CN108680175A - Synchronous superposition method and device based on rodent models - Google Patents
Synchronous superposition method and device based on rodent models Download PDFInfo
- Publication number
- CN108680175A CN108680175A CN201810265940.1A CN201810265940A CN108680175A CN 108680175 A CN108680175 A CN 108680175A CN 201810265940 A CN201810265940 A CN 201810265940A CN 108680175 A CN108680175 A CN 108680175A
- Authority
- CN
- China
- Prior art keywords
- access point
- wireless access
- wifi
- current
- rodent models
- 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
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01C—MEASURING DISTANCES, LEVELS OR BEARINGS; SURVEYING; NAVIGATION; GYROSCOPIC INSTRUMENTS; PHOTOGRAMMETRY OR VIDEOGRAMMETRY
- G01C21/00—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00
- G01C21/26—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00 specially adapted for navigation in a road network
- G01C21/28—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00 specially adapted for navigation in a road network with correlation of data from several navigational instruments
- G01C21/30—Map- or contour-matching
- G01C21/32—Structuring or formatting of map data
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01C—MEASURING DISTANCES, LEVELS OR BEARINGS; SURVEYING; NAVIGATION; GYROSCOPIC INSTRUMENTS; PHOTOGRAMMETRY OR VIDEOGRAMMETRY
- G01C21/00—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00
- G01C21/26—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00 specially adapted for navigation in a road network
- G01C21/34—Route searching; Route guidance
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W4/00—Services specially adapted for wireless communication networks; Facilities therefor
- H04W4/02—Services making use of location information
Landscapes
- Engineering & Computer Science (AREA)
- Radar, Positioning & Navigation (AREA)
- Remote Sensing (AREA)
- Automation & Control Theory (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Position Fixing By Use Of Radio Waves (AREA)
Abstract
The present invention provides a kind of synchronous superposition method and device based on rodent models, and method includes：Obtain the Current vision scene image information of robot；According to the rodent models built in advance, the posture information that there is maximum scene similarity with Current vision scene image information is matched from the visual information base built in advance；When scene similarity is less than given threshold, the current WIFI signal intensity set of robot is obtained；According to rodent models, the posture information that there is maximum fingerprint similarity with current WIFI signal intensity set is matched from the WIFI fingerprint maps built in advance；Positioning and map structuring are synchronized to robot according to maximum scene similarity and/maximum fingerprint similarity corresponding posture information.
Description
Technical field
The present invention relates to robot control field more particularly to a kind of synchronous positioning and ground based on rodent models
Figure construction method and device.
Background technology
Synchronous superposition is the great difficult problem that mobile robot faces at present.Because mobile robot is substantial
The sensor platform exactly moved, although sensor type and ability are had nothing in common with each other, be widely present odometer drift and not
With noise the problems such as.Constantly probing by scholars later, bio-robot (using the robot of bionics techniques control)
Good application prospect is gradually highlighted, perfect biorational and the high degree of adaptability to natural environment are shown.Wherein,
Bio-robot uses rodent models to carry out bionics techniques realization mostly.
Wherein, rodent models perceive visual odometry information and the integrated appearance in place of visual scene image information
In cell model, so that mobile robot has certain update predictive ability, and time, spatial position, row are set up
For etc. information experience figure.Currently, rodent models are widely used in the location navigation work of robot, solve
Numerous synchronous superpositions (Simultaneous Localization and Mapping, MAP) are insoluble to ask
Topic, but rodent models obtain visual scene image information and visual odometry information exist it is a degree of
Error, although there is scholar to introduce FAB-MAP (fast appearance based for the error of visual odometry
Mapping), this closed loop detection algorithm based on historical models can improve the steady of system by the matching of real time critical frame
Precision and unstable that is qualitative, but positioning, and robustness is not strong.So individual rodent models are in positioning accuracy
And it is remained to be further improved in terms of robustness.
Invention content
The technical problem to be solved in the present invention is to provide a kind of to synchronize positioning and map structure based on rodent models
Construction method and device blend WIFI fingerprint technology and the bionical location technology based on rodent models, realize contraposition
The amendment of appearance cellular network then realizes being accurately positioned to robot to obtain optimal path experience figure.
In order to solve the above technical problems, technical solution provided by the invention is：
In a first aspect, the present invention provides a kind of synchronous superposition method based on rodent models, side
Method includes：
Obtain the Current vision scene image information of robot；
According to the rodent models built in advance, matched from the visual information base built in advance and Current vision
Scene image information has the posture information of maximum scene similarity；
When scene similarity is less than given threshold, the current WIFI signal intensity set of robot is obtained；
According to rodent models, matched from the WIFI fingerprint maps built in advance and current WIFI signal intensity
Gathering has the posture information of maximum fingerprint similarity；
Positioning is synchronized to robot according to maximum scene similarity posture information corresponding with/maximum fingerprint similarity
With map structuring.
Further, method further includes：
Visual information base is updated according to maximum scene similarity；
WIFI fingerprint maps are updated according to maximum fingerprint similarity.
Further, the current WIFI signal intensity set of robot is obtained, including：
Determine valid wireless access point；
Receive the current received signal strength mean value that each valid wireless access point is generated in current location；
The corresponding current received signal strength mean value of all valid wireless access points is determined as the current of robot
WIFI signal intensity set.
Further, it is determined that valid wireless access point, including：
Determine the quantity of valid wireless access point；
Two wireless access point are selected from the wireless access point set that all wireless access point form as reference at random
Access point；
Two are calculated with reference to the first mutual information between access point；
The wireless access point of the second mutual information minimum can be made by being obtained from wireless access point set；
The wireless access point of third mutual information minimum can be made by being obtained from wireless access point set；
And so on, until getting sufficient amount of wireless access point.
Further, according to rodent models, matched from the WIFI fingerprint maps built in advance with currently
WIFI signal intensity set has the posture information of maximum fingerprint similarity, including：
Machine is obtained according to current WIFI signal intensity set, and according to the Bayesian posterior estimation model built in advance
The estimated location of people；
According to estimated location, WIFI fingerprint maps, and according to the pose cellular network in rodent models, carry out
Posture information matches, to match the posture information for having maximum fingerprint similarity with current WIFI signal intensity set.
Further, pose cellular network carries out posture information matching, including：
According to estimated location, at least one experience unit adjacent with estimated location is extracted from WIFI fingerprint maps；
Calculate the Euclidean distance between the WIFI signal intensity and current WIFI signal intensity of each experience unit；
The posture information pointed by the corresponding experience unit of maximum Euclidean distance is obtained, and the posture information got is true
It is set to the current posture information of robot.
Further, the calculation formula of Euclidean distance is：
Wherein, (xpc,ypc,θpc) it is the corresponding pose cell coordinate of experience unit；(xi,yi,θi) be and current location pair
The pose cell coordinate answered；raFor the zonal constant of (x, y) plane, θaFor the zonal constant in θ dimensions.
Further, the structure of WIFI fingerprints map, including：
Selected reference point；
In each reference point, it is averaging after carrying out continuous sampling to the signal strength of each preset wireless access point,
To obtain received signal strength mean value of each wireless access point at reference point；
According to the received signal strength mean value of the corresponding all wireless access point of each reference point, structure is carried out according to preset rules
It builds WIFI fingerprints map and stores.
Further, the data store organisation of WIFI fingerprints map is：
IM={ φ, A, M, MACi}；Wherein,
φ={ L1,L2,…,Li,…,Lk}；A={ AP1,AP2,…,APi,…,APR}；
MACiIndicate the MAC Address value of i-th of reference point；
Wherein, IM indicates WIFI fingerprint maps；Li=(xi,yi) indicate i-th of reference point position, k be reference point number
Amount, φ indicate the location sets of all reference points, indicate that the composition set of all wireless access point observed in map, R are
The quantity of the wireless access point observed, M are the set for the received signal strength mean value that each reference point answers each wireless access point,
WhereinIt is the R wireless access point in reference point LkThe received signal strength mean value at place.
Second aspect, the present invention provide a kind of robot synchronous superposition dress based on rodent models
It sets, device includes：
Information acquisition unit, the Current vision scene image information for obtaining robot；
First matching unit, for the rodent models that basis is built in advance, from the visual information base built in advance
In match and Current vision scene image information has the posture information of maximum scene similarity；
Data determining unit, for when scene similarity is less than given threshold, obtaining the current WIFI signal of robot
Intensity set；
Second matching unit, for according to rodent models, being matched from the WIFI fingerprint maps built in advance
There is the posture information of maximum fingerprint similarity with current WIFI signal intensity set；
Synchronous positioning unit, for according to maximum scene similarity and the corresponding posture information pair of/maximum fingerprint similarity
Robot synchronizes positioning and map structuring.
Synchronous superposition method and device provided by the invention based on rodent models, WIFI is referred to
Line technology is blended with the bionical location technology based on rodent models, the amendment to pose cellular network is realized, to obtain
Optimal path experience figure is obtained, then realizes being accurately positioned to robot.
Description of the drawings
Fig. 1 is the flow chart of method provided in an embodiment of the present invention；
Fig. 2 is the block diagram of device provided in an embodiment of the present invention；
Fig. 3 is utilization WIFI fingerprint matching procedure relation schematic diagrames provided in an embodiment of the present invention.
Specific implementation mode
It is further illustrated the present invention below by specific embodiment, it should be understood, however, that, these embodiments are only
It is used for specifically describing in more detail, and is not to be construed as limiting the present invention in any form.
Embodiment one
In conjunction with Fig. 1, the synchronous superposition method provided in this embodiment based on rodent models, method
Including：
Step S1 obtains the Current vision scene image information of robot；
Step S2, according to the rodent models built in advance, matched from the visual information base built in advance with
Current vision scene image information has the posture information of maximum scene similarity；
Step S3 obtains the current WIFI signal intensity set of robot when scene similarity is less than given threshold；
Step S4 is matched and current WIFI according to rodent models from the WIFI fingerprint maps built in advance
Signal strength set has the posture information of maximum fingerprint similarity；
Step S5 carries out robot according to maximum scene similarity posture information corresponding with/maximum fingerprint similarity
Synchronous superposition.
Synchronous superposition method provided in an embodiment of the present invention based on rodent models, WIFI is referred to
Line technology is blended with the bionical location technology based on rodent models, the amendment to pose cellular network is realized, to obtain
Optimal path experience figure is obtained, then realizes being accurately positioned to robot.
Preferably, method further includes：
Visual information base is updated according to maximum scene similarity；
WIFI fingerprint maps are updated according to maximum fingerprint similarity.
In the present embodiment, specifically, maximum scene similarity is R1, and maximum fingerprint similarity is R2, λ1, λ2To set in advance
Two fixed scene similarity thresholds, ε1, ε2For preset two fingerprint similarity thresholds, ε1＜ ε2.More specifically,
Work as R1＞ λ2When, successful match need not carry out WIFI fingerprint matchings, is not updated to visual information base at this time；Work as R1＜ λ1
When, it fails to match, is updated to visual information base；λ1＜ R1＜ λ2And R2＞ ε2When, successful match, without update；Work as λ1
＜ R1＜ λ2And ε1＜ R2＜ ε2When, successful match, without update；Work as λ1＜ R1＜ λ2And R2＜ ε1When, it fails to match, right
WIFI fingerprint maps are updated.
It is further preferred that the current WIFI signal intensity set of robot is obtained, including：
Determine valid wireless access point；
Receive the current received signal strength mean value that each valid wireless access point is generated in current location；
The corresponding current received signal strength mean value of all valid wireless access points is determined as the current of robot
WIFI signal intensity set.
It, can be in conjunction with actual needs to the quantity of valid wireless access point (Access Point, AP) in the present embodiment
It is set, in this way, the dimension of signal space can be made to reduce, reduces calculation amount.Specifically, in the present embodiment, for example, certain is indoor
Under environment, there are T available AP, need therefrom to choose best S effective AP, after effective AP is determined, successively
Measure each reference point from different AP received signal strength value (Received Signal Strength Indication,
RSSI) the signal characteristic number as reference point AP, and be recorded in location fingerprint database by certain format, the database
It is referred to as location fingerprint map or WIFI fingerprint images.
Specifically, the structure of WIFI fingerprints map, including：
Selected reference point；
In each reference point, it is averaging after carrying out continuous sampling to the signal strength of each preset wireless access point,
To obtain received signal strength mean value of each wireless access point at reference point；
According to the received signal strength mean value of the corresponding all wireless access point of each reference point, structure is carried out according to preset rules
It builds WIFI fingerprints map and stores.
Further specifically, the data store organisation of WIFI fingerprints map is：
IM={ φ, A, M, MACi}；Wherein,
φ={ L1,L2,…,Li,…,Lk}；A={ AP1,AP2,…,APi,…,APR}；
MACiIndicate the MAC Address value of i-th of reference point；
Wherein, IM indicates WIFI fingerprint maps；Li=(xi,yi) indicate i-th of reference point position, k be reference point number
Amount, φ indicate the location sets of all reference points, indicate that the composition set of all wireless access point observed in map, R are
The quantity of the wireless access point observed, M are the set for the received signal strength mean value that each reference point answers each wireless access point,
WhereinIt is the R wireless access point in reference point LkThe received signal strength mean value at place.
In the present embodiment, the method for building up of WIFI fingerprint images is as follows under indoor environment, according to a set pattern in localizing environment
Then choose reference point, and in each reference point to the signal strength continuous sampling of AP for a period of time, obtain the mean value of each APIt is stored in database, constitutes location fingerprint figure IM, whereinWherein,
Li=(xi,yi) indicate reference point position, k be reference point quantity,Indicate the location sets of all reference points；A={ AP1,
AP2,...,APRIndicate all composition set for observing AP in map.
M is the set of all mean values in location fingerprint, whereinIt is j-th of AP in reference point LiThe mean value at place, MACi
The MAC Address value of i-th of reference point is indicated, specifically, shown in the data structure such as formula (1) of M.
More specifically, valid wireless access point is determined, including：
Determine the quantity of valid wireless access point；
Two wireless access point are selected from the wireless access point set that all wireless access point form as reference at random
Access point；
Two are calculated with reference to the first mutual information between access point；
The wireless access point of the second mutual information minimum can be made by being obtained from wireless access point set；
The wireless access point of third mutual information minimum can be made by being obtained from wireless access point set；
And so on, until getting sufficient amount of wireless access point.
In the present embodiment, more specifically, setting available AP number of reference point of indoor positioning environment is T, chooses wherein S
The dimension of signal space then can be dropped to S dimensions by the majorized subset of a AP from T dimensions, thus can reduce calculation amount.Specifically, originally
Embodiment uses minimum mutual information AP Selection Strategies, and is as follows：
1) combination of two is carried out for S AP of selection, calculates the mutual information each combined according to the following formula, finds out mutual trust
Cease minimum combination, corresponding APm, APnAs two initial reference point AP；
MI(APm,APn)=H (APm)+H(APn)-H(APm,APn) (2)
In formula (2)：MI(APm,APn) indicate two difference AP mutual information, that is, the first mutual information, H (APm,APn) table
Show the combined information entropy of two AP.
2) mutual information that certain AP is combined with two initial AP is calculated according to formula (3).
MI(APm,APn,APi)=H (APm,APn)+H(APi)-H(APm,APn,APi) (3)
Find out can so that MI minimum AP as optimization AP subsets third AP.
3) next optimal AP is chosen according to the form of the 2) step successively, successively iteration, until selecting S optimal AP
Until.It should be noted that the selection formula of R optimal AP is as follows：
MI(AP1,AP2,…,APR)=H (AP1,AP2,…,APR-1)+H(APR)-H(APm,APn,…,APR) (4)
Preferably, it according to rodent models, is matched and current WIFI from the WIFI fingerprint maps built in advance
Signal strength set has the posture information of maximum fingerprint similarity, including：
Machine is obtained according to current WIFI signal intensity set, and according to the Bayesian posterior estimation model built in advance
The estimated location of people；
According to estimated location, WIFI fingerprint maps, and according to the pose cellular network in rodent models, carry out
Posture information matches, to match the posture information for having maximum fingerprint similarity with current WIFI signal intensity set.
In the present embodiment, using Bayes's location estimation strategy, and specifically, for above-mentioned minimum mutual information AP choosings
Strategy is taken, Bayesian posterior estimation is further used and is combined optimization so that the location estimation essence of WIFI fingerprinting localization algorithms
Degree and reliability greatly promote.
The basic principle of Bayesian posterior estimation is shown below：
In formula：RSSI indicate multiple AP position estimation point RSSI observations；p(Li| RSSI) indicate position LiTo
Determine the conditional probability under RSSI, i.e., in the case where observing RSSI vectors, anchor point appears in LiProbability；p(RSSI|Li)
Indicate position LiProbability；p(Li) indicate position LiProbability, do not consider the difference between fingerprint point usually, i.e., fingerprint point etc. is general
Rate；P (RSSI) indicates that the full probability that RSSI occurs, formula are as follows：
Wherein, C (RSSI1, RSSI2 ..., RSSIM) indicates the number for the specified RSSI vectors that fingerprint point observes；K tables
Show fingerprint point epoch of observation number.
By above-mentioned full probability formula back substitution to Bayesian posterior estimator, to calculate posteriority conditional probability.Using more
Bayes's weight location estimation formula of a fingerprint point can calculate the position of location estimation point within a short period of time, enable estimation point
Position be p, then the calculation formula of estimated location is as follows：
In formula：(x, y) indicates the two-dimensional coordinate of location estimation point, (xi,yi) indicate i-th of fingerprint point coordinate, ωiTable
The weighting weight of i-th of fingerprint point, the as probability of Bayesian posterior condition are levied, K indicates neighbor point number
Preferably, pose cellular network carries out posture information matching, including：
According to estimated location, at least one experience unit adjacent with estimated location is extracted from WIFI fingerprint maps；
Calculate the Euclidean distance between the WIFI signal intensity and current WIFI signal intensity of each experience unit；
The posture information pointed by the corresponding experience unit of maximum Euclidean distance is obtained, and the posture information got is true
It is set to the current posture information of robot.
It is further preferred that the calculation formula of Euclidean distance is：
Wherein, (xpc,ypc,θpc) it is the corresponding pose cell coordinate of experience unit；(xi,yi,θi) be and current location pair
The pose cell coordinate answered；raFor the zonal constant of (x, y) plane, θaFor the zonal constant in θ dimensions.
In the present embodiment, each unit tool that undergoes there are one activity level, activity level by pose perception cell and
Degree of closeness is determined between energy peak and each experience unit in WIFI fingerprints.Each experience pose perception cell and
There are one relevant active regions in WIFI fingerprints.When energy peak is in these active regions, which is activated at once, this
A little regions are continuous in pose perception cell interior, and the relevant range in WIFI fingerprints is discrete.Each warp
Go through eiBy experience activity level Ei, WIFI signal intensity RiIt is determined.Wherein, ei={ Ei,Ri,
The energy level E of one experience unitxyθWith total energy level E of i-th of experience unitiBy formula (9) and formula (10)
It can be calculated.
Wherein, xpcypcAnd θpcFor the coordinate of maximum activity posture cell；xi、yi、θiFor with the relevant pose sense of the experience
Know the coordinate of cell；raFor the zonal constant of (x, y) plane；θaFor the zonal constant in θ dimensions.RcurrIt is strong for current WIFI signal
Degree；RiFor with experience the relevant WIFI signal intensity of i.
It should be noted that in the present embodiment, as illustrated in fig. 3, WIFI fingerprint matching procedure relation schematic diagrames are utilized.It will
Wireless signal network WIFI is used as a kind of sensor in rodent models, and there are three major parts for location model
Composition, respectively WIFI fingerprints, pose cellular network and experience are schemed.WIFI fingerprints obtain the WIFI signal intensity of environment, are claimed
For WIFI signal intensity template.WIFI finger print informations are used for recognizing known environment.When the WIFI signal strength information newly inputted
When with already present WIFI signal intensity template matches, the active factors of pose cellular network are activated, and the two combines can be very
The generation that erroneous matching is prevented in big degree generates more accurate experience figure.
Embodiment two
In conjunction with Fig. 2, the robot provided in an embodiment of the present invention based on rodent models synchronizes positioning and map structure
Device is built, device includes：
Information acquisition unit 1, the Current vision scene image information for obtaining robot；
First matching unit 2, for the rodent models that basis is built in advance, from the visual information base built in advance
In match and Current vision scene image information has the posture information of maximum scene similarity；
Data determining unit 3, for when scene similarity is less than given threshold, obtaining the current WIFI signal of robot
Intensity set；
Second matching unit 4, for according to rodent models, being matched from the WIFI fingerprint maps built in advance
There is the posture information of maximum fingerprint similarity with current WIFI signal intensity set；
Synchronous positioning unit 5, for according to maximum scene similarity and the corresponding posture information pair of/maximum fingerprint similarity
Robot synchronizes positioning and map structuring.
Synchronous superposition device provided in an embodiment of the present invention based on rodent models, WIFI is referred to
Line technology is blended with the bionical location technology based on rodent models, the amendment to pose cellular network is realized, to obtain
Optimal path experience figure is obtained, then realizes being accurately positioned to robot.
Preferably, device further includes data updating unit, is used for
Visual information base is updated according to maximum scene similarity；
WIFI fingerprint maps are updated according to maximum fingerprint similarity.
In the present embodiment, specifically, maximum scene similarity is R1, and maximum fingerprint similarity is R2, λ1, λ2To set in advance
Two fixed scene similarity thresholds, ε1, ε2For preset two fingerprint similarity thresholds, ε1＜ ε2.More specifically,
Work as R1＞ λ2When, successful match need not carry out WIFI fingerprint matchings, is not updated to visual information base at this time；Work as R1＜ λ1
When, it fails to match, is updated to visual information base；λ1＜ R1＜ λ2And R2＞ ε2When, successful match, without update；Work as λ1
＜ R1＜ λ2And ε1＜ R2＜ ε2When, successful match, without update；Work as λ1＜ R1＜ λ2And R2＜ ε1When, it fails to match, right
WIFI fingerprint maps are updated.
It is further preferred that the current WIFI signal intensity set of robot is obtained, including：
Determine valid wireless access point；
Receive the current received signal strength mean value that each valid wireless access point is generated in current location；
The corresponding current received signal strength mean value of all valid wireless access points is determined as the current of robot
WIFI signal intensity set.
It, can be in conjunction with actual needs to the quantity of valid wireless access point (Access Point, AP) in the present embodiment
It is set, in this way, the dimension of signal space can be made to reduce, reduces calculation amount.Specifically, in the present embodiment, for example, certain is indoor
Under environment, there are T available AP, need therefrom to choose best S effective AP, after effective AP is determined, successively
Measure each reference point from different AP received signal strength value (Received Signal Strength Indication,
RSSI) the signal characteristic number as reference point AP, and be recorded in location fingerprint database by certain format, the database
It is referred to as location fingerprint map or WIFI fingerprint images.
Specifically, the structure of WIFI fingerprints map, including：
Selected reference point；
In each reference point, it is averaging after carrying out continuous sampling to the signal strength of each preset wireless access point,
To obtain received signal strength mean value of each wireless access point at reference point；
According to the received signal strength mean value of the corresponding all wireless access point of each reference point, structure is carried out according to preset rules
It builds WIFI fingerprints map and stores.
Further specifically, the data store organisation of WIFI fingerprints map is：
IM={ φ, A, M, MACi}；Wherein,
φ={ L1,L2,…,Li,…,Lk}；A={ AP1,AP2,…,APi,…,APR}；
MACiIndicate the MAC Address value of i-th of reference point；
Wherein, IM indicates WIFI fingerprint maps；Li=(xi,yi) indicate i-th of reference point position, k be reference point number
Amount, φ indicate the location sets of all reference points, indicate that the composition set of all wireless access point observed in map, R are
The quantity of the wireless access point observed, M are the set for the received signal strength mean value that each reference point answers each wireless access point,
WhereinIt is the R wireless access point in reference point LkThe received signal strength mean value at place.
In the present embodiment, the method for building up of WIFI fingerprint images is as follows under indoor environment, according to a set pattern in localizing environment
Then choose reference point, and in each reference point to the signal strength continuous sampling of AP for a period of time, obtain the mean value of each AP
It is stored in database, constitutes location fingerprint figure IM, whereinWherein,Li
=(xi,yi) indicate reference point position, k be reference point quantity,Indicate the location sets of all reference points；A={ AP1,
AP2,...,APRIndicate all composition set for observing AP in map.
M is the set of all mean values in location fingerprint, whereinIt is j-th of AP in reference point LiThe mean value at place, MACi
The MAC Address value of i-th of reference point is indicated, specifically, shown in the data structure such as formula (1) of M.
More specifically, valid wireless access point is determined, including：
Determine the quantity of valid wireless access point；
Two wireless access point are selected from the wireless access point set that all wireless access point form as reference at random
Access point；
Two are calculated with reference to the first mutual information between access point；
The wireless access point of the second mutual information minimum can be made by being obtained from wireless access point set；
The wireless access point of third mutual information minimum can be made by being obtained from wireless access point set；
And so on, until getting sufficient amount of wireless access point.
In the present embodiment, more specifically, setting available AP number of reference point of indoor positioning environment is T, chooses wherein S
The dimension of signal space then can be dropped to S dimensions by the majorized subset of a AP from T dimensions, thus can reduce calculation amount.Specifically, originally
Embodiment uses minimum mutual information AP Selection Strategies, and is as follows：
1) combination of two is carried out for S AP of selection, calculates the mutual information each combined according to the following formula, finds out mutual trust
Cease minimum combination, corresponding APm, APnAs two initial reference point AP；
MI(APm,APn)=H (APm)+H(APn)-H(APm,APn) (2)
In formula (2)：MI(APm,APn) indicate two difference AP mutual information, that is, the first mutual information, H (APm,APn) table
Show the combined information entropy of two AP.
2) mutual information that certain AP is combined with two initial AP is calculated according to formula (3).
MI(APm,APn,APi)=H (APm,APn)+H(APi)-H(APm,APn,APi) (3)
Find out can so that MI minimum AP as optimization AP subsets third AP.
3) next optimal AP is chosen according to the form of the 2) step successively, successively iteration, until selecting S optimal AP
Until.It should be noted that the selection formula of R optimal AP is as follows：
MI(AP1,AP2,…,APR)=H (AP1,AP2,…,APR-1)+H(APR)-H(APm,APn,…,APR) (4)
Preferably, the second matching unit 4 is specifically used for,
Machine is obtained according to current WIFI signal intensity set, and according to the Bayesian posterior estimation model built in advance
The estimated location of people；
According to estimated location, WIFI fingerprint maps, and according to the pose cellular network in rodent models, carry out
Posture information matches, to match the posture information for having maximum fingerprint similarity with current WIFI signal intensity set.
In the present embodiment, using Bayes's location estimation strategy, and specifically, for above-mentioned minimum mutual information AP choosings
Strategy is taken, Bayesian posterior estimation is further used and is combined optimization so that the location estimation essence of WIFI fingerprinting localization algorithms
Degree and reliability greatly promote.
The basic principle of Bayesian posterior estimation is shown below：
In formula：RSSI indicate multiple AP position estimation point RSSI observations；p(Li| RSSI) indicate position LiTo
Determine the conditional probability under RSSI, i.e., in the case where observing RSSI vectors, anchor point appears in LiProbability；p(RSSI|Li)
Indicate position LiProbability；p(Li) indicate position LiProbability, do not consider the difference between fingerprint point usually, i.e., fingerprint point etc. is general
Rate；P (RSSI) indicates that the full probability that RSSI occurs, formula are as follows：
Wherein, C (RSSI1, RSSI2 ..., RSSIM) indicates the number for the specified RSSI vectors that fingerprint point observes；K tables
Show fingerprint point epoch of observation number.
By above-mentioned full probability formula back substitution to Bayesian posterior estimator, to calculate posteriority conditional probability.Using more
Bayes's weight location estimation formula of a fingerprint point can calculate the position of location estimation point within a short period of time, enable estimation point
Position be p, then the calculation formula of estimated location is as follows：
In formula：(x, y) indicates the two-dimensional coordinate of location estimation point, (xi,yi) indicate i-th of fingerprint point coordinate, ωiTable
The weighting weight of i-th of fingerprint point, the as probability of Bayesian posterior condition are levied, K indicates neighbor point number
Preferably, pose cellular network carries out posture information matching, including：
According to estimated location, at least one experience unit adjacent with estimated location is extracted from WIFI fingerprint maps；
Calculate the Euclidean distance between the WIFI signal intensity and current WIFI signal intensity of each experience unit；
The posture information pointed by the corresponding experience unit of maximum Euclidean distance is obtained, and the posture information got is true
It is set to the current posture information of robot.
It is further preferred that the calculation formula of Euclidean distance is：
Wherein, (xpc,ypc,θpc) it is the corresponding pose cell coordinate of experience unit；(xi,yi,θi) be and current location pair
The pose cell coordinate answered；raFor the zonal constant of (x, y) plane, θaFor the zonal constant in θ dimensions.
In the present embodiment, each unit tool that undergoes there are one activity level, activity level by pose perception cell and
Degree of closeness is determined between energy peak and each experience unit in WIFI fingerprints.Each experience pose perception cell and
There are one relevant active regions in WIFI fingerprints.When energy peak is in these active regions, which is activated at once, this
A little regions are continuous in pose perception cell interior, and the relevant range in WIFI fingerprints is discrete.Each warp
Go through eiBy experience activity level Ei, WIFI signal intensity RiIt is determined.Wherein, ei={ Ei,Ri,
The energy level E of one experience unitxyθWith total energy level E of i-th of experience unitiBy formula (9) and formula (10)
It can be calculated.
Wherein, xpcypcAnd θpcFor the coordinate of maximum activity posture cell；xi、yi、θiFor with the relevant pose sense of the experience
Know the coordinate of cell；raFor the zonal constant of (x, y) plane；θaFor the zonal constant in θ dimensions.RcurrIt is strong for current WIFI signal
Degree；RiFor with experience the relevant WIFI signal intensity of i.
It should be noted that in the present embodiment, as illustrated in fig. 3, WIFI fingerprint matching procedure relation schematic diagrames are utilized.It will
Wireless signal network WIFI is used as a kind of sensor in rodent models, and there are three major parts for location model
Composition, respectively WIFI fingerprints, pose cellular network and experience are schemed.WIFI fingerprints obtain the WIFI signal intensity of environment, are claimed
For WIFI signal intensity template.WIFI finger print informations are used for recognizing known environment.When the WIFI signal strength information newly inputted
When with already present WIFI signal intensity template matches, the active factors of pose cellular network are activated, and the two combines can be very
The generation that erroneous matching is prevented in big degree generates more accurate experience figure.
Although present invention has been a degree of descriptions, it will be apparent that, do not departing from the spirit and scope of the present invention
Under the conditions of, the appropriate variation of each condition can be carried out.It is appreciated that the present invention is not limited to the embodiments, and it is attributed to right
It is required that range comprising the equivalent replacement of each factor.
Claims (10)
1. a kind of synchronous superposition method based on rodent models, which is characterized in that the method includes：
Obtain the Current vision scene image information of robot；
According to the rodent models built in advance, matched from the visual information base built in advance and the Current vision
Scene image information has the posture information of maximum scene similarity；
When the scene similarity is less than given threshold, the current WIFI signal intensity set of the robot is obtained；
According to the rodent models, matched from the WIFI fingerprint maps built in advance and the current WIFI signal
Intensity set has the posture information of maximum fingerprint similarity；
According to the maximum scene similarity and the corresponding posture information of/maximum fingerprint similarity to robot progress
Synchronous superposition.
2. the synchronous superposition method according to claim 1 based on rodent models, feature exist
In the method further includes：
The visual information base is updated according to the maximum scene similarity；
The WIFI fingerprints map is updated according to the maximum fingerprint similarity.
3. the synchronous superposition method according to claim 1 based on rodent models, feature exist
In, the current WIFI signal intensity set for obtaining the robot, including：
Determine valid wireless access point；
Receive the current received signal strength mean value that each valid wireless access point is generated in current location；
The corresponding current received signal strength mean value of all valid wireless access points is determined as the current of the robot
WIFI signal intensity set.
4. the synchronous superposition method according to claim 3 based on rodent models, feature exist
In, the determining valid wireless access point, including：
Determine the quantity of valid wireless access point；
Two wireless access point are selected at random from the wireless access point set that all wireless access point form to be used as with reference to access
Point；
Calculate two first mutual informations with reference between access point；
The wireless access point of the second mutual information minimum can be made by being obtained from the wireless access point set；
The wireless access point of third mutual information minimum can be made by being obtained from the wireless access point set；
And so on, until getting sufficient amount of wireless access point.
5. the synchronous superposition method according to claim 1 based on rodent models, feature exist
In, it is described according to the rodent models, it is matched from the WIFI fingerprint maps built in advance and the current WIFI
Signal strength set has the posture information of maximum fingerprint similarity, including：
Machine is obtained according to the current WIFI signal intensity set, and according to the Bayesian posterior estimation model built in advance
The estimated location of people；
According to the estimated location, the WIFI fingerprints map, and according to the pose cell in the rodent models
Network carries out posture information matching, has maximum fingerprint similarity with the current WIFI signal intensity set to match
Posture information.
6. the synchronous superposition method according to claim 5 based on rodent models, feature exist
In, the pose cellular network carries out posture information matching, including：
According to the estimated location, at least one warp adjacent with the estimated location is extracted from the WIFI fingerprints map
Go through unit；
Calculate the Euclidean distance between the WIFI signal intensity and the current WIFI signal intensity of each experience unit；
The posture information pointed by the corresponding experience unit of maximum Euclidean distance is obtained, and the posture information got is determined as
The current posture information of robot.
7. the synchronous superposition method according to claim 6 based on rodent models, feature exist
In the calculation formula of the Euclidean distance is：
Wherein, (xpc,ypc,θpc) it is the corresponding pose cell coordinate of experience unit；(xi,yi,θi) it is corresponding with current location
Pose cell coordinate；raFor the zonal constant of (x, y) plane, θaFor the zonal constant in θ dimensions.
8. the synchronous superposition method according to claim 1 based on rodent models, feature exist
In, the structure of the WIFI fingerprints map, including：
Selected reference point；
In each reference point, it is averaging after carrying out continuous sampling to the signal strength of each preset wireless access point, to obtain
Take received signal strength mean value of each wireless access point at reference point；
According to the received signal strength mean value of the corresponding all wireless access point of each reference point, built according to preset rules
WIFI fingerprints map simultaneously stores.
9. the synchronous superposition method according to claim 1 based on rodent models, feature exist
In the data store organisation of the WIFI fingerprints map is：
IM={ φ, A, M, MACi}；Wherein,
φ={ L1,L2,…,Li,…,Lk}；A={ AP1,AP2,…,APi,…,APR}；
MACiIndicate the MAC Address value of i-th of reference point；
Wherein, IM indicates WIFI fingerprint maps；Li=(xi,yi) indicate i-th of reference point position, k be reference point quantity,
φ indicates the location sets of all reference points, indicates that the composition set of all wireless access point observed in map, R are observation
The quantity of the wireless access point arrived, M are the set for the received signal strength mean value that each reference point answers each wireless access point, whereinIt is the R wireless access point in reference point LkThe received signal strength mean value at place.
10. a kind of robot synchronous superposition device based on rodent models, which is characterized in that the dress
Set including：
Information acquisition unit, the Current vision scene image information for obtaining robot；
First matching unit, for according to the rodent models that build in advance, from the visual information base built in advance
Allot the posture information that there is maximum scene similarity with the Current vision scene image information；
Data determining unit, for when the scene similarity is less than given threshold, obtaining the current WIFI of the robot
Signal strength set；
Second matching unit, for according to the rodent models, being matched from the WIFI fingerprint maps built in advance
There is the posture information of maximum fingerprint similarity with the current WIFI signal intensity set；
Synchronous positioning unit, for according to the maximum scene similarity and the corresponding pose letter of/maximum fingerprint similarity
Breath synchronizes positioning and map structuring to the robot.
Applications Claiming Priority (4)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711320886 | 2017-12-12 | ||
| CN2017113208868 | 2017-12-12 | ||
| CN2018101033693 | 2018-02-01 | ||
| CN201810103369 | 2018-02-01 |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108680175A true CN108680175A (en) | 2018-10-19 |
Family
ID=63799613
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810265940.1A Pending CN108680175A (en) | 2017-12-12 | 2018-03-28 | Synchronous superposition method and device based on rodent models |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108680175A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112230243A (en) * | 2020-10-28 | 2021-01-15 | 西南科技大学 | Indoor map construction method for mobile robot |
| CN112325883A (en) * | 2020-10-19 | 2021-02-05 | 湖南大学 | Indoor positioning method for mobile robot with WiFi and visual multi-source integration |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104244409A (en) * | 2014-09-30 | 2014-12-24 | 智慧城市信息技术有限公司 | Server and method and system for implementing positioning by utilizing wireless signals |
| CN104507050A (en) * | 2014-11-24 | 2015-04-08 | 北京航空航天大学 | A method for probability type fingerprint matching in WiFi (Wireless Fidelity) indoor positioning |
| CN104640201A (en) * | 2014-12-31 | 2015-05-20 | 智慧城市信息技术有限公司 | Positioning method and device based on WIFI (wireless fidelity) fingerprint technology |
| CN105378809A (en) * | 2013-03-14 | 2016-03-02 | 谷歌公司 | Device, methods and related information processing for security in a smart-sensored home |
| CN106793075A (en) * | 2016-12-21 | 2017-05-31 | 武汉大学 | A kind of WiFi indoor orientation methods based on domain cluster |
| CN106814737A (en) * | 2017-01-20 | 2017-06-09 | 安徽工程大学 | A kind of SLAM methods based on rodent models and RTAB Map closed loop detection algorithms |
| CN107409051A (en) * | 2015-03-31 | 2017-11-28 | 深圳市大疆创新科技有限公司 | For generating the Verification System and method of air traffic control |
-
2018
- 2018-03-28 CN CN201810265940.1A patent/CN108680175A/en active Pending
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105378809A (en) * | 2013-03-14 | 2016-03-02 | 谷歌公司 | Device, methods and related information processing for security in a smart-sensored home |
| CN104244409A (en) * | 2014-09-30 | 2014-12-24 | 智慧城市信息技术有限公司 | Server and method and system for implementing positioning by utilizing wireless signals |
| CN104507050A (en) * | 2014-11-24 | 2015-04-08 | 北京航空航天大学 | A method for probability type fingerprint matching in WiFi (Wireless Fidelity) indoor positioning |
| CN104640201A (en) * | 2014-12-31 | 2015-05-20 | 智慧城市信息技术有限公司 | Positioning method and device based on WIFI (wireless fidelity) fingerprint technology |
| CN107409051A (en) * | 2015-03-31 | 2017-11-28 | 深圳市大疆创新科技有限公司 | For generating the Verification System and method of air traffic control |
| CN106793075A (en) * | 2016-12-21 | 2017-05-31 | 武汉大学 | A kind of WiFi indoor orientation methods based on domain cluster |
| CN106814737A (en) * | 2017-01-20 | 2017-06-09 | 安徽工程大学 | A kind of SLAM methods based on rodent models and RTAB Map closed loop detection algorithms |
Non-Patent Citations (3)
| Title |
|---|
| RAFAEL BERKVENS1, ET AL: "Biologically Inspired SLAM Using Wi-Fi", 《2014 IEEE/RSJ INTERNATIONAL CONFERENCE ON INTELLIGENT ROBOTS AND SYSTEMS (IROS 2014)》 * |
| RAFAEL BERKVENS1, ET AL: "Biologically Inspired SLAM Using Wi-Fi", 《2014 IEEE/RSJ INTERNATIONAL CONFERENCE ON INTELLIGENT ROBOTS AND SYSTEMS》 * |
| 高晓颖: "《源于自然的机器人导航 基于啮齿类动物模型的同步定位地图构建和路径规划》", 31 January 2016, 国防工业出版社 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112325883A (en) * | 2020-10-19 | 2021-02-05 | 湖南大学 | Indoor positioning method for mobile robot with WiFi and visual multi-source integration |
| CN112230243A (en) * | 2020-10-28 | 2021-01-15 | 西南科技大学 | Indoor map construction method for mobile robot |
| CN112230243B (en) * | 2020-10-28 | 2022-04-08 | 西南科技大学 | Indoor map construction method for mobile robot |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Luo et al. | Enriched indoor map construction based on multisensor fusion approach for intelligent service robot | |
| CN104869639B (en) | A kind of indoor orientation method and device | |
| CN104807460B (en) | Unmanned plane indoor orientation method and system | |
| US20140171107A1 (en) | System and method for wireless indoor localization based on inertial measurement unit and map information | |
| CN105704652B (en) | Fingerprint base acquisition and optimization method in a kind of positioning of WLAN/ bluetooth | |
| CN107635204A (en) | A kind of indoor fusion and positioning method and device of motor behavior auxiliary, storage medium | |
| CN104902562B (en) | A kind of indoor orientation method based on multilayer fingerprint matching | |
| Pathak et al. | Fast 3D mapping by matching planes extracted from range sensor point-clouds | |
| CN106199500B (en) | Fingerprint characteristic localization method and device | |
| CN108351217A (en) | It is generated the mobile device navigating, track and position with automatic map in global positioning system denied access or global positioning system inaccuracy region | |
| CN105636201B (en) | Indoor orientation method based on sparse signal fingerprint database | |
| CN103761539B (en) | Indoor locating method based on environment characteristic objects | |
| Ding et al. | Overview of received signal strength based fingerprinting localization in indoor wireless LAN environments | |
| CN109275095A (en) | A kind of indoor locating system based on bluetooth, positioning device and localization method | |
| CN108521627A (en) | The indoor locating system and method for wifi and earth magnetism fusion based on HMM | |
| CN105120479B (en) | The signal intensity difference modification method of terminal room Wi-Fi signal | |
| CN108680177B (en) | Synchronous positioning and map construction method and device based on rodent model | |
| CN110375739A (en) | A kind of mobile terminal vision fusion and positioning method, system and electronic equipment | |
| CN109212476A (en) | A kind of RFID indoor positioning algorithms based on DDPG | |
| Lin et al. | An indoor Wi-Fi localization algorithm using ranging model constructed with transformed RSSI and BP neural network | |
| CN108680175A (en) | Synchronous superposition method and device based on rodent models | |
| CN109239661A (en) | A kind of RFID indoor locating system and algorithm based on depth Q network | |
| Tian et al. | HiQuadLoc: A RSS fingerprinting based indoor localization system for quadrotors | |
| CN109839613B (en) | Radio frequency positioning method and device using path information calibration | |
| CN112729301A (en) | Indoor positioning method based on multi-source data fusion |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20181019 |
|
| WD01 | Invention patent application deemed withdrawn after publication |