CN109764876A - Multimodal fusion positioning method for unmanned platform - Google Patents
Multimodal fusion positioning method for unmanned platform Download PDFInfo
- Publication number
- CN109764876A CN109764876A CN201910130418.7A CN201910130418A CN109764876A CN 109764876 A CN109764876 A CN 109764876A CN 201910130418 A CN201910130418 A CN 201910130418A CN 109764876 A CN109764876 A CN 109764876A
- Authority
- CN
- China
- Prior art keywords
- positioning
- information
- positioning system
- matrix
- fusion
- 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
- 230000004927 fusion Effects 0.000 title claims abstract description 56
- 238000000034 method Methods 0.000 title claims abstract description 52
- 239000011159 matrix material Substances 0.000 claims abstract description 71
- 238000013528 artificial neural network Methods 0.000 claims abstract description 34
- 230000004807 localization Effects 0.000 claims description 20
- 238000012549 training Methods 0.000 claims description 13
- 238000013461 design Methods 0.000 claims description 7
- 238000013507 mapping Methods 0.000 claims description 7
- 238000009432 framing Methods 0.000 claims description 6
- 230000009466 transformation Effects 0.000 claims description 4
- 238000013527 convolutional neural network Methods 0.000 claims description 3
- 241000208340 Araliaceae Species 0.000 claims description 2
- 235000005035 Panax pseudoginseng ssp. pseudoginseng Nutrition 0.000 claims description 2
- 235000003140 Panax quinquefolius Nutrition 0.000 claims description 2
- 235000008434 ginseng Nutrition 0.000 claims description 2
- 238000013480 data collection Methods 0.000 claims 1
- 230000008569 process Effects 0.000 description 18
- 238000005516 engineering process Methods 0.000 description 10
- 238000010586 diagram Methods 0.000 description 6
- 230000006870 function Effects 0.000 description 5
- 238000012545 processing Methods 0.000 description 5
- 230000008901 benefit Effects 0.000 description 2
- 230000008878 coupling Effects 0.000 description 2
- 238000010168 coupling process Methods 0.000 description 2
- 238000005859 coupling reaction Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 238000002156 mixing Methods 0.000 description 2
- 206010044565 Tremor Diseases 0.000 description 1
- 230000000295 complement effect Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 238000007499 fusion processing Methods 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 230000001537 neural effect Effects 0.000 description 1
- 238000007500 overflow downdraw method Methods 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Landscapes
- Control Of Position, Course, Altitude, Or Attitude Of Moving Bodies (AREA)
Abstract
本发明提供了一种无人平台的多模态融合定位方法。该方法包括：在无人平台人中搭载多个定位系统，分别学习出每个定位系统的用于描述误差模型的神经网络参数，根据所述神经网络参数得到用以输出定位算法的误差信息矩阵，基于各个定位系统的信息矩阵得到各个定位系统输出的定位结果；将每个定位系统的定位结果与信息矩阵输入到信息滤波器，所示信息滤波器输出所述无人平台的融合定位结果。本发明实施例的方法可以有效地融合多模块定位系统输出，获得精准稳定的融合定位结果，为移动机器人等无人平台执行其他复杂任务提供了基础。可扩展性强,可应用于多个模块融合定位。
The invention provides a multi-modal fusion positioning method of an unmanned platform. The method includes: carrying a plurality of positioning systems in an unmanned platform person, learning neural network parameters for describing an error model of each positioning system respectively, and obtaining an error information matrix for outputting a positioning algorithm according to the neural network parameters , based on the information matrix of each positioning system, the positioning results output by each positioning system are obtained; the positioning results and information matrix of each positioning system are input into the information filter, and the shown information filter outputs the fusion positioning result of the unmanned platform. The method of the embodiment of the present invention can effectively fuse the outputs of the multi-module positioning system, obtain accurate and stable fusion positioning results, and provide a basis for unmanned platforms such as mobile robots to perform other complex tasks. It has strong scalability and can be applied to fusion positioning of multiple modules.
Description
Technical field
The present invention relates to field of locating technology more particularly to a kind of multi-modal fusion localization methods of unmanned platform.
Background technique
In the development process of mobile robot, location technology realizes the basic technology of all kinds of complex tasks for it, such as leads
Performing effectively for the tasks such as boat, planning all relies on accurately positioning input.In recent years, with unmanned field and sensor
The development of technology has emerged a variety of positioning strategies for relying on disparate modules sensor.Due to the survey of different types of sensor
The difference of information and location algorithm is measured, single module location technology respectively has superiority and inferiority, and intelligent vehicle platform need to merge multimode in complex scene
Block location technology exports location information to ensure reliable and stablely.
Existing multimode localization method is mostly rule of thumb design rule progress module switching, is substantially equivalent to single mode
Block positioning takes turns to operate, once it is i.e. no longer reliable to be detached from experience usable condition.Though Bayes Fusion Method based on probability
It can realize that information merges in itself, but its basic condition is to require each module positioning to have reliable error model, and require
Each module, which is located in fusion process, to be mutually adapted.For existing single module location technology, except limited several as inertia is led
Boat, dead reckoning etc. can be estimated outside its error model with calibration technique and Propagation Theory, be based on environment sensing for majority
Location technology for, error model is related often with standby scene since its internal nonlinearity characteristic processing link is difficult to export
Property, as laser positioning method is more accurate in the scene of feature rich.
For multimode location technology, the purpose is to realize that modules fusion is complementary, but still lacks now and effectively melt
Conjunction means, thus multi-source positioning fusion problem is still a great problem of positioning field.
Summary of the invention
The embodiment provides a kind of multi-modal fusion localization methods of unmanned platform, to overcome the prior art
Problem.
To achieve the goals above, this invention takes following technical solutions.
A kind of multi-modal fusion localization method of unmanned platform, comprising:
Carry multiple positioning systems in unmanned platform people, learn respectively each positioning system out for describing error mould
The neural network parameter of type obtains being based on to the control information matrix for exporting location algorithm according to the neural network parameter
The information matrix of each positioning system obtains the positioning result of each positioning system output；
The positioning result of each positioning system and information matrix are input to information filter, shown information filter output
The fusion positioning result of the unmanned platform.
Preferably, the neural network parameter for being used to describe error model for learning each positioning system out respectively,
The control information matrix to export location algorithm is obtained according to the neural network parameter, comprising:
The input data for acquiring the location algorithm of each positioning system makes the training number of each positioning system using input data
According to collection, it is directed to the corresponding neural network of each Location System Design respectively, each positioning system is indicated in the form of neural network
Positioning scene data utilize the training dataset and positioning scene number of each positioning system to the mapping relations between error model
According to the neural network for being used to describe error model for the mapping relations between error model, learning each positioning system out respectively
Parameter obtains the control information matrix to export location algorithm according to the neural network parameter.
Preferably, when the positioning system is laser odometer, the use for learning each positioning system out respectively
In the neural network parameter of description error model, believed according to the error that the neural network parameter is obtained to export location algorithm
Cease matrix, comprising:
Single frames two dimensional laser scanning data projection to two-dimensional surface is obtained by single frames and is accounted for by two-dimensional localization orientation problem
There is grating map, grating map is inputted into CNN neural network, exports 6 independent elements, it will be under 6 independent elements composition
The lower triangular matrix is multiplied by triangular matrix with its transposed matrix, positive semidefinite information matrix is obtained, by the positive semidefinite information square
Information matrix of the battle array as the laser odometer.
Preferably, the positioning result by each positioning system and information matrix are input to information filter, shown
Information filter output fusion positioning result, comprising:
Information filter obtains the information vector and information matrix of former frame fusion positioning result, by information vector and information
Matrixing was set to the local coordinate system of zero point to former framing bit, will it is described before framing bit be set to the local coordinate system of zero point and each
The positioning result and information matrix of a positioning system are input to information filter, and the fusion that information filter exports present frame is opposite
Positioning.
Preferably, the positioning result by each positioning system and information matrix are input to information filter, shown
Information filter output fusion positioning result, comprising:
The positioning system of certain known error model is selected as fusion reference position system, for positioning system to be learned,
Merge reference position system and the stateful relationship of positioning system to be learned:
Wherein xtIndicate the relative positioning value relative to preceding frame position, utFor the relatively fixed of fusion reference position system output
Position estimation, ztFor the relative positioning estimation of positioning system to be learned output, εtFor the relative positioning estimation for merging reference position system
Gaussian error, RtFor the covariance matrix for merging reference position system, δtFor positioning system to be learned relative positioning estimate
Gaussian error, QtFor the covariance matrix of Positioning System Error to be learned；
If the two-dimensional localization input data of t moment information filter is zt,ut,Rt, then the information filter of information filter
Wave fusion steps include:
1. carrying out coordinate system transformation
2. carrying out information matrix prediction
3. carrying out information vector prediction
4. carrying out information matrix update according to observation
5. carrying out information vector update according to observation
6. obtaining fusion positioning solution
Export μt,Ωt,ξt
Wherein, for two-dimensional localization solution μt=(μx,t,μy,t,μθ,t)TThere is two-dimensional localization coordinate transform battle array
ξtFor the information vector in information filter, there is ξt=Ωtμt。
As can be seen from the technical scheme provided by the above-mentioned embodiment of the present invention, the method for the embodiment of the present invention can be effective
Ground merges the output of multimode positioning system, obtains the metastable fusion positioning result of essence, holds for unmanned platforms such as mobile robots
Other complex tasks of row provide the foundation.Scalability is strong, can be applied to multiple module fusion positioning.
The additional aspect of the present invention and advantage will be set forth in part in the description, these will become from the following description
Obviously, or practice through the invention is recognized.
Detailed description of the invention
In order to illustrate the technical solution of the embodiments of the present invention more clearly, required use in being described below to embodiment
Attached drawing be briefly described, it should be apparent that, drawings in the following description are only some embodiments of the invention, for this
For the those of ordinary skill of field, without creative efforts, it can also be obtained according to these attached drawings others
Attached drawing.
Fig. 1 is a kind of realization principle signal of multi-modal fusion localization method of unmanned platform provided in an embodiment of the present invention
Figure；
Fig. 2 is a kind of process flow diagram of the multi-modal fusion localization method of unmanned platform provided in an embodiment of the present invention；
Fig. 3 is a kind of realization principle figure of the learning method of the information matrix of positioning system provided in an embodiment of the present invention；
Fig. 4 is the process schematic that a kind of information filter provided in an embodiment of the present invention carries out use processing:
Fig. 5 is a kind of process flow diagram of the scene of laser odometer-error map algorithm provided in an embodiment of the present invention；
Fig. 6 is a kind of error model of the scene of laser odometer-error map algorithm provided in an embodiment of the present invention
Practise the process flow diagram of algorithm.
Specific embodiment
Embodiments of the present invention are described below in detail, the example of the embodiment is shown in the accompanying drawings, wherein from beginning
Same or similar element or element with the same or similar functions are indicated to same or similar label eventually.Below by ginseng
The embodiment for examining attached drawing description is exemplary, and for explaining only the invention, and is not construed as limiting the claims.
Those skilled in the art of the present technique are appreciated that unless expressly stated, singular " one " used herein, " one
It is a ", " described " and "the" may also comprise plural form.It is to be further understood that being arranged used in specification of the invention
Diction " comprising " refer to that there are the feature, integer, step, operation, element and/or component, but it is not excluded that in the presence of or addition
Other one or more features, integer, step, operation, element, component and/or their group.It should be understood that when we claim member
Part is " connected " or when " coupled " to another element, it can be directly connected or coupled to other elements, or there may also be
Intermediary element.In addition, " connection " used herein or " coupling " may include being wirelessly connected or coupling.Wording used herein
"and/or" includes one or more associated any cells for listing item and all combinations.
Those skilled in the art of the present technique are appreciated that unless otherwise defined, all terms used herein (including technology
Term and scientific term) there is meaning identical with the general understanding of those of ordinary skill in fields of the present invention.Also answer
It should be appreciated that those terms such as defined in the general dictionary should be understood that have in the context of the prior art
The consistent meaning of meaning, and unless defined as here, it will not be explained in an idealized or overly formal meaning.
In order to facilitate understanding of embodiments of the present invention, it is done by taking several specific embodiments as an example below in conjunction with attached drawing further
Explanation, and each embodiment does not constitute the restriction to the embodiment of the present invention.
Embodiment one
The embodiment of the present invention positions fusion problem for the multimode of mobile robot, proposes a kind of based on machine learning
End-to-end error model learning method.This method can learn all kinds of positioning systems based on Bayesian filter fusion frame
The error model of system output carries out positioning fusion using the error model, and fusion positioning solution is precisely reliable, the continuous nothing of positioning track
It trembles.In fusion frame, without carrying out complicated error model design for each disparate modules and assuming, it is ensured that each module
It works independently.In learning process, without relying on continuous, high-precision positioning true value information, only rely on rough grade track starting point,
Endpoint information can carry out once learning renewal process, therefore can be used for unmanned platform and carry out on-line study.
A kind of realization principle schematic diagram of the multi-modal fusion localization method of mobile robot provided in an embodiment of the present invention
As shown in Figure 1, specific process flow is as shown in Figure 2, comprising the following steps:
The first step acquires each positioning if carrying N in mobile robot covers positioning system in mobile robot usage scenario
The input data of the location algorithm of system utilizes the training dataset of each positioning system of the data creating of acquisition；
It is opposite to learn each positioning system as unit of frame respectively for second step, the positioning fusion benchmark for selecting each positioning system
The information matrix of previous frame, the information matrix are used to describe the error distribution of positioning system.It is provided in an embodiment of the present invention a kind of fixed
The realization principle of the learning method of the information matrix of position system is as shown in figure 3, include following treatment process: each positioning system of acquisition
The input data of the location algorithm of system makes the training dataset of each positioning system using input data.Due to position error one
As it is related with the scene information in the input of its algorithm, for arbitrarily needing the positioning system X of learning error model, respectively for each
The corresponding neural network of Location System Design indicates the positioning scene data of each positioning system to error in the form of neural network
Mapping relations between model.Then, the training dataset of each positioning system and positioning scene data to error model are utilized
Between mapping relations, learn the information matrix of the opposite previous frame of each positioning system out.
Learn the treatment process packet of the information matrix of each positioning system out using training dataset and above-mentioned mapping relations
It includes:
1. the positioning system A (such as inertial navigation system) of certain known error model is selected as fusion baseline system, if its
Data input is D1, positioning system A output phase is to positioning result and error co-variance matrix u, and (covariance matrix is information square to R
The inverse matrix of battle array)；
2. couple any positioning system X for needing learning error model (sets its data to input as D2, export positioning result and error
Information matrix is z, Q) initialize its error model neural network.Choose the scene related data that may influence its position error
D2-sceneIt is inputted as neural network, such as visual odometry, that is, it is input that realtime graphic, which may be selected, and output is error letter
Cease matrix(can refer to embodiment two)；
3. pair random length is the training data Traj of T, if the global position true value of its terminus isIt is desirable
Obtain the positioning input data D of t frame in data1,t,D2,tAnd neural network required input data D2-scene,t, in neural network
The relative positioning result and its error co-variance matrix <u of continuous each frame can be obtained under Parameter Conditions Θt,Rt>,<zt,Qt> (suitable
Information matrix need to be converted to), the relative positioning result of continuous each frame and its error co-variance matrix are inputted into fusion shown in FIG. 1
Frame can be obtained each frame fusion relative positioning result and information matrix < μt,Ωt>.Using coordinate system transformation,Position base
To μ on plinth1,μ2,...,μTIt is accumulative to carry out relative positioning, can be obtained the Global localization position under Current Situation of Neural Network parameter
Estimation
4. set loss function asSeek the J value of training data Traj.Neural network parameter, which may be updated, isWherein loss function can need to design according to experiment, such as can be
5. the different training data track of more than loading, parameter learning can be carried out by repeating step 3-4.Reach setting most
After big frequency of training, i.e. study terminates.At this point, the neural network parameter can be used, for any D2-scene,tIt maps out available
Information matrix
Then, the information matrix based on each positioning system obtains the positioning result of each positioning system output.
Then third step is merged the positioning result of each positioning system with information matrix input information filter
Positioning result.Information filter uses the error model learning outcome of each positioning system, realizes fusion positioning, output fusion positioning
As a result.Information filter can online updating in application process, continue to optimize error model.
A kind of information filter provided in an embodiment of the present invention carries out process schematic such as Fig. 4 institute of use processing
Show, including following treatment process:
Information filter obtains the information vector and information matrix of former frame fusion positioning result, by information vector and information
Matrixing was set to the local coordinate system of zero point to former framing bit, will it is above-mentioned before framing bit be set to the local coordinate system of zero point and each
The positioning result and information matrix of a positioning system are input to information filter, and the fusion that information filter exports present frame is opposite
Positioning.
Information filter is the exemplary filter in existing document, and the specific work process of information filter includes: to the
Two positioning systems in two steps, i.e. fusion reference position system and the stateful relationship of positioning system to be learned.
Wherein xtIndicate the relative positioning value relative to preceding frame position, utFor the relatively fixed of fusion reference position system output
Position estimation, ztFor the relative positioning estimation of positioning system to be learned output, εt,δtRespectively indicate the relative positioning of two positioning systems
The Gaussian error of estimation, the covariance matrix of error are respectively Rt,Qt。
If t moment obtains two-dimensional positioning data and inputs zt,ut,Rt, then time information filtering fusion steps are as follows:
7. carrying out coordinate system transformation
8. carrying out information matrix prediction
9. carrying out information vector prediction
10. carrying out information matrix update according to observation
11. carrying out information vector update according to observation
12. obtaining fusion positioning solution
Export μt,Ωt,ξt
Wherein, for two-dimensional localization solution μt=(μx,t,μy,t,μθ,t)TThere is two-dimensional localization (3DOF) coordinate transform battle array
ξtFor the information vector in information filter, there is ξt=Ωtμt
Embodiment two
The embodiment is described in detail by taking the study of the error model of laser odometer as an example for technical detail.Due to boat
The error model of position predication method has steadiness, can be obtained by initial sensor calibration according to Propagation Theory relatively reliable
Error model, therefore the error model in dead reckoning is selected to merge benchmark, it is merged with laser odometer.
A kind of scene-error map algorithm process flow such as Fig. 5 institute of laser odometer provided in an embodiment of the present invention
Show, concrete processing procedure includes: that single frames two dimensional laser scanning data projection to two-dimensional surface is obtained single frames with occupying grid
Grating map is inputted CNN neural network, exports 6 independent elements by figure.Then, 6 independent elements are formed into lower three angular moments
Battle array, which is multiplied with its transposed matrix, acquisition positive semidefinite information matrix, using the positive semidefinite information matrix as
The information matrix of the laser odometer.
Fig. 6 is a kind of error model of the scene of laser odometer-error map algorithm provided in an embodiment of the present invention
The process flow diagram of algorithm is practised, concrete processing procedure includes:
Step 1: initialization error Model Neural parameter, setting maximum number of iterations are M, the number of iterations i=0,
Set accumulative step number T
Step 2: being loaded into i-th trained track, data and track head and the tail section global position true value needed for obtaining T frame alignment
Step 3: initialization starting point global position estimation initializes relative positioning blending algorithm, and positioning has added up step number t=
0
Step 4: calculating dead reckoning relative positioning result and information matrix
Step 5: calculating laser mileage relative positioning result
Step 6: laser odometer information matrix is calculated according to laser odometer scene-error map algorithm
Step 7: calculating laser odometer using relative positioning blending algorithm and merge solution with the positioning of dead reckoning
Step 8: by t=0 moment global position true value, obtaining the accumulative global position of present frame
Whether the current accumulative step number t of step 9, judgement is less than accumulative step number T, if so, returning to step 4；Otherwise,
Execute step 10；
Step 10, the global position true value for obtaining the T moment calculate target loss function
Network graphics drawing is unfolded step 11 according to the time, carries out backpropagation T step using loss function value, obtains tired
Count gradient
Step 12 uses accumulative gradient updating network parameter
Step 13 judges whether the number of iterations i is less than maximum number of iterations for M, if so, returning to step 2；It is no
Then, process terminates.
In conclusion the method for the embodiment of the present invention can effectively merge the output of multimode positioning system, obtain accurate
Stable fusion positioning result, fills up the blank both at home and abroad in terms of multimode positions fusion, and for mobile robot etc., nobody is flat
Platform executes other complex tasks and provides the foundation.
Error model learning method proposed by the present invention carries out under probability fusion frame completely, solves in the prior art
Above-mentioned significant problem, have an advantage that 1) eliminating complicated error model design derives work, can be with positioning accuracy
Target carries out end-to-end study, and 2) scalability is strong, can be applied to multiple module fusions and positions, and 3) for carrying pre-training network
Multimode position unmanned platform for, the application prospect with on-line study, 4) syncretizing effect is good, can get higher positioning
Precision, positioning track are smooth.
Those of ordinary skill in the art will appreciate that: attached drawing is the schematic diagram of one embodiment, module in attached drawing or
Process is not necessarily implemented necessary to the present invention.
All the embodiments in this specification are described in a progressive manner, same and similar portion between each embodiment
Dividing may refer to each other, and each embodiment focuses on the differences from other embodiments.Especially for device or
For system embodiment, since it is substantially similar to the method embodiment, so describing fairly simple, related place is referring to method
The part of embodiment illustrates.Apparatus and system embodiment described above is only schematical, wherein the conduct
The unit of separate part description may or may not be physically separated, component shown as a unit can be or
Person may not be physical unit, it can and it is in one place, or may be distributed over multiple network units.It can root
According to actual need that some or all of the modules therein is selected to achieve the purpose of the solution of this embodiment.Ordinary skill
Personnel can understand and implement without creative efforts.
The foregoing is only a preferred embodiment of the present invention, but scope of protection of the present invention is not limited thereto,
In the technical scope disclosed by the present invention, any changes or substitutions that can be easily thought of by anyone skilled in the art,
It should be covered by the protection scope of the present invention.Therefore, protection scope of the present invention should be with scope of protection of the claims
Subject to.
Claims (5)
1. a kind of multi-modal fusion localization method of unmanned platform characterized by comprising
Carry multiple positioning systems in unmanned platform people, learn respectively each positioning system out for describing error model
Neural network parameter obtains the control information matrix to export location algorithm according to the neural network parameter, based on each
The information matrix of positioning system obtains the positioning result of each positioning system output；
The positioning result of each positioning system and information matrix are input to information filter, described in shown information filter output
The fusion positioning result of unmanned platform.
2. the method according to claim 1, wherein it is described learn respectively each positioning system out for retouching
The neural network parameter for stating error model obtains the control information square to export location algorithm according to the neural network parameter
Battle array, comprising:
The input data for acquiring the location algorithm of each positioning system makes the training data of each positioning system using input data
Collection is directed to the corresponding neural network of each Location System Design respectively, determining for each positioning system is indicated in the form of neural network
Position contextual data utilizes the training dataset and positioning scene data of each positioning system to the mapping relations between error model
To the mapping relations between error model, learn the neural network ginseng for describing error model of each positioning system out respectively
Number, obtains the control information matrix to export location algorithm according to the neural network parameter.
3. according to the method described in claim 2, it is characterized in that, when the positioning system be laser odometer when, it is described
The neural network parameter for being used to describe error model for learning each positioning system out respectively, obtains according to the neural network parameter
To the control information matrix to export location algorithm, comprising:
Single frames two dimensional laser scanning data projection to two-dimensional surface is obtained by single frames and occupies grid by two-dimensional localization orientation problem
Grating map is inputted CNN neural network, exports 6 independent elements by lattice map, and 6 independent elements are formed lower triangle
The lower triangular matrix is multiplied by matrix with its transposed matrix, obtains positive semidefinite information matrix, and the positive semidefinite information matrix is made
For the information matrix of the laser odometer.
4. according to claim 1 or method described in 2 or 3, which is characterized in that the positioning by each positioning system
As a result information filter, shown information filter output fusion positioning result are input to information matrix, comprising:
Information filter obtains the information vector and information matrix of former frame fusion positioning result, by information vector and information matrix
Be converted into the local coordinate system that former framing bit is set to zero point, will it is described before framing bit be set to the local coordinate system of zero point and each fixed
The positioning result and information matrix of position system are input to information filter, and the fusion that information filter exports present frame is relatively fixed
Position.
5. according to the method described in claim 4, it is characterized in that, the positioning result and information by each positioning system
For Input matrix to information filter, shown information filter, which exports, merges positioning result, comprising:
The positioning system of certain known error model is selected as fusion reference position system, for positioning system to be learned, fusion
Reference position system and the stateful relationship of positioning system to be learned:
Wherein xtIndicate the relative positioning value relative to preceding frame position, utRelative positioning for fusion reference position system output is estimated
Meter, ztFor the relative positioning estimation of positioning system to be learned output, εtTo merge the height that the relative positioning of reference position system is estimated
This error, RtFor the covariance matrix for merging reference position system, δtThe Gauss estimated for the relative positioning of positioning system to be learned
Error, QtFor the covariance matrix of Positioning System Error to be learned；
If the two-dimensional localization input data of t moment information filter is zt,ut,Rt, then the information filter fusion of information filter
Step includes:
1. carrying out coordinate system transformation
2. carrying out information matrix prediction
3. carrying out information vector prediction
4. carrying out information matrix update according to observation
5. carrying out information vector update according to observation
6. obtaining fusion positioning solution
Export μt,Ωt,ξt
Wherein, for two-dimensional localization solution μt=(μx,t,μy,t,μθ,t)TThere is two-dimensional localization coordinate transform battle array
ξtFor the information vector in information filter, there is ξt=Ωtμt。
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910130418.7A CN109764876B (en) | 2019-02-21 | 2019-02-21 | Multi-mode fusion positioning method of unmanned platform |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910130418.7A CN109764876B (en) | 2019-02-21 | 2019-02-21 | Multi-mode fusion positioning method of unmanned platform |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109764876A true CN109764876A (en) | 2019-05-17 |
| CN109764876B CN109764876B (en) | 2021-01-08 |
Family
ID=66456303
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910130418.7A Expired - Fee Related CN109764876B (en) | 2019-02-21 | 2019-02-21 | Multi-mode fusion positioning method of unmanned platform |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109764876B (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111367282A (en) * | 2020-03-09 | 2020-07-03 | 山东大学 | A robot navigation method and system based on multimodal perception and reinforcement learning |
| CN111680596A (en) * | 2020-05-29 | 2020-09-18 | 北京百度网讯科技有限公司 | Positioning truth value verification method, device, equipment and medium based on deep learning |
| CN113543305A (en) * | 2020-04-22 | 2021-10-22 | 维沃移动通信有限公司 | Positioning method, communication equipment and network equipment |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103983263A (en) * | 2014-05-30 | 2014-08-13 | 东南大学 | Inertia/visual integrated navigation method adopting iterated extended Kalman filter and neural network |
| CA2973038A1 (en) * | 2015-12-01 | 2017-06-08 | China University Of Mining And Technology | Combined inertial navigation and laser scanning coal shearer positioning device and method |
| CN106980133A (en) * | 2017-01-18 | 2017-07-25 | 中国南方电网有限责任公司超高压输电公司广州局 | The GPS INS Combinated navigation methods and system for being compensated and being corrected using neural network algorithm |
| CN108692701A (en) * | 2018-05-28 | 2018-10-23 | 佛山市南海区广工大数控装备协同创新研究院 | Mobile robot Multi-sensor Fusion localization method based on particle filter |
| CN108716917A (en) * | 2018-04-16 | 2018-10-30 | 天津大学 | A kind of indoor orientation method merging inertia and visual information based on ELM |
| CN108871336A (en) * | 2018-06-20 | 2018-11-23 | 湘潭大学 | A kind of vehicle location estimating system and method |
| CN109059912A (en) * | 2018-07-31 | 2018-12-21 | 太原理工大学 | A kind of GPS/INS integrated positioning method based on wavelet neural network |
-
2019
- 2019-02-21 CN CN201910130418.7A patent/CN109764876B/en not_active Expired - Fee Related
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103983263A (en) * | 2014-05-30 | 2014-08-13 | 东南大学 | Inertia/visual integrated navigation method adopting iterated extended Kalman filter and neural network |
| CA2973038A1 (en) * | 2015-12-01 | 2017-06-08 | China University Of Mining And Technology | Combined inertial navigation and laser scanning coal shearer positioning device and method |
| CN106980133A (en) * | 2017-01-18 | 2017-07-25 | 中国南方电网有限责任公司超高压输电公司广州局 | The GPS INS Combinated navigation methods and system for being compensated and being corrected using neural network algorithm |
| CN108716917A (en) * | 2018-04-16 | 2018-10-30 | 天津大学 | A kind of indoor orientation method merging inertia and visual information based on ELM |
| CN108692701A (en) * | 2018-05-28 | 2018-10-23 | 佛山市南海区广工大数控装备协同创新研究院 | Mobile robot Multi-sensor Fusion localization method based on particle filter |
| CN108871336A (en) * | 2018-06-20 | 2018-11-23 | 湘潭大学 | A kind of vehicle location estimating system and method |
| CN109059912A (en) * | 2018-07-31 | 2018-12-21 | 太原理工大学 | A kind of GPS/INS integrated positioning method based on wavelet neural network |
Non-Patent Citations (4)
| Title |
|---|
| HUAQUN GUO: "《Automotive Informatics and Communicative Systems: Principles in Vehicular Networks and Data Exchange》", 30 April 2009 * |
| SIAVASH HOSSEINYALAMDARY: "Deep Kalman Filter: Simultaneous Multi-Sensor Integration and Modelling; A GNSS/IMU Case Study", 《SENSORS》 * |
| 刘洞波: "《移动机器人粒子滤波定位与地图创建》", 30 September 2016 * |
| 吴晓晓: "基于信息融合的室外移动机器人定位理论及技术研究", 《中国优秀硕士学位论文全文数据库》 * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111367282A (en) * | 2020-03-09 | 2020-07-03 | 山东大学 | A robot navigation method and system based on multimodal perception and reinforcement learning |
| CN113543305A (en) * | 2020-04-22 | 2021-10-22 | 维沃移动通信有限公司 | Positioning method, communication equipment and network equipment |
| CN111680596A (en) * | 2020-05-29 | 2020-09-18 | 北京百度网讯科技有限公司 | Positioning truth value verification method, device, equipment and medium based on deep learning |
| CN111680596B (en) * | 2020-05-29 | 2023-10-13 | 北京百度网讯科技有限公司 | Positioning true value verification method, device, equipment and medium based on deep learning |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109764876B (en) | 2021-01-08 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Shen et al. | Observability analysis and adaptive information fusion for integrated navigation of unmanned ground vehicles | |
| Hasberg et al. | Simultaneous localization and mapping for path-constrained motion | |
| Fang et al. | Graph optimization approach to range-based localization | |
| CN107084714B (en) | A multi-robot collaborative target positioning method based on RoboCup3D | |
| CN108810133A (en) | A kind of intelligent robot localization method and positioning system based on UWB and TDOA algorithms | |
| CN106772524B (en) | A Rank Filter Based Integrated Navigation Information Fusion Method for Agricultural Robots | |
| CN110726406A (en) | An Improved Nonlinear Optimization Method for Monocular Inertial Navigation SLAM | |
| CN108319293A (en) | A kind of UUV Realtime collision free planing methods based on LSTM networks | |
| CN107727098A (en) | A kind of unmanned water surface ship paths planning method for multiple target locations of patrolling successively | |
| Tully et al. | Leap-frog path design for multi-robot cooperative localization | |
| Yuan et al. | An entorhinal-hippocampal model for simultaneous cognitive map building | |
| CN109764876A (en) | Multimodal fusion positioning method for unmanned platform | |
| CN103369466A (en) | Map matching-assistant indoor positioning method | |
| CN103644917B (en) | Traverse measurement platform laser radar rotation and translation calculation method of parameters | |
| CN112857370A (en) | Robot map-free navigation method based on time sequence information modeling | |
| CN105300383A (en) | Unmanned aerial vehicle air refueling position and attitude estimation method based on backtracking and searching | |
| He et al. | Camera-odometer calibration and fusion using graph based optimization | |
| CN114879660A (en) | Robot environment sensing method based on target driving | |
| Neloy et al. | Alpha-N-V2: Shortest path finder automated delivery robot with obstacle detection and avoiding system | |
| Li et al. | Exploring the Potential of the Deep-Learning-Aided Kalman Filter for GNSS/INS Integration: A Study on 2-D Simulation Datasets | |
| Huang et al. | Neural inertial navigation system on pedestrian | |
| Wang et al. | Applying SLAM algorithm based on nonlinear optimized monocular vision and IMU in the positioning method of power inspection robot in complex environment | |
| Xu et al. | A novel heading angle estimation methodology for land vehicles based on deep learning and enhanced digital map | |
| Wang et al. | Information‐Fusion Methods Based Simultaneous Localization and Mapping for Robot Adapting to Search and Rescue Postdisaster Environments | |
| Olson | Landmark selection for terrain matching |
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 | ||
| CF01 | Termination of patent right due to non-payment of annual fee | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20210108 |