CN109931940A - A kind of robot localization method for evaluating confidence based on monocular vision - Google Patents
A kind of robot localization method for evaluating confidence based on monocular vision Download PDFInfo
- Publication number
- CN109931940A CN109931940A CN201910056724.0A CN201910056724A CN109931940A CN 109931940 A CN109931940 A CN 109931940A CN 201910056724 A CN201910056724 A CN 201910056724A CN 109931940 A CN109931940 A CN 109931940A
- Authority
- CN
- China
- Prior art keywords
- model
- uncertainty
- integrity degree
- node
- localization 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.)
- Granted
Links
Landscapes
- Length Measuring Devices By Optical Means (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
The invention discloses a kind of robot localization method for evaluating confidence based on monocular vision, initially set up environment expression model, then equally distributed Parameter Expression form is used, the distributed model statistical modeling that the model of the foundation is parameterized, and compared with preferably model parameter is uniformly distributed, obtain the qualitative assessment of integrity degree;Then according to the mark that the information matrix between model interior joint and side has been established, the uncertainty that the node of model has been established is calculated;It is last that the weighted average of the two is carried out to some given position according to the integrity degree being calculated and uncertainty, complete positioning confidence level estimation.The present invention provides prior information for the global path planning of vision guided navigation, to realize that reliable, safety vision guided navigation provides Safety Factors Assessment early period.
Description
Technical field
The present invention relates to the technical field of robot localization more particularly to a kind of robot localizations based on monocular vision
Method for evaluating confidence.
Background technique
Positioning is realized in existing environmental model using monocular vision, is that mobile robot realizes the important of vision guided navigation
Component part.In vision guided navigation problem, the positioning in real time for obtaining mobile robot is most important.Current monocular vision navigation
Scheme often carries out global path planning using existing map, is realized and is moved using the matching that characteristic point and feature point description accord with
The vision positioning of mobile robot, and then realize according to planning path and the Navigation Control of positioning scenarios.
However in monocular vision orientation problem, vision data matching is the process of a not robust, therefore, insecure
An important factor for vision positioning is limitation vision guided navigation development.In order to solve this problem, current vision guided navigation path planning,
Other than range the considerations of barrier is included in global path planning, the confidence of vision positioning on the path is often also considered
Degree namely vision positioning some specific position potential feasibility, to realize that reliable vision is led on the particular path
Boat.
Summary of the invention
It is an object of the invention to overcome the deficiencies in the prior art, propose that a kind of robot localization based on monocular vision is set
Reliability appraisal procedure.This method using established environmental model as algorithm input, assess the uncertainty of the modeler model with
And integrity degree, so that the vision positioning confidence level to any of model position does off-line analysis.The it is proposed of this method is vision
The global path planning of navigation provides prior information, is to realize that reliable, safety vision guided navigation provides early period safely
Number assessment.
To achieve the above object, technical solution provided by the present invention are as follows:
A kind of robot localization method for evaluating confidence based on monocular vision, initially sets up environment expression model, then
Using equally distributed Parameter Expression form, to the distributed model statistical modeling that the model of the foundation is parameterized, and with
It is preferably uniformly distributed model parameter to compare, obtains the qualitative assessment of integrity degree;Then according to model interior joint has been established
The mark of information matrix between side calculates the uncertainty that the node of model has been established;It is last complete according to being calculated
Degree and uncertainty carry out the weighted average of the two to some given position, complete positioning confidence level estimation.
Further, topological form expression of the model of the foundation to scheme;Wherein the node expression of figure has been collected into
Each image;The relativeness of existing node is expressed on side in figure;An acquisition figure in the model of each node correspondence establishment
Picture;The foundation on side is calculated by the pixel data association of node corresponding image.
Further, specific step is as follows for the acquisition modeler model integrity degree qualitative assessment:
The model that environmental modeling is obtained carries out gridding processing, calculates each grid giThe angle appearance of middle acquired image
StateStatistical distribution;Count the angle set of all image postures in a gridMean valueAnd its standard deviation
And the mean value m for being distributed the two values and desired homogeneousidealWith standard deviation videalEuclidean distance is calculated, to obtain built
Mould model the grid gesture distribution between ideal gesture distribution at a distance from;Herein desired homogeneous distribution, be defined as U (-
π, π), thus the Parameter Expression m of the U (- π, π) distributionidealAnd videalIt can obtain;
According to this distance, assess the integrity degree of Image Acquisition in single grid: the two distance is bigger, illustrate currently to model and
The case where idealization models difference is bigger, thus its integrity degree is also lower;
For grid giIntegrity degreeCalculating is defined as:
Further, the uncertainty of some specific position, be defined as node corresponding to the acquisition image of the position with
The inverse of the sum of the mark of information matrix on all sides that other nodes are established, the uncertainty of the model interior joint is expressed asFor the information matrix on the side of each node link, mark numerical value is bigger, and uncertainty is lower.
Both further, the integrity degree and uncertainty that the basis is calculated, some given position is carried out
Weighted average, complete positioning confidence level estimation specific step is as follows:
Since node each in model possesses the posture information of three-dimensional space, information matrix is the square matrix of 6*6, this
Place's expression are as follows:
The then mark of information matrix corresponding to the side are as follows:
Assuming that the total quantity with the side of node link to be assessed is k, thenCalculating are as follows:
According to integrity degreeWith uncertaintyThe vision positioning confidence level of set point in modelCalculating definition
Are as follows:
WhereinWithFor weight, respectivelyWith
Compared with prior art, this programme principle and advantage is as follows:
This programme summarizes some given position as environmental model using integrity degree and the two weighted sum of uncertainty
Vision positioning confidence level estimation quantitative criteria.
Specially step are as follows:
Environment expression model is initially set up, equally distributed Parameter Expression form is then used, to the model of the foundation
The distributed model statistical modeling parameterized, and compared with preferably model parameter is uniformly distributed, obtain integrity degree
It is quantitatively evaluated;Then according to the mark that the information matrix between model interior joint and side has been established, the node that model has been established is calculated
Uncertainty;The last weighting according to the integrity degree being calculated and uncertainty, to both some given position progress
It is average, complete positioning confidence level estimation.
This programme provides prior information for the global path planning of vision guided navigation, for the vision for realizing reliable safety
Navigation provides Safety Factors Assessment early period.
Detailed description of the invention
Fig. 1 is a kind of flow chart of the robot localization method for evaluating confidence based on monocular vision of the present invention.
Specific embodiment
The present invention is further explained in the light of specific embodiments:
It is shown in Figure 1, a kind of robot localization method for evaluating confidence based on monocular vision described in the present embodiment,
It is specific as follows:
Initially set up environment expression model:
Topological form expression of the model of foundation to scheme;Wherein the node of figure indicates each image being collected into;In figure
Side, express the relativeness of existing node;An acquisition image in the model of each node correspondence establishment;The foundation on side is led to
The pixel data association for crossing node corresponding image is calculated.
Then equally distributed Parameter Expression form is used, the distributed model parameterized to the model of the foundation is united
Meter modeling, and compared with preferably model parameter is uniformly distributed, obtain the qualitative assessment of integrity degree;Specific step is as follows:
The model that environmental modeling is obtained carries out gridding processing, calculates each grid giThe angle appearance of middle acquired image
StateStatistical distribution;Count the angle set of all image postures in a gridMean valueAnd its standard deviation
And the mean value m for being distributed the two values and desired homogeneousidealWith standard deviation videalEuclidean distance is calculated, to obtain built
Mould model the grid gesture distribution between ideal gesture distribution at a distance from;Herein desired homogeneous distribution, be defined as U (-
π, π), thus the Parameter Expression m of the U (- π, π) distributionidealAnd videalIt can obtain;
According to this distance, assess the integrity degree of Image Acquisition in single grid: the two distance is bigger, illustrate currently to model and
The case where idealization models difference is bigger, thus its integrity degree is also lower;
For grid giIntegrity degreeCalculating is defined as:
Then according to the mark that the information matrix between model interior joint and side has been established, the node that model has been established is calculated
Uncertainty;
Specifically, the uncertainty of some specific position is defined as node corresponding to the acquisition image of the position and its
The inverse of the sum of the mark of information matrix on all sides that his node is established, the uncertainty of the model interior joint is expressed asFor the information matrix on the side of each node link, mark numerical value is bigger, and uncertainty is lower.
The integrity degree and uncertainty that last basis is calculated, the weighting for carrying out the two to some given position are flat
, positioning confidence level estimation is completed, the specific steps are as follows:
Since node each in model possesses the posture information of three-dimensional space, information matrix is the square matrix of 6*6, this
Place's expression are as follows:
The then mark of information matrix corresponding to the side are as follows:
Assuming that the total quantity with the side of node link to be assessed is k, thenCalculating are as follows:
According to integrity degreeWith uncertaintyThe vision positioning confidence level of set point in modelCalculating is defined as:
WhereinWithFor weight, respectivelyWith
The present embodiment summarizes some to positioning using integrity degree and the two weighted sum of uncertainty, as environmental model
The quantitative criteria for the vision positioning confidence level estimation set.Prior information is provided for the global path planning of vision guided navigation, is real
Now reliable, safety vision guided navigation provides Safety Factors Assessment early period.
The examples of implementation of the above are only the preferred embodiments of the invention, and implementation model of the invention is not limited with this
It encloses, therefore all shapes according to the present invention, changes made by principle, should all be included within the scope of protection of the present invention.
Claims (5)
1. a kind of robot localization method for evaluating confidence based on monocular vision, which is characterized in that initially set up environment expression
Then model uses equally distributed Parameter Expression form, the distributed model parameterized to the model of the foundation counts
Modeling, and compared with preferably model parameter is uniformly distributed, obtain the qualitative assessment of integrity degree;Then according to built formwork erection
The mark of information matrix between type interior joint and side calculates the uncertainty that the node of model has been established;Finally according to calculating
The integrity degree and uncertainty arrived carries out the weighted average of the two to some given position, completes positioning confidence level estimation.
2. a kind of robot localization method for evaluating confidence based on monocular vision according to claim 1, feature exist
In topological form expression of the model of the foundation to scheme;Wherein the node of figure indicates each image being collected into;In figure
The relativeness of existing node is expressed on side;An acquisition image in the model of each node correspondence establishment;The foundation on side, passes through
The pixel data association of node corresponding image is calculated.
3. a kind of robot localization method for evaluating confidence based on monocular vision according to claim 1, feature exist
In specific step is as follows for the acquisition modeler model integrity degree qualitative assessment:
The model that environmental modeling is obtained carries out gridding processing, calculates each grid giThe angular pose of middle acquired image
Statistical distribution;Count the angle set of all image postures in a gridMean valueAnd its standard deviationAnd it will
The mean value m of the two values and desired homogeneous distributionidealWith standard deviation videalEuclidean distance is calculated, to modeled mould
Type the grid gesture distribution between ideal gesture distribution at a distance from;Herein desired homogeneous distribution, be defined as U (- π,
π), thus the Parameter Expression m of the U (- π, π) distributionidealAnd videalIt can obtain;
According to this distance, assess the integrity degree of Image Acquisition in single grid: the two distance is bigger, illustrates currently to model and ideal
The case where changing modeling difference is bigger, thus its integrity degree is also lower;
For grid giIntegrity degreeCalculating is defined as:
4. a kind of robot localization method for evaluating confidence based on monocular vision according to claim 1, feature exist
In the uncertainty of some specific position is defined as node corresponding to the acquisition image of the position and is established with other nodes
The sum of the mark of information matrix on all sides inverse, the uncertainty of the model interior joint is expressed asFor each section
The information matrix on the side of point link, mark numerical value is bigger, and uncertainty is lower.
5. a kind of robot localization method for evaluating confidence based on monocular vision according to claim 1, feature exist
In the integrity degree and uncertainty that the basis is calculated carry out the weighted average of the two to some given position, complete
Positioning confidence level estimation, specific step is as follows:
Since node each in model possesses the posture information of three-dimensional space, information matrix is the square matrix of 6*6, herein table
It reaches are as follows:
The then mark of information matrix corresponding to the side are as follows:
Assuming that the total quantity with the side of node link to be assessed is k, thenCalculating are as follows:
According to integrity degreeWith uncertaintyThe vision positioning confidence level of set point in modelCalculating is defined as:
WhereinWithFor weight, respectivelyWith
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910056724.0A CN109931940B (en) | 2019-01-22 | 2019-01-22 | Robot positioning position reliability assessment method based on monocular vision |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910056724.0A CN109931940B (en) | 2019-01-22 | 2019-01-22 | Robot positioning position reliability assessment method based on monocular vision |
Publications (2)
Publication Number | Publication Date |
---|---|
CN109931940A true CN109931940A (en) | 2019-06-25 |
CN109931940B CN109931940B (en) | 2022-04-19 |
Family
ID=66985012
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910056724.0A Active CN109931940B (en) | 2019-01-22 | 2019-01-22 | Robot positioning position reliability assessment method based on monocular vision |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109931940B (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110738706A (en) * | 2019-09-17 | 2020-01-31 | 杭州电子科技大学 | quick robot vision positioning method based on track conjecture |
CN112258577A (en) * | 2020-10-26 | 2021-01-22 | 武汉中海庭数据技术有限公司 | Method and system for evaluating vehicle-end monocular vision mapping measurement confidence |
CN112325770A (en) * | 2020-10-26 | 2021-02-05 | 武汉中海庭数据技术有限公司 | Method and system for evaluating confidence of relative precision of monocular vision measurement at vehicle end |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20080063237A1 (en) * | 2006-09-08 | 2008-03-13 | Advanced Fuel Research, Inc. | Image analysis by object addition and recovery |
CN102402288A (en) * | 2010-09-07 | 2012-04-04 | 微软公司 | System for fast, probabilistic skeletal tracking |
CN103926927A (en) * | 2014-05-05 | 2014-07-16 | 重庆大学 | Binocular vision positioning and three-dimensional mapping method for indoor mobile robot |
CN104573646A (en) * | 2014-12-29 | 2015-04-29 | 长安大学 | Detection method and system, based on laser radar and binocular camera, for pedestrian in front of vehicle |
CN108647182A (en) * | 2018-05-11 | 2018-10-12 | 河南科技大学 | Based on the probability conversion method that can distribute degree of certainty in a kind of evidence theory |
CN109241855A (en) * | 2018-08-10 | 2019-01-18 | 西安交通大学 | Intelligent vehicle based on stereoscopic vision can travel area detection method |
-
2019
- 2019-01-22 CN CN201910056724.0A patent/CN109931940B/en active Active
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20080063237A1 (en) * | 2006-09-08 | 2008-03-13 | Advanced Fuel Research, Inc. | Image analysis by object addition and recovery |
CN102402288A (en) * | 2010-09-07 | 2012-04-04 | 微软公司 | System for fast, probabilistic skeletal tracking |
CN103926927A (en) * | 2014-05-05 | 2014-07-16 | 重庆大学 | Binocular vision positioning and three-dimensional mapping method for indoor mobile robot |
CN104573646A (en) * | 2014-12-29 | 2015-04-29 | 长安大学 | Detection method and system, based on laser radar and binocular camera, for pedestrian in front of vehicle |
CN108647182A (en) * | 2018-05-11 | 2018-10-12 | 河南科技大学 | Based on the probability conversion method that can distribute degree of certainty in a kind of evidence theory |
CN109241855A (en) * | 2018-08-10 | 2019-01-18 | 西安交通大学 | Intelligent vehicle based on stereoscopic vision can travel area detection method |
Non-Patent Citations (3)
Title |
---|
CHATELAIN, P.等: "Confidence-driven control of an ultrasound probe", 《IEEE TRANSACTIONS ON ROBOTICS》 * |
姚拓中: "结合主动学习的视觉场景理解", 《中国优秀博硕士学位论文全文数据库(博士)信息科技辑》 * |
靳太明: "基于双目视觉的运动目标深度信息提取方法研究", 《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 (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110738706A (en) * | 2019-09-17 | 2020-01-31 | 杭州电子科技大学 | quick robot vision positioning method based on track conjecture |
CN110738706B (en) * | 2019-09-17 | 2022-03-29 | 杭州电子科技大学 | Rapid robot visual positioning method based on track conjecture |
CN112258577A (en) * | 2020-10-26 | 2021-01-22 | 武汉中海庭数据技术有限公司 | Method and system for evaluating vehicle-end monocular vision mapping measurement confidence |
CN112325770A (en) * | 2020-10-26 | 2021-02-05 | 武汉中海庭数据技术有限公司 | Method and system for evaluating confidence of relative precision of monocular vision measurement at vehicle end |
Also Published As
Publication number | Publication date |
---|---|
CN109931940B (en) | 2022-04-19 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN112132972B (en) | Three-dimensional reconstruction method and system for fusing laser and image data | |
CN105096386B (en) | A wide range of complicated urban environment geometry map automatic generation method | |
CN107516326B (en) | Robot positioning method and system fusing monocular vision and encoder information | |
CN112859859A (en) | Dynamic grid map updating method based on three-dimensional obstacle object pixel object mapping | |
CN110146099B (en) | Synchronous positioning and map construction method based on deep learning | |
CN106338736B (en) | A kind of full 3D based on laser radar occupies volume elements terrain modeling method | |
CN106826813B (en) | A kind of hexapod robot stable motion control method | |
CN103020989B (en) | A kind of various visual angles method for tracking target based on online scene characteristic cluster | |
CN105865449A (en) | Laser and vision-based hybrid location method for mobile robot | |
CN106940704A (en) | A kind of localization method and device based on grating map | |
CN111539994A (en) | Particle filter repositioning method based on semantic likelihood estimation | |
CN109931940A (en) | A kind of robot localization method for evaluating confidence based on monocular vision | |
CN110223351B (en) | Depth camera positioning method based on convolutional neural network | |
CN103413352A (en) | Scene three-dimensional reconstruction method based on RGBD multi-sensor fusion | |
CN114820932B (en) | Panoramic three-dimensional scene understanding method based on graph neural network and relation optimization | |
CN113108773A (en) | Grid map construction method integrating laser and visual sensor | |
CN112818925B (en) | Urban building and crown identification method | |
CN108680177B (en) | Synchronous positioning and map construction method and device based on rodent model | |
CN111489392B (en) | Single target human motion posture capturing method and system in multi-person environment | |
CN116518960B (en) | Road network updating method, device, electronic equipment and storage medium | |
CN111998862A (en) | Dense binocular SLAM method based on BNN | |
CN113392584A (en) | Visual navigation method based on deep reinforcement learning and direction estimation | |
CN107564065A (en) | The measuring method of man-machine minimum range under a kind of Collaborative environment | |
CN109443200A (en) | A kind of mapping method and device of overall Vision coordinate system and mechanical arm coordinate system | |
CN111812978B (en) | Cooperative SLAM method and system for multiple unmanned aerial vehicles |
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 | ||
TA01 | Transfer of patent application right |
Effective date of registration: 20210826 Address after: 528253 room 3, 803, floor 8, block 3, Tian'an center, No. 31, Jihua East Road, Guicheng Street, Nanhai District, Foshan City, Guangdong Province (residence declaration) Applicant after: Jiutian innovation (Guangdong) Intelligent Technology Co.,Ltd. Address before: No. 100, Waihuan West Road, University Town, Guangzhou, Guangdong 510062 Applicant before: GUANGDONG University OF TECHNOLOGY |
|
TA01 | Transfer of patent application right | ||
GR01 | Patent grant | ||
GR01 | Patent grant |