q in function to pixel to p contribution weighting
Function；
In tree construction, S (p, q) is defined as:
Wherein σ is the parameter for distance adjustment；
Cost value of the pixel p at the horizontal d of each parallax:
8. the positioning of the autonomous crusing robot of a kind of binocular camera according to claim 1 and inertial navigation fusion and three-dimensional map
Construction method, it is characterised in that: in map structuring step, the pose of current key frame is T, and corresponding Lie algebra is ξ, depth
For zcPixel (u, v) corresponding three-dimensional points Pw are as follows:
The corresponding normal vector of the cloud are as follows:
The radius of initial error range are as follows:
R=zc/f；
Initial confidence level are as follows:
C=exp (- γ2/2σ2),
Wherein, γ is the depth data after normalization；σ is generally taken as 0.6.
9. the positioning of the autonomous crusing robot of a kind of binocular camera according to claim 1 and inertial navigation fusion and three-dimensional map
Construction method, it is characterised in that: interior point and exterior point are calculated using stochastical sampling consistency in map structuring step, to solve ground
Facial plane equation:
Ax+by+cz+d=0,
Wherein, a, b, c are the coordinate of normal vector in Hessian normal form, and d is constant.
10. the autonomous crusing robot of a kind of binocular camera according to claim 1 and inertial navigation fusion positioning with dimensionally
Figure construction method, it is characterised in that: define the relative pose between two frame of image in winding detecting step:
The residual error of i-th frame and jth frame is defined as:
All sequences while and overall goals function when closed loop is defined as:
Wherein S is sequence side, and L is closed loop side, and h is robust kernel function.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910042766.9A CN109764869A (en) | 2019-01-16 | 2019-01-16 | A kind of positioning of autonomous crusing robot and the three-dimensional map construction method of binocular camera and inertial navigation fusion |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910042766.9A CN109764869A (en) | 2019-01-16 | 2019-01-16 | A kind of positioning of autonomous crusing robot and the three-dimensional map construction method of binocular camera and inertial navigation fusion |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109764869A true CN109764869A (en) | 2019-05-17 |
Family
ID=66454120
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910042766.9A Pending CN109764869A (en) | 2019-01-16 | 2019-01-16 | A kind of positioning of autonomous crusing robot and the three-dimensional map construction method of binocular camera and inertial navigation fusion |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109764869A (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110196044A (en) * | 2019-05-28 | 2019-09-03 | 广东亿嘉和科技有限公司 | It is a kind of based on GPS closed loop detection Intelligent Mobile Robot build drawing method |
| CN110400267A (en) * | 2019-07-03 | 2019-11-01 | 安徽继远软件有限公司 | A kind of preprocess method based on inspection image |
| CN110633843A (en) * | 2019-08-23 | 2019-12-31 | 广州杰赛科技股份有限公司 | Park inspection method, device, equipment and storage medium |
| CN111897332A (en) * | 2020-07-30 | 2020-11-06 | 国网智能科技股份有限公司 | Semantic intelligent substation robot humanoid inspection operation method and system |
| CN112179361A (en) * | 2019-07-02 | 2021-01-05 | 华为技术有限公司 | Method, device and storage medium for updating work map of mobile robot |
| CN113587916A (en) * | 2021-07-27 | 2021-11-02 | 北京信息科技大学 | Real-time sparse visual odometer, navigation method and system |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107193279A (en) * | 2017-05-09 | 2017-09-22 | 复旦大学 | Robot localization and map structuring system based on monocular vision and IMU information |
| CN108460743A (en) * | 2018-03-19 | 2018-08-28 | 西安因诺航空科技有限公司 | A kind of unmanned plane image defogging algorithm based on dark |
| 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 |
-
2019
- 2019-01-16 CN CN201910042766.9A patent/CN109764869A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107193279A (en) * | 2017-05-09 | 2017-09-22 | 复旦大学 | Robot localization and map structuring system based on monocular vision and IMU information |
| CN108460743A (en) * | 2018-03-19 | 2018-08-28 | 西安因诺航空科技有限公司 | A kind of unmanned plane image defogging algorithm based on dark |
| 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 |
Non-Patent Citations (5)
| Title |
|---|
| CHANG CHEN，: ""A Review of Visual-Inertial Simultaneous Localization and Mapping from Filtering-Based and Optimization-Based Perspectives"", 《ROBOTICS》 * |
| QINGXIONG YANG: ""A Non-Local Cost Aggregation Method for Stereo Matching"", 《2012 IEEE CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION》 * |
| 张国良: ""考虑多位姿估计约束的双目视觉里程计"", 《控制与决策》 * |
| 张国良: "《移动机器人的SLAM与VSLAM方法》", 30 September 2018 * |
| 陶学宇: ""单目视觉移动机器人行走控制技术研究与实现"", 《中国优秀博硕士学位论文全文数据库(硕士) 信息科技辑》 * |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110196044A (en) * | 2019-05-28 | 2019-09-03 | 广东亿嘉和科技有限公司 | It is a kind of based on GPS closed loop detection Intelligent Mobile Robot build drawing method |
| CN112179361A (en) * | 2019-07-02 | 2021-01-05 | 华为技术有限公司 | Method, device and storage medium for updating work map of mobile robot |
| WO2021000630A1 (en) * | 2019-07-02 | 2021-01-07 | 华为技术有限公司 | Method and apparatus for updating working map of mobile robot, and storage medium |
| US11896175B2 (en) | 2019-07-02 | 2024-02-13 | Huawei Technologies Co., Ltd. | Method and apparatus for updating working map of mobile robot, and storage medium |
| CN110400267A (en) * | 2019-07-03 | 2019-11-01 | 安徽继远软件有限公司 | A kind of preprocess method based on inspection image |
| CN110633843A (en) * | 2019-08-23 | 2019-12-31 | 广州杰赛科技股份有限公司 | Park inspection method, device, equipment and storage medium |
| CN110633843B (en) * | 2019-08-23 | 2022-04-12 | 广州杰赛科技股份有限公司 | Park inspection method, device, equipment and storage medium |
| CN111897332A (en) * | 2020-07-30 | 2020-11-06 | 国网智能科技股份有限公司 | Semantic intelligent substation robot humanoid inspection operation method and system |
| CN113587916A (en) * | 2021-07-27 | 2021-11-02 | 北京信息科技大学 | Real-time sparse visual odometer, navigation method and system |
| CN113587916B (en) * | 2021-07-27 | 2023-10-03 | 北京信息科技大学 | Real-time sparse vision odometer, navigation method and system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109764869A (en) | A kind of positioning of autonomous crusing robot and the three-dimensional map construction method of binocular camera and inertial navigation fusion | |
| CN111739063B (en) | Positioning method of power inspection robot based on multi-sensor fusion | |
| CN110070615B (en) | Multi-camera cooperation-based panoramic vision SLAM method | |
| CN106840148B (en) | Wearable positioning and path guiding method based on binocular camera under outdoor working environment | |
| CN110097553B (en) | Semantic mapping system based on instant positioning mapping and three-dimensional semantic segmentation | |
| CN110044354A (en) | A kind of binocular vision indoor positioning and build drawing method and device | |
| CN105865454B (en) | A kind of Navigation of Pilotless Aircraft method generated based on real-time online map | |
| CN110262546A (en) | A kind of tunnel intelligent unmanned plane cruising inspection system and method | |
| KR20180079428A (en) | Apparatus and method for automatic localization | |
| CN107167139A (en) | A kind of Intelligent Mobile Robot vision positioning air navigation aid and system | |
| CN105953796A (en) | Stable motion tracking method and stable motion tracking device based on integration of simple camera and IMU (inertial measurement unit) of smart cellphone | |
| CN110553648A (en) | method and system for indoor navigation | |
| Xiao et al. | Monocular vehicle self-localization method based on compact semantic map | |
| CN109781092B (en) | Mobile robot positioning and mapping method in dangerous chemical engineering accident | |
| CN109848996A (en) | Extensive three-dimensional environment map creating method based on figure optimum theory | |
| CN111288989A (en) | Visual positioning method for small unmanned aerial vehicle | |
| CN110766785A (en) | Real-time positioning and three-dimensional reconstruction device and method for underground pipeline | |
| CN113188557B (en) | Visual inertial integrated navigation method integrating semantic features | |
| CN111595334A (en) | Indoor autonomous positioning method based on tight coupling of visual point-line characteristics and IMU (inertial measurement Unit) | |
| CN114019552A (en) | Bayesian multi-sensor error constraint-based location reliability optimization method | |
| Zhang et al. | ST-VIO: Visual-inertial odometry combined with image segmentation and tracking | |
| CN105738909B (en) | One kind being suitable for low latitude plant protection unmanned helicopter operation boundary extraction method | |
| CN108871409A (en) | A kind of fault detection method and system | |
| CN112179373A (en) | Measuring method of visual odometer and visual odometer | |
| Herath et al. | Fusion-dhl: Wifi, imu, and floorplan fusion for dense history of locations in indoor environments |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190517 |
|
| RJ01 | Rejection of invention patent application after publication |