CN111556432A - Crowdsourcing fingerprint database construction method based on map information screening and matching - Google Patents
Crowdsourcing fingerprint database construction method based on map information screening and matching Download PDFInfo
- Publication number
- CN111556432A CN111556432A CN202010319187.7A CN202010319187A CN111556432A CN 111556432 A CN111556432 A CN 111556432A CN 202010319187 A CN202010319187 A CN 202010319187A CN 111556432 A CN111556432 A CN 111556432A
- Authority
- CN
- China
- Prior art keywords
- road
- particle
- map
- starting point
- positions
- 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
Images
Classifications
-
- 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
- H04W4/021—Services related to particular areas, e.g. point of interest [POI] services, venue services or geofences
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/29—Geographical information databases
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04B—TRANSMISSION
- H04B17/00—Monitoring; Testing
- H04B17/30—Monitoring; Testing of propagation channels
- H04B17/309—Measuring or estimating channel quality parameters
- H04B17/318—Received signal strength
-
- 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
- H04W4/023—Services making use of location information using mutual or relative location information between multiple location based services [LBS] targets or of distance thresholds
-
- 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
- H04W4/025—Services making use of location information using location based information parameters
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W4/00—Services specially adapted for wireless communication networks; Facilities therefor
- H04W4/30—Services specially adapted for particular environments, situations or purposes
- H04W4/33—Services specially adapted for particular environments, situations or purposes for indoor environments, e.g. buildings
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W64/00—Locating users or terminals or network equipment for network management purposes, e.g. mobility management
Abstract
The invention discloses a crowd-sourced fingerprint database construction method based on map information screening and matching. And by combining the RSS data collected in the walking process, an offline fingerprint database is quickly and accurately constructed, and online positioning is assisted. The invention can construct the fingerprint database more quickly and intensively and has higher positioning precision.
Description
Technical Field
The invention belongs to the technical field of mobile communication and indoor positioning, and particularly relates to a crowdsourcing fingerprint database construction method based on map information screening and matching.
Background
With the rapid development of Wireless communication technology, due to the increasing use of mobile terminals, from the practical perspective of technology maturation and large-scale application, Wireless Local Area Network (WLAN) positioning becomes the current mainstream and is one of the most promising indoor positioning methods in the future.
The location technology based on location fingerprints does not need to know propagation parameters and location coordinates of wireless Access Points (APs) in an acquisition environment in advance, and performs location matching by establishing a feature library. Due to the complexity of the Received Signal Strength (RSS) value indoors, there is usually no case where two locations have similar RSS characteristics, so RSS can be used as a location fingerprint to achieve indoor positioning. The fingerprint positioning is generally divided into an off-line stage and an on-line stage, sampling nodes are uniformly planned in an indoor reachable area in the off-line stage, then AP signal characteristics at the nodes are collected, and a position-characteristic fingerprint database is established. And the on-line stage system compares the real-time signal characteristic scanning result of the mobile terminal with the fingerprints in the fingerprint database, and searches the node with the highest similarity as a positioning result. However, the offline fingerprint sampling requires large manpower and time overhead, so that the large-scale popularization of the positioning technology is restricted, and the application in large buildings is difficult.
In addition, due to the development of intelligent terminals, various sensors such as an acceleration sensor, a gyroscope, a geomagnetic sensor and the like are smaller and smaller, and functions are stronger and stronger. The Pedestrian trajectory can therefore be located using a Pedestrian Dead Reckoning (PDR) algorithm. The prior art considers building a radiomap using the PDR method, but the PDR technique has the disadvantages of mainly cumulative nature of errors, poor effect when used alone, and the need to know the exact initial position when positioning, which is difficult to obtain in practice. Without knowing this information, the PDR trajectory sampling locations are only relative locations and not absolute locations, which cannot be used to construct a radiomap; the PDR technology can infer the position of each step in the continuous walking process, but needs to know the starting point, has large accumulated errors of the positions along with the time, and cannot accurately construct a fingerprint database.
Disclosure of Invention
The technical problem to be solved by the invention is to provide a crowdsourcing fingerprint database construction method based on map information screening and matching aiming at the defects in the prior art, data acquisition at fixed points is not needed, the position of each step can be matched in any walking route, and starting point information is not needed to be additionally known, so that the purpose of quickly and accurately constructing an offline fingerprint database is achieved, the deployment cost is effectively reduced, and the problem that the time and labor are wasted when the traditional manual offline fixed-point data acquisition is used for constructing the fingerprint database is solved.
The invention adopts the following technical scheme:
a crowd-sourced fingerprint database construction method based on map information screening and matching comprises the following steps:
s1, keeping uniform pace speed at any position of the road, continuously collecting sensor data and corresponding received signal strength samples of each AP, repeating the steps for multiple times, and selecting one part of the path as offline training data and the other part of the path as online test data;
s2, detecting the step number of the sensor data acquired in the step S1 by utilizing a pedestrian dead reckoning algorithm, and processing gyroscope data to obtain the direction of the corresponding step;
s3, map direction matching and correction are carried out on the gyroscope direction obtained in the step S2 to be used as the advancing direction of each step;
s4, extracting the position coordinates of the walkable area and the key point area on the map, and the traveling direction of the key point;
s5, scattering points on all key point areas obtained in the step S4, and screening out all particles meeting the standard by combining the steps determined in the step S2 and the forward direction information determined in the step S3 and a route corresponding to map matching;
s6, reversely calculating the position of the starting point of the particle obtained in the step S5, and taking the position of the particle meeting the requirement as the starting point of the first step;
s7, randomly generating particles by taking the starting point position determined in the step S6 as the center, and obtaining each step position corresponding to the sensor data acquired in the step S1 by adopting a particle filter algorithm;
s8, matching the corresponding positions of all AP received signal strength samples acquired at each moment in the corresponding path by using the position obtained in each step in the step S7, taking the AP received signal strength samples of each corresponding position and corresponding moment as a record pair, and recording the record pair as a record pairThe method is constructed into an off-line database, and when the on-line positioning is carried out, the position of the on-line test data is calculated by combining the off-line database and utilizing a WKNN algorithm.
Specifically, in step S2, if n sampling points are set in the k-th step in each step detected by the pedestrian dead reckoning algorithm, the gyro integral value θ of the n sampling points is calculated1,θ2,…,θnCalculating the average valueAs the average direction of the k step, the calculation of circumference average is adoptedComprises the following steps:
wherein the content of the first and second substances,to be the average of the sine values,is the average cosine value.
Specifically, step S3 specifically includes:
s301, for each step of gyroscope direction obtained in step S2Before the direction is matched, allCorrecting the value of (A) to a corresponding positive value to ensure that the direction is positioned in an interval needing comparison;
s302, correcting the direction of each step of gyroscope in the step S301Map direction matching is carried out, a road traveling direction is randomly distributed to the ith particle to serve as an initial traveling direction, and when the updated position is calculated, the direction of each step is determined by adding the random initial traveling direction of the particle to the direction of the gyroscope to serve as the absolute direction of the particle
S303, enabling the ith particle to be in the absolute direction of the kth stepMapDeriction of all possible directions of roadjComparing j to 1, and taking the road direction with the minimum difference as the actual walking direction
S304, before the comparison in the normal direction, matching is carried out in the 360-degree direction for one time, and the obtained direction of each step is the result of the normal comparison in the step S303And the minimum value in the 360-degree direction comparison result tempmin;
s305, aiming at the actual walking direction of each step obtained in the step S304And (5) performing turning correction.
Further, step S305 specifically includes:
when the matched direction of the k stepDirection of step k-1At different times, the following two conditions were calculated:
min(|k-TurnStep|)>Thrturn
wherein, ThrstraightIs the decision threshold for straight-going, Turnstep is the sequence number of all turn steps, ThrturnIs the tolerance threshold for the turn step; when the above two conditions are satisfied simultaneously, the direction of the k step is adjustedCorrection to the direction of step k-1Otherwise, the direction of the matched k step is not changed
Specifically, step S4 specifically includes:
s401, respectively extracting position coordinates Cango (cangox, cangoy) and key point region position coordinates KeyPoint (Keypointx, Keypointy) of a walkable region on the map according to the RGB colors of the map;
s402, setting a road width threshold RoadWidth according to the map road width information, respectively advancing RoadWidth steps along all road directions MapDiection for each key point, and recording all the coincident road directions as passable directions KeyDir of the key point if each step is within the range of a walkable region Cango.
Specifically, step S5 specifically includes:
s501, scattering points on all key point areas obtained in the step S4 to serve as initial point positions, and sequentially updating all particle position information from the first turning step to the last turning step by using a PDR position updating formula according to the step number obtained in the step S2 and the corresponding step direction information obtained in the step S3;
s502, in the process of updating the position in each step of the step S501, for the position of the ith particle in the k step, if the position is updated, the position is updated Add 1 to Score ifRecording the step number k of the first wall collision, not adding Score, and not adding Score for each subsequent step in the route segment with the consistent direction, if the position is still in the Cango area;
s503, sequentially updating the positions according to the step S501, stopping continuously updating the positions when the direction is changed and the vehicle turns to enter the next section of path, and performing whole-section translation correction on the road section with the wall collision condition again;
s504, when CandidateEnd exists, for a line segment which starts from a starting point in CandidateBegin and is connected with an end point, selecting a candidate starting point CandidateBegin corresponding to the minimum value in DisBegin as a finally selected winning key point WinBegin, wherein the CandidateEnd is a line segment which is in line connection with the end point and accords with the traveling direction of a map road line;
s505, calculating a straight line passing through a key point WinBegin in the passing direction of the current road section and a straight line passing through a starting point, namely the t step, in the other road direction perpendicular to the current road section, wherein the Intersection interaction of the two straight lines is the required first-step correction position;
s506, calculating the integral Shift distance needed to translate, namely the t-th step of the Intersection interaction and the starting pointThe difference of each coordinate in (1);
s507, adding a translation distance Shift to all positions from the starting point to the end point, namely from the t step to the q step, to obtain an updated position; sequentially calculating whether the updated position set belongs to the Cango range, if so, adding 1 to Score, and otherwise, recording as wall collision;
when the wall collision exceeds 5 times, directly ending the position updating process of all the subsequent steps of the segment of particles, and jumping to the step S501 to perform the position updating process of the next particle; when the collision with the wall does not exceed 5 times, continuing to start the position updating process of all the subsequent steps of the particle by the corrected road section terminal point;
and S508, after the positions of all the particles are updated in sequence, sequencing the Score of all the particles, and selecting the corresponding particles of which the Score meets the requirements as a possible starting point set MayStartIndex.
Further, in step S503, the wall collision step k is compared in sequenceAndif the two steps are consistent, finding the number of steps with changed directions as the first step of the section, assuming the t step, and updating the Score, wherein the Score is Score- (k-t); for the position of the t stepCalculating the distances to all key points as follows:
setting a maximum distance threshold KeyPointradius and a passable direction KeyDir of the key point; when the following two conditions are satisfied:
DisBegin≤KeyPointRadius
then all corresponding key points KeyPoint are used as candidate starting points candidat;
in the presence of CandidateBegin, assume that the last step is the q-th step, and the position of the q-th stepDistances to all keypoints are also calculated:
when the following two conditions are satisfied:
DisEnd≤KeyPointRadius
all corresponding keypoints KeyPoint are taken as candidate end points candidated.
Further, in step S504, when CandidateEnd does not exist, the candidate starting point CandidateBegin corresponding to the minimum value in DisBegin is directly selected as the finally selected winning key point WinBegin.
Specifically, in step S6, the possible starting point set MayStartIndex obtained in step S508 is expressed by the following formula:
and sequentially reversely calculating the positions of the first step, and simultaneously calculating whether the position of each step belongs to the Cango range, if so, recording the position of the first step reversely pushed by the particles as the matched road section starting point position MayStartPosition.
Specifically, in step S8, according to the position of each step obtained in step S7, the time corresponding to the position of each step, and the pair of the collected received signal strength samples of all APsCalculating the corresponding position (x, y) of the received signal strength samples of all APs by linear interpolation according to time, and taking the position (x, y) and the received signal strength samples of the corresponding APs Rssi1,Rssi2,…,RssinFor a pair, an offline database is constructed.
Compared with the prior art, the invention has at least the following beneficial effects:
the invention relates to a crowdsourcing fingerprint database construction method based on map information screening and matching, which is characterized in that gyroscope information and map related information are processed and corrected step by step, an optimal position is obtained by utilizing particle scoring and two-layer screening, and an offline fingerprint database is constructed by linear interpolation of positions obtained by particle filtering, wherein the correctness of each step of link ensures the correctness of the finally obtained offline fingerprint database.
Further, step S2 calculates the average value of the angle information obtained from the gyroscope data by means of circumferential averaging instead of arithmetic averaging, so that the correct average value can be calculated more accurately, and errors caused by data of 0 ° and 360 ° can be avoided.
Further, step S3 performs hard matching of the gyro data with the map-travelable direction, and can reduce the direction angle shift phenomenon due to gyro accumulated errors.
Further, in step S305, the matched direction is corrected using the turning and straight information, and thus hard matching errors due to deviation of the direction angle can be reduced.
Further, step S4 extracts the positions of the key points and the passable directions thereof, so that the area range of the initial starting point scattering points can be narrowed, the corresponding positions of the route can be matched quickly, and the direction information of the key points can assist in position correction in the matching process.
Further, in step S5, the optimal particle is screened out by updating and correcting the positions of all the particles and then scoring, so that the correct position can be obtained quickly and accurately.
Furthermore, the step S503 performs the whole translation correction on the road section colliding with the wall instead of performing the position correction individually at each step, which considers the integrity of the road section, is more reasonable compared with the single step correction, and can more accurately match the correct path.
Further, in step S504, the key points used in the road segment correction process are subjected to head-to-tail double screening, so that it can be relatively correctly ensured that the selected key points are right.
Further, by reversely extrapolating the position of the start point of the link in step S6, the accuracy of the position of the particle screened in step S5 can be further verified, and the correct start point of the link can be obtained.
Further, step S8 performs linear interpolation using the time relationship, so that the corresponding position of the sampling time can be matched relatively correctly, and a relatively accurate offline fingerprint database can be constructed.
In summary, the invention matches the position corresponding to the path quickly and accurately by extracting and utilizing the map information, thereby constructing the corresponding off-line fingerprint database.
The technical solution of the present invention is further described in detail by the accompanying drawings and embodiments.
Drawings
FIG. 1 is a block diagram of a positioning implementation of the method of the present invention;
FIG. 2 is a flowchart of a gyroscope orientation matching and correction algorithm;
FIG. 3 is a flow chart of matching trajectory starting points;
fig. 4 is a plan view of a test environment.
Detailed Description
The invention provides a crowd-sourced fingerprint database construction method based on map information screening and matching. And by combining the RSS data collected in the walking process, an offline fingerprint database is quickly and accurately constructed, and online positioning is assisted. Compared with the traditional fixed-point acquisition and fingerprint database construction mode, the method can construct the fingerprint database more quickly and intensively, and has higher positioning precision.
Referring to fig. 1, the method for constructing a crowdsourcing fingerprint database based on map information screening matching of the present invention includes the following steps:
s1, collecting sensor data;
walking at a uniform pace for a distance at any position of a road, and continuously acquiring data of each sensor and corresponding received signal strength samples Rssi of each AP in the process by the mobile phone1,Rssi2,…,RssinWherein n is the number of samples. And repeatedly walking for dozens of times, wherein all passable roads are required to be covered, and samples are respectively selected as offline training data and online testing data.
S2, detecting the step number of the sensor data acquired in the step S1 by using a Pedestrian Dead Reckoning (PDR) algorithm, and processing gyroscope data to obtain the direction of the corresponding step;
setting n sampling points in the kth step for each step detected by the PDR algorithm, and then integrating values theta of gyroscopes of the n sampling points1,θ2,…,θnCalculating the average valueAs the average direction of the k-th step, wherein,the calculation is carried out in a circumferential average mode, and the calculation formula is as follows:
wherein the content of the first and second substances,is flatThe value of the mean sine is calculated,is the average cosine value.
S3, map direction matching and correction are carried out on the gyroscope direction obtained in the step S2 to be used as the advancing direction of each step;
s301, for each step of gyroscope direction obtained in step S2Before the direction is matched, all the components need to be matchedCorrecting the value of (A) to a corresponding positive value to ensure that the direction is positioned in a range needing to be compared; the correction formula is as follows:
s302, correcting the direction of each step of the gyroscope obtained in the step S301And carrying out map direction matching. Since the direction of the gyroscope is a relative direction, a road traveling direction is randomly assigned to the ith particle as an initial traveling direction, and when calculating the updated position, the direction of each step should be determined by adding the random initial traveling direction of the particle to the direction of the gyroscope as the "absolute direction" of the particle "
S303, making the ith particle in the 'absolute direction' of the kth step "MapDeriction of all possible directions of roadj(j ═ 1.. m) is compared, and the road direction with the smallest difference is taken as the actual walking direction of the step
S304, in addition, since 0 ° and 360 ° are actually the same direction, and the stored road direction MapDirection does not include 360 °, it is necessary to perform matching for 360 ° direction separately before normal direction comparison, and finally the obtained direction of each step is the result of normal comparison in step S303And the minimum value in the 360-degree direction comparison result tempmin;
s305, aiming at the actual walking direction of each step obtained in the step S304Performing turning correction;
the specific operation is as follows: when the matched direction of the k stepDirection of step k-1At different times, the following two conditions were calculated:
min(|k-TurnStep|)>Thrturn
wherein, ThrstraightIs the decision threshold for straight-going, Turnstep is the sequence number of all turn steps, ThrturnIs the tolerance threshold for the turn step.
When the above two conditions are satisfied simultaneously, the direction of the k step is adjustedCorrection to the direction of step k-1Otherwise, it is not changedChanging the direction of the matched k-th step
S4, extracting the position coordinates of the walkable area and the key point area on the map and the key point travelable direction;
s401, respectively extracting position coordinates Cango (cangox, cangoy) and key point region position coordinates KeyPoint (Keypointx, Keypointy) of a walkable region on the map according to the RGB colors of the map;
s402, extracting the passable direction of each key point. The method specifically comprises the steps of setting a road width threshold RoadWidth according to the map road width information, respectively advancing RoadWidth steps along all road directions MapDiection for each key point, and recording all the coincident road directions as passable directions KeyDir of the key point if each step is within the range of a walkable area Cango.
S5, scattering points on all key point areas obtained in the step S4, and screening out all particles meeting the standard by utilizing the step number obtained in the step S2 and the direction information obtained in the step S3 in combination with a route corresponding to map matching;
s501, scattering points on all the key point regions obtained in step S4 as initial point positions, and sequentially updating all the particle position information from the first turning step to the last turning step by using the step number obtained in step S2 and the corresponding step direction information obtained in step S3 using a PDR position update formula, which is calculated as follows:
wherein the content of the first and second substances,is the position coordinate of the ith particle at step k-1,represents the step size of the ith particle in the kth step;
S502in the process of updating the position in each step of step S501, for the position of the ith particle in the k step, if so, the position is updated Add 1 to Score ifRecording the step number k of the first wall collision, not scoring Score, and not scoring Score any more if the position of each subsequent step in the route segment with the consistent direction is still in the Cango area;
s503, sequentially updating the positions according to the step S501, stopping continuously updating the positions when the direction is changed and the vehicle turns to enter the next section of path, and performing whole-section translation correction on the road section with the wall collision condition again;
the specific method comprises the following steps: step k of wall collision, comparing in sequenceAndand if the two steps are consistent, the step number of the found direction change is the first step of the section, and the Score is updated under the assumption that the step number is the t-th step, and the Score is equal to Score- (k-t). For the position of the t stepCalculating the distance to all key points, and the formula is as follows:
and setting a maximum distance threshold KeyPointradius and a passable direction KeyDir of the key point.
When the following two conditions are satisfied:
DisBegin≤KeyPointRadius
all corresponding keypoints KeyPoint are taken as candidate starting points candidin. In the presence of CandidateBegin, assume that the last step of the segment is the q-th step, and the position of the q-th stepDistances to all keypoints are also calculated:
when the following two conditions are satisfied:
DisEnd≤KeyPointRadius
taking all corresponding key points KeyPoint as candidate terminal CandidateEnd;
s504, when CandidateEnd exists, for a line segment which starts from a starting point in CandidateBegin and is connected with an end point, selecting a candidate starting point CandidateBegin corresponding to the minimum value in DisBegin as a finally selected winning key point WinBegin, wherein the CandidateEnd is a line segment which is in line connection with the end point and accords with the traveling direction of a map road line;
under the condition that CandidateEnd does not exist, directly selecting a candidate starting point CandidateBegin corresponding to the minimum value in the DisBegin as a finally selected winning key point WinBegin;
s505, calculating a first correction position of the section;
calculating a straight line passing through WinBegin in the passing direction of the current road section and a straight line passing through the starting point of the section, namely the t step, in the direction of the other road perpendicular to the current road section, wherein the Intersection interaction of the two straight lines is the required first-step correction position;
s506, calculating the distance Shift of the whole section needing translation, namely the Intersection interaction and the t-th step of the starting point of the sectionThe difference of each coordinate in (1);
s507, adding a translation distance Shift to all positions from the starting point to the end point of the section, namely from the t step to the q step, integrally, namely to obtain an updated position; sequentially calculating whether the updated position set belongs to the Cango range, if so, adding 1 to Score, otherwise, marking as wall collision, and when the wall collision exceeds 5 times, directly ending the position updating process of all subsequent steps of the segment of particles, and jumping to the step S501 to carry out the position updating process of the next particle; if the collision with the wall does not exceed 5 times, continuing to start the position updating process of all the subsequent steps of the particle by the corrected road section terminal;
and S508, after the positions of all the particles are updated in sequence, sequencing the Score of all the particles, and selecting the corresponding particles of which the Score meets the requirements as a possible starting point set MayStartIndex.
S6, reversely calculating the position of the starting point of the particle obtained in the step S5, and if the position of the particle meets the requirement, keeping the position of the particle as the starting point of the first step;
for the possible starting point set MayStartIndex obtained in step S508, the following formula is followed:
and sequentially reversely calculating the positions of the first step, simultaneously calculating whether the position of each step belongs to the Cango range, and if so, recording the position of the first step of the particle reverse extrapolation as the matched starting point position MayStartPosition of the road section.
S7, randomly generating particles within a radius range consistent with the road width by taking the starting point position obtained in the step S6 as the center, and obtaining each step position corresponding to the path sensor data collected in the step S1 by adopting a particle filter algorithm;
and S8, matching the corresponding positions of the received signal strength samples of all APs acquired in the path by using the position obtained in each step in the step S7, constructing an off-line database, and calculating the position of the on-line test data by using a WKNN algorithm.
S801, calculating the corresponding positions (x, y) of the received signal strength samples of all APs by linear interpolation according to the time corresponding to each step position and the corresponding time of the received signal strength samples of all APs obtained from step S7, and using the positions (x, y) and the received signal strength samples Rssi of the corresponding APs1,Rssi2,…,RssinFor a pair, an offline database is constructed. The position is calculated as follows:
where t is the time taken to collect a sample of the received signal strength, t1The time of the corresponding step with the smallest difference from t, (x)1,y1) Is the position of the step, t2Time of the corresponding step which is the second smallest difference from t, (x)2,y2) Is the location of the step;
and S802, calculating the position of the online test data by using the offline fingerprint library constructed in the step S801 and adopting a traditional WKNN algorithm.
For an AP received signal strength vector set P ═ (P) collected in the online test stage1,p2,…,pW),pjRepresenting the signal strength value from the jth AP, calculating P and M fingerprints RSS in the offline fingerprint libraryi(i is 1,2, …, M), whereinRepresenting the signal strength value, N, from the jth AP acquired at the ith anchor pointiIndicating the total number of APs that can receive the signal at the ith anchor point. During calculation, an intersection is taken from the AP sets corresponding to the two vectors, and the Euclidean distance is calculated by using the received signal strength corresponding to the AP in the intersection:
wherein W represents the intersection number of APs between the offline fingerprint and the signal strength vector received in the online test stage
Sorting all fingerprints according to Euclidean distance, dis (P, RSS)i1)≤dis(P,RSSi2)≤…≤dis(P,RSSiM) Wherein i isj∈{1,2,…,M},ijAnd a corresponding serial number representing the fingerprint with the Euclidean distance arranged at the j-th bit.
Selecting the first K off-line fingerprints { i) with the minimum Euclidean distancejJ is 1,2,.. K }, and the set of position coordinates corresponding to the K offline fingerprints isAnd taking the reciprocal of the Euclidean distance corresponding to the K selected positions as a weight, and then normalizing:
the estimated location of the end user is then expressed as:
in order to make the objects, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are some, but not all, embodiments of the present invention. The components of the embodiments of the present invention generally described and illustrated in the figures herein may be arranged and designed in a wide variety of different configurations. Thus, the following detailed description of the embodiments of the present invention, presented in the figures, is not intended to limit the scope of the invention, as claimed, but is merely representative of selected embodiments of the invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
Examples
Referring to fig. 4, a typical shopping mall is shown, an environmental area is 35.2 mx 106.2m, a signal acquisition terminal is an Android smartphone, and a specific operation process is as follows:
1) collecting sensor data
A plurality of paths are planned in advance on any position of a road, all passable road sections of a shopping mall are covered, and the path length is different from 118m to 236 m. On the paths, the mobile phone walks at a uniform pace, and continuously acquires the data of each sensor and the corresponding received signal strength sample Rssi of each AP in the process1,Rssi2,…,RssinWherein n is the number of samples. Repeatedly walking for dozens of times, automatically selecting a part of paths as offline training data, and taking the other part of paths as online testing data;
2) step number detection is carried out on the sensor data acquired in the step 1) by utilizing a pedestrian dead reckoning algorithm (PDR algorithm), and gyroscope data are processed in a circumferential average mode to obtain the direction of the corresponding step. Setting n sampling points for the kth step detected by the PDR algorithm, and then using the gyroscope integral value theta of the n sampling points1,θ2,…,θnAverage value of (2)As the average direction of the k step;
3) the direction of each step of gyroscope obtained in the step 2)InThe value of (A) is corrected to a corresponding positive value, so that the direction is positioned in the interval [0 degrees and 360 degrees ] required to be compared;
4) correcting the direction of each step of gyroscope obtained in the step 3)And carrying out map direction matching. Since the direction of the gyroscope is the relative direction, a road row is randomly allocated to the ith particleThe advancing direction MapDirection is used as an initial traveling direction, and in this example, the road traveling direction MapDirection of the map has four values of 0 °,90 °,180 °,270 °, which represent traveling in the right, up, left, and down directions, respectively. In calculating the updated position, the direction of each step should be the "absolute direction" of the particle, which is the gyroscope direction plus the random initial travel direction assigned by the particle "
5) Since 0 ° and 360 ° are actually the same direction, and the stored road direction MapDirection does not contain 360 °, it is necessary to match the 360 ° direction once before the comparison of the normal directions, and calculate the "absolute direction" of the ith particle at the k step "Tempmin from 360 °. Then map direction with all possible directions of roadjComparing (j ═ 1.. m) to obtain the smallest difference value, and taking the road direction corresponding to the smallest value in said value and tempmin as the actual walking direction of said step
6) For the actual walking direction of each step obtained by 5)And (5) performing turning correction. When the matched direction of the k stepDirection of step k-1At different times, the following two conditions were calculated:
min(|k-TurnStep|)>Thrturn
wherein, ThrstraightIs a straight-going judgment threshold value, and Thr is taken in the experimentstraightTurn step is all turn step sequence number, Thr 10turnIs the tolerance threshold of the turning step, and Thr is taken in the experimentturn2. When the above two conditions are satisfied simultaneously, the direction of the k step is adjustedCorrection to the direction of step k-1Otherwise, the direction of the matched k step is not changed
7) And extracting the position coordinates of the walkable area and the key point area on the map and the travelable direction of the key point. Respectively extracting walking area position coordinates Cango (cangox, cangoy) and key point area position coordinates KeyPoint (Keypointx, Keypointy) on the map according to RGB color values of different areas in the map, wherein the number of key point positions in the experimental environment is 24;
8) and extracting the passable direction of each key point. Specifically, a road width threshold RoadWidth is set according to the map road width information, and the RoadWidth is taken as 5 in the experiment. For each key point, advancing a RoadWidth step along all road directions MapDiection respectively, and if each step is within the range of a walkable area Cango, recording all the coincident road directions as passable directions KeyDir of the key point;
9) scattering dots on all key point areas obtained in the step 8), and obtaining 24 × 4-96 particles in total. Calculating the Score of each particle by using the number of steps obtained in the step 2) and the direction information obtained in the step 6) in combination with the route corresponding to the map matching.
Taking the 96 particle positions as initial point positions, and adopting a PDR position updating formula to sequentially update all particle position information from the first turning step to the last step; updating the position at each stepIf the position of the ith particle in the k stepAdd 1 to Score ifRecording the step number k of the first wall collision, not scoring Score, and not scoring Score any more if the position of each subsequent step in the route segment with the consistent direction is still in the Cango area;
and sequentially updating the positions until the direction is changed and the turning is carried out to enter the next section of path, stopping continuously updating the positions, and carrying out whole-section translation correction on the road section with the wall collision condition again. The method comprises sequentially comparing wall collision steps kAndand if the two steps are consistent, the step number of the found direction change is the first step of the section, and the Score is updated under the assumption that the step number is the t-th step, and the Score is equal to Score- (k-t). For the position of the t stepCalculating the distances to all key points; setting a maximum distance threshold KeyPointradius to the key point as 3m and a passable direction KeyDir of the key point;
when the following two conditions are satisfied:
DisBegin≤KeyPointRadius
all corresponding keypoints KeyPoint are taken as candidate starting points candidin. In the presence of CandidateBegin, assume that the last step of the segment is the q-th step, and the position of the q-th stepDistances to all key points are also calculated when the following two conditions are met:
DisEnd≤KeyPointRadius
taking all corresponding key points KeyPoint as candidate terminal CandidateEnd;
when CandidateEnd exists, for a line segment which starts from a starting point in CandidateBegin and is connected by an end point, selecting a candidate starting point CandidateBegin corresponding to the minimum value in DisBegin as a finally selected winning key point WinBegin, wherein the line segment conforms to the traveling direction of a map road line; under the condition that CandidateEnd does not exist, directly selecting a candidate starting point CandidateBegin corresponding to the minimum value in the DisBegin as a finally selected winning key point WinBegin;
and calculating a straight line passing through WinBegin in the passing direction of the current road section, a straight line passing through the starting point of the section, namely the t-th step, in the direction of the other road perpendicular to the current road section, and an Intersection interaction of the two straight lines, namely the calculated first-step corrected position. Calculating the distance Shift of the whole section needing translation, namely the Intersection point interaction and the t-th step of the starting point of the sectionThe difference of each coordinate in (1);
adding translation distance Shift to all positions from the starting point to the end point, namely from the t step to the q step, namely the updated position, sequentially calculating whether the updated position set belongs to the Cango range, if so, adding 1 to Score, otherwise, marking as wall collision, and when the wall collision exceeds 5 times, directly ending the position updating process of all subsequent steps of the segment of particles and carrying out the position updating process of the next particle; if the collision does not exceed 5 times, the position updating process of all the subsequent steps of the particle is continuously started by the corrected road section end point.
10) After the positions of all the particles are updated in sequence, the scores of all the particles are sorted, and the selected Score meets the following requirements: the difference of the total steps corresponding to the route section is not more than 5. The corresponding particle serves as the set of possible starting points MayStartIndex.
And reversely calculating the starting point position of the possible starting point set MayStartIndex, if the position of the particle is in accordance with the requirement, reserving the position of the particle as the starting point of the first step, sequentially reversely calculating the position of the first step, simultaneously calculating whether the position of each step belongs to the Cango range, and if the position of each step is met, recording the position of the first step reversely deduced by the particle as the matched starting point position MayStartPosition of the road section.
11) Randomly generating 100 particles within a radius of 5m by taking the starting point position obtained in the step 10) as a center, and obtaining each step position corresponding to the off-line sensor data acquired in the step 1) by adopting a particle filtering algorithm;
12) and matching the corresponding positions of the received signal strength samples of all APs acquired in the path by using the position obtained in each step in the step 11) to construct an offline database. Calculating the corresponding positions (x, y) of the received signal strength samples of all the APs by utilizing linear interpolation according to the time corresponding to each step position and the corresponding time of the received signal strength samples of all the APs, and calculating the corresponding positions (x, y) of the received signal strength samples of all the APs according to the positions (x, y) and the received signal strength samples Rssi of the corresponding APs1,Rssi2,…,RssinConstructing an offline database for each pair;
13) and (3) calculating the position of the online test data by using the offline fingerprint library constructed by 12) and adopting a traditional WKNN algorithm.
TABLE 1
The positioning result of this embodiment is shown in table 1, and the comparison method is to position the fingerprint database constructed by the conventional fixed-point acquisition method; as can be seen from Table 1, the average positioning error and the 95% positioning error of the invention are both obviously reduced, which proves that the invention can effectively improve the average precision of the positioning system.
In summary, according to the crowdsourcing fingerprint database construction method based on map information screening and matching, by utilizing map information including key points, the direction of a road capable of traveling and the like, a two-layer screening and scoring mechanism is adopted, the real corresponding position of a path is quickly matched, and an offline fingerprint database can be quickly and accurately constructed without single-step acquisition and knowing the starting point position information of the path.
The above-mentioned contents are only for illustrating the technical idea of the present invention, and the protection scope of the present invention is not limited thereby, and any modification made on the basis of the technical idea of the present invention falls within the protection scope of the claims of the present invention.
Claims (10)
1. A crowdsourcing fingerprint database construction method based on map information screening and matching is characterized by comprising the following steps:
s1, keeping uniform pace speed at any position of the road, continuously collecting sensor data and corresponding received signal strength samples of each AP, repeating the steps for multiple times, and selecting one part of the path as offline training data and the other part of the path as online test data;
s2, detecting the step number of the sensor data acquired in the step S1 by utilizing a pedestrian dead reckoning algorithm, and processing gyroscope data to obtain the direction of the corresponding step;
s3, map direction matching and correction are carried out on the gyroscope direction obtained in the step S2 to be used as the advancing direction of each step;
s4, extracting the position coordinates of the walkable area and the key point area on the map, and the traveling direction of the key point;
s5, scattering points on all key point areas obtained in the step S4, and screening out all particles meeting the standard by combining the steps determined in the step S2 and the forward direction information determined in the step S3 and a route corresponding to map matching;
s6, reversely calculating the position of the starting point of the particle obtained in the step S5, and taking the position of the particle meeting the requirement as the starting point of the first step;
s7, randomly generating particles by taking the starting point position determined in the step S6 as the center, and obtaining each step position corresponding to the sensor data acquired in the step S1 by adopting a particle filter algorithm;
s8, matching the corresponding positions of all AP received signal strength samples acquired at each moment in the corresponding path by using the position obtained in each step in the step S7, taking the AP received signal strength samples of each corresponding position and corresponding moment as a record pair, and recording the record pair as a record pairThe method is constructed into an off-line database, and when the on-line positioning is carried out, the position of the on-line test data is calculated by combining the off-line database and utilizing a WKNN algorithm.
2. The method for constructing a crowdsourced fingerprint database based on map information screening matching according to claim 1, wherein in step S2, if n sampling points are set in the kth step in each step detected by the pedestrian dead reckoning algorithm, a gyroscope integral value θ of the n sampling points is determined1,θ2,…,θnCalculating the average valueAs the average direction of the k step, the calculation of circumference average is adoptedComprises the following steps:
3. The method for constructing the crowdsourced fingerprint database based on map information screening matching according to claim 1, wherein the step S3 specifically comprises:
s301, for each step of gyroscope direction obtained in step S2Before the direction is matched, allCorrecting the value of (A) to a corresponding positive value to ensure that the direction is positioned in an interval needing comparison;
s302, correcting the direction of each step of gyroscope in the step S301Map direction matching is carried out, a road traveling direction is randomly distributed to the ith particle to serve as an initial traveling direction, and when the updated position is calculated, the direction of each step is determined by adding the random initial traveling direction of the particle to the direction of the gyroscope to serve as the absolute direction of the particle
S303, enabling the ith particle to be in the absolute direction of the kth stepMapDeriction of all possible directions of roadjComparing j to 1, and taking the road direction with the minimum difference as the actual walking direction
S304, before the comparison in the normal direction, matching is carried out in the 360-degree direction for one time, and the obtained direction of each step is the result of the normal comparison in the step S303And the minimum value in the 360-degree direction comparison result tempmin;
4. The method for constructing the crowdsourced fingerprint database based on map information screening matching according to claim 3, wherein the step S305 specifically comprises:
when the matched direction of the k stepDirection of step k-1At different times, the following two conditions were calculated:
min(|k-TurnStep|)>Thrturn
wherein, ThrstraightIs the decision threshold for straight-going, Turnstep is the sequence number of all turn steps, ThrturnIs the tolerance threshold for the turn step; when the above two conditions are satisfied simultaneously, the direction of the k step is adjustedCorrection to the direction of step k-1Otherwise, the direction of the matched k step is not changed
5. The method for constructing the crowdsourced fingerprint database based on map information screening matching according to claim 1, wherein the step S4 specifically comprises:
s401, respectively extracting position coordinates Cango (cangox, cangoy) and key point region position coordinates KeyPoint (Keypointx, Keypointy) of a walkable region on the map according to the RGB colors of the map;
s402, setting a road width threshold RoadWidth according to the map road width information, respectively advancing RoadWidth steps along all road directions MapDiection for each key point, and recording all the coincident road directions as passable directions KeyDir of the key point if each step is within the range of a walkable region Cango.
6. The method for constructing the crowdsourced fingerprint database based on map information screening matching according to claim 1, wherein the step S5 specifically comprises:
s501, scattering points on all key point areas obtained in the step S4 to serve as initial point positions, and sequentially updating all particle position information from the first turning step to the last turning step by using a PDR position updating formula according to the step number obtained in the step S2 and the corresponding step direction information obtained in the step S3;
s502, in the process of updating the position in each step of the step S501, for the position of the ith particle in the k step, if the position is updated, the position is updatedAdd 1 to Score ifThe step number k of the first wall collision is recorded, Score is not added, and the first direction is oneEach subsequent step in the route segment, if the position is still in the Cango area, not adding Score;
s503, sequentially updating the positions according to the step S501, stopping continuously updating the positions when the direction is changed and the vehicle turns to enter the next section of path, and performing whole-section translation correction on the road section with the wall collision condition again;
s504, when CandidateEnd exists, for a line segment which starts from a starting point in CandidateBegin and is connected with an end point, selecting a candidate starting point CandidateBegin corresponding to the minimum value in DisBegin as a finally selected winning key point WinBegin, wherein the CandidateEnd is a line segment which is in line connection with the end point and accords with the traveling direction of a map road line;
s505, calculating a straight line passing through a key point WinBegin in the passing direction of the current road section and a straight line passing through a starting point, namely the t step, in the other road direction perpendicular to the current road section, wherein the Intersection interaction of the two straight lines is the required first-step correction position;
s506, calculating the integral Shift distance needed to translate, namely the t-th step of the Intersection interaction and the starting pointThe difference of each coordinate in (1);
s507, adding a translation distance Shift to all positions from the starting point to the end point, namely from the t step to the q step, to obtain an updated position; sequentially calculating whether the updated position set belongs to the Cango range, if so, adding 1 to Score, and otherwise, recording as wall collision;
when the wall collision exceeds 5 times, directly ending the position updating process of all the subsequent steps of the segment of particles, and jumping to the step S501 to perform the position updating process of the next particle; when the collision with the wall does not exceed 5 times, continuing to start the position updating process of all the subsequent steps of the particle by the corrected road section terminal point;
and S508, after the positions of all the particles are updated in sequence, sequencing the Score of all the particles, and selecting the corresponding particles of which the Score meets the requirements as a possible starting point set MayStartIndex.
7. The method as claimed in claim 6, wherein the step S503 is performed by sequentially comparing the step K of wall collisionAndif the two steps are consistent, finding the number of steps with changed directions as the first step of the section, assuming the t step, and updating the Score, wherein the Score is Score- (k-t); for the position of the t stepCalculating the distances to all key points as follows:
setting a maximum distance threshold KeyPoint Radius and a passable direction KeyDir of the key point; when the following two conditions are satisfied:
DisBegin≤KeyPoint Radius
then all corresponding key points KeyPoint are used as candidate starting points candidat;
in the presence of CandidateBegin, assume that the last step is the q-th step, and the position of the q-th stepDistances to all keypoints are also calculated:
when the following two conditions are satisfied:
DisEnd≤KeyPointRadius
all corresponding keypoints KeyPoint are taken as candidate end points candidated.
8. The method for constructing the crowdsourced fingerprint library based on map information screening matching according to claim 6, wherein in step S504, when CandidateEnd does not exist, a candidate starting point CandidateBegin corresponding to a minimum value in DisBegin is directly selected as a finally selected winning key point WinBegin.
9. The method for constructing the crowd-sourced fingerprint library based on map information screening matching according to claim 1, wherein in step S6, the possible starting point set MayStartIndex obtained in step S508 is obtained according to the following formula:
and sequentially reversely calculating the positions of the first step, and simultaneously calculating whether the position of each step belongs to the Cango range, if so, recording the position of the first step reversely pushed by the particles as the matched road section starting point position MayStartPosition.
10. The method as claimed in claim 1, wherein in step S8, the positions (x, y) corresponding to the received signal strength samples of all APs are calculated by linear interpolation according to the positions of each step obtained in step S7, the time corresponding to each position of each step, and the corresponding time of the received signal strength samples of all APs, and the positions (x, y) and the received signal strength samples Rssi of the corresponding APs are used to calculate the positions (x, y)1,Rssi2,…,RssinConstruct an offline number for a pairA database.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010319187.7A CN111556432B (en) | 2020-04-21 | 2020-04-21 | Crowdsourcing fingerprint database construction method based on map information screening and matching |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010319187.7A CN111556432B (en) | 2020-04-21 | 2020-04-21 | Crowdsourcing fingerprint database construction method based on map information screening and matching |
Publications (2)
Publication Number | Publication Date |
---|---|
CN111556432A true CN111556432A (en) | 2020-08-18 |
CN111556432B CN111556432B (en) | 2022-02-01 |
Family
ID=72005824
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202010319187.7A Active CN111556432B (en) | 2020-04-21 | 2020-04-21 | Crowdsourcing fingerprint database construction method based on map information screening and matching |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN111556432B (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113704377A (en) * | 2021-08-30 | 2021-11-26 | 中国地质大学(武汉) | Road network matching method combining artificial interaction and active learning |
CN113810850A (en) * | 2021-08-11 | 2021-12-17 | 荣耀终端有限公司 | Method for creating positioning database and electronic equipment |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20160025498A1 (en) * | 2014-07-28 | 2016-01-28 | Google Inc. | Systems and Methods for Performing a Multi-Step Process for Map Generation or Device Localizing |
CN105704652A (en) * | 2016-02-16 | 2016-06-22 | 中国科学院光电研究院 | Method for building and optimizing fingerprint database in WLAN/Bluetooth positioning processes |
CN107396321A (en) * | 2017-08-02 | 2017-11-24 | 江南大学 | Unsupervised formula indoor orientation method based on mobile phone sensor and iBeacon |
CN108496057A (en) * | 2016-01-19 | 2018-09-04 | 飞利浦照明控股有限公司 | It is positioned based on light source |
CN108632761A (en) * | 2018-04-20 | 2018-10-09 | 西安交通大学 | A kind of indoor orientation method based on particle filter algorithm |
CN110260855A (en) * | 2019-06-24 | 2019-09-20 | 北京壹氢科技有限公司 | A kind of indoor pedestrian navigation localization method merging pedestrian's dead reckoning, Geomagnetism Information and indoor map information |
-
2020
- 2020-04-21 CN CN202010319187.7A patent/CN111556432B/en active Active
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20160025498A1 (en) * | 2014-07-28 | 2016-01-28 | Google Inc. | Systems and Methods for Performing a Multi-Step Process for Map Generation or Device Localizing |
CN108496057A (en) * | 2016-01-19 | 2018-09-04 | 飞利浦照明控股有限公司 | It is positioned based on light source |
CN105704652A (en) * | 2016-02-16 | 2016-06-22 | 中国科学院光电研究院 | Method for building and optimizing fingerprint database in WLAN/Bluetooth positioning processes |
CN107396321A (en) * | 2017-08-02 | 2017-11-24 | 江南大学 | Unsupervised formula indoor orientation method based on mobile phone sensor and iBeacon |
CN108632761A (en) * | 2018-04-20 | 2018-10-09 | 西安交通大学 | A kind of indoor orientation method based on particle filter algorithm |
CN110260855A (en) * | 2019-06-24 | 2019-09-20 | 北京壹氢科技有限公司 | A kind of indoor pedestrian navigation localization method merging pedestrian's dead reckoning, Geomagnetism Information and indoor map information |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113810850A (en) * | 2021-08-11 | 2021-12-17 | 荣耀终端有限公司 | Method for creating positioning database and electronic equipment |
CN113704377A (en) * | 2021-08-30 | 2021-11-26 | 中国地质大学(武汉) | Road network matching method combining artificial interaction and active learning |
CN113704377B (en) * | 2021-08-30 | 2024-03-01 | 中国地质大学(武汉) | Road network matching method combining manual interaction and active learning |
Also Published As
Publication number | Publication date |
---|---|
CN111556432B (en) | 2022-02-01 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109405839B (en) | Traffic network off-line map matching algorithm based on multiple paths | |
US10466056B2 (en) | Trajectory matching using ambient signals | |
CN108534779B (en) | Indoor positioning map construction method based on track correction and fingerprint improvement | |
CN111912416B (en) | Method, device and equipment for positioning equipment | |
CN107092877B (en) | Remote sensing image roof contour extraction method based on building base vector | |
CN110958575B (en) | Positioning method and system based on WiFi fusion prediction | |
CN108632761B (en) | Indoor positioning method based on particle filter algorithm | |
KR101311100B1 (en) | Method for recognizing the self position of a mobile robot unit using arbitrary ceiling features on the ceiling image/feature map | |
CN111491367B (en) | Indoor positioning method based on crowd sensing and multi-fusion technology | |
CN111556432B (en) | Crowdsourcing fingerprint database construction method based on map information screening and matching | |
CN109815831B (en) | Vehicle orientation obtaining method and related device | |
CN106767854B (en) | Mobile device, garage map forming method and system | |
CN110569861B (en) | Image matching positioning method based on point feature and contour feature fusion | |
CN110470295B (en) | Indoor walking navigation system and method based on AR positioning | |
Du et al. | CRCLoc: A crowdsourcing-based radio map construction method for WiFi fingerprinting localization | |
CN104501807B (en) | Indoor location method based on geomagnetic field and historical localization track | |
CN114639085A (en) | Traffic signal lamp identification method and device, computer equipment and storage medium | |
CN109739830B (en) | Position fingerprint database rapid construction method based on crowdsourcing data | |
CN111105435A (en) | Marker matching method and device and terminal equipment | |
CN111123333B (en) | Vehicle track positioning method fusing bayonet and GPS data | |
CN113554705B (en) | Laser radar robust positioning method under changing scene | |
CN111210628B (en) | Boundary road section selection method and device in urban reachable area calculation | |
Ye et al. | Self-localization of parking robots using square-like landmarks | |
CN110602658B (en) | Continuous positioning method | |
CN114608560B (en) | Passive combined indoor positioning system and method based on intelligent terminal sensor |
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 |