CN114630266A - Multimode data fusion indoor positioning system based on neural network - Google Patents
Multimode data fusion indoor positioning system based on neural network Download PDFInfo
- Publication number
- CN114630266A CN114630266A CN202011466281.1A CN202011466281A CN114630266A CN 114630266 A CN114630266 A CN 114630266A CN 202011466281 A CN202011466281 A CN 202011466281A CN 114630266 A CN114630266 A CN 114630266A
- Authority
- CN
- China
- Prior art keywords
- fingerprint
- segmented
- positioning
- sequence
- fingerprints
- 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
- 238000013528 artificial neural network Methods 0.000 title claims abstract description 10
- 230000004927 fusion Effects 0.000 title claims abstract description 7
- 238000000034 method Methods 0.000 claims abstract description 29
- 230000011218 segmentation Effects 0.000 claims abstract description 23
- 238000013145 classification model Methods 0.000 claims abstract description 17
- 238000013507 mapping Methods 0.000 claims description 13
- 238000012549 training Methods 0.000 claims description 11
- 238000000605 extraction Methods 0.000 claims description 5
- 238000011176 pooling Methods 0.000 claims description 4
- 230000008569 process Effects 0.000 claims description 4
- 230000005540 biological transmission Effects 0.000 claims description 2
- 230000000295 complement effect Effects 0.000 claims description 2
- 230000005358 geomagnetic field Effects 0.000 abstract description 2
- 230000006798 recombination Effects 0.000 abstract description 2
- 238000005215 recombination Methods 0.000 abstract description 2
- 239000011159 matrix material Substances 0.000 description 8
- 238000012545 processing Methods 0.000 description 8
- 238000005070 sampling Methods 0.000 description 8
- 238000010586 diagram Methods 0.000 description 6
- 238000005516 engineering process Methods 0.000 description 5
- 230000006870 function Effects 0.000 description 4
- 238000013527 convolutional neural network Methods 0.000 description 3
- 238000013135 deep learning Methods 0.000 description 3
- 238000011160 research Methods 0.000 description 3
- 230000001133 acceleration Effects 0.000 description 2
- 230000004913 activation Effects 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 2
- 238000006243 chemical reaction Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000001914 filtration Methods 0.000 description 2
- 238000005457 optimization Methods 0.000 description 2
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 description 1
- 230000000903 blocking effect Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000013480 data collection Methods 0.000 description 1
- 238000013136 deep learning model Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000004069 differentiation Effects 0.000 description 1
- 230000008034 disappearance Effects 0.000 description 1
- 230000005389 magnetism Effects 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 230000000873 masking effect Effects 0.000 description 1
- 210000002569 neuron Anatomy 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
- 238000002864 sequence alignment Methods 0.000 description 1
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
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/043—Architecture, e.g. interconnection topology based on fuzzy logic, fuzzy membership or fuzzy inference, e.g. adaptive neuro-fuzzy inference systems [ANFIS]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/044—Recurrent networks, e.g. Hopfield networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- 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
- H04W4/30—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
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Signal Processing (AREA)
- Computer Networks & Wireless Communication (AREA)
- Software Systems (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Biomedical Technology (AREA)
- Mathematical Physics (AREA)
- Health & Medical Sciences (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Automation & Control Theory (AREA)
- Computational Mathematics (AREA)
- Fuzzy Systems (AREA)
- Mathematical Analysis (AREA)
- Mathematical Optimization (AREA)
- Pure & Applied Mathematics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Collating Specific Patterns (AREA)
Abstract
The invention discloses a multi-mode data fusion indoor positioning system based on a neural network. The invention combines the geomagnetic field intensity value, the Wi-Fi signal intensity value and the traveling direction value together to form a multi-mode fingerprint, and has higher uniqueness in open space. The method adopts a Mi-character grid to divide a main path in an indoor space, then designs a fixed length fingerprint segmentation method based on sequence extreme points to segment the fingerprint of each main path, takes the extreme points as the marks for segmenting and aligning the segmented fingerprint sequence, and reduces complex and diverse tracks by a segmentation recombination idea while solving the problem of the matching dislocation of the segmented fingerprints. In order to more accurately identify the multi-modal fingerprints formed by combining a plurality of segmented fingerprints according to time sequence, the invention provides a multi-modal fingerprint classification model based on convolution and long-short term memory network, which can accurately match diversified fingerprint sequences and improve the positioning accuracy of the system in open space.
Description
Technical Field
The invention belongs to the field of indoor positioning, and relates to a deep learning-based multi-mode indoor positioning system and an implementation method thereof.
Background
In recent years, in an outdoor environment, Location-Based Service (LBS) applications that rely on a Global Positioning System (GPS) are emerging and are being penetrated into people's clothing and housing. However, in an Indoor environment, due to signal blocking and multipath effects, the GPS cannot meet the requirements of people for Indoor Location-Based Service (ILBS). Therefore, indoor positioning technology has been rapidly developed, and various signals are used in the indoor positioning technology, such as Wi-Fi (Wireless-Fidelity), Bluetooth (Bluetooth), Radio Frequency Identification (RFID), infrared, Ultra Wide Band (UWB), terrestrial magnetism, visible light, ultrasonic waves, Frequency Modulation (FM), Inertial Navigation System (INS), image, and multi-mode fusion technologies. The indoor positioning scheme needs to consider the deployment and maintenance cost of the scheme while considering the positioning accuracy. In contrast, the geomagnetic field-based positioning technology not only has stable signals and high positioning accuracy, but also does not require additional equipment, and is gradually the focus of research.
Most of the existing geomagnetic positioning schemes aim at unidirectional one-dimensional positioning, and in open indoor spaces such as airport terminal buildings, stations, shopping centers and the like, the indoor environment range has large span, is complex and changeable, and the walking track of pedestrians is also changed from one dimension to two dimensions and is difficult to enumerate one by one, so that the one-dimensional positioning algorithm is greatly limited in the environment. The existing research has the following defects: 1) the fingerprint acquisition cost is higher: the existing research needs the staff to survey in the location place, manually collect the fingerprint data in the space, and mark the position, establish the mapping relation of fingerprint and position. 2) The difficulty in establishing the mapping relationship between the complex trajectory and the fingerprint sequence is high: the wide indoor space is often larger in area and more in paths, tracks become complex and diverse due to unpredictability of walking directions of pedestrians, and the mapping relation between various fingerprint sequences and the corresponding tracks is difficult to establish. 3) Fingerprint sequence matching dislocation problem: the segmented fingerprints need to be aligned when being matched to ensure the accuracy of positioning, the sliding window is used for aligning to increase the calculation cost, the regression method is used for positioning, and due to the fact that the step length of each person is different, the sampling of each step is staggered with the data in the training set in actual positioning, and the problem of dislocation still exists.
Disclosure of Invention
The invention aims to solve the problems that the existing geomagnetic indoor positioning algorithm has high fingerprint acquisition cost, is difficult to establish the mapping relation between a complex track and a fingerprint sequence, has mismatching fingerprint sequences and the like. A fingerprint classification model is trained through a deep learning method, and a mapping relation between a position coordinate point and a geomagnetic fingerprint is established, so that a low-cost, efficient and accurate geomagnetic type indoor positioning system is realized.
The core technical idea of the invention is to realize positioning by matching and identifying geomagnetic fingerprints through a deep learning model, and the system comprises two stages of off-line training and on-line positioning. 1) An off-line training stage: collecting geomagnetic fingerprint information in space, training a geomagnetic fingerprint classification model after processing, and establishing a mapping relation between geomagnetic fingerprints and positions. 2) And (3) in an online positioning stage: after receiving a positioning request of a user, calculating a positioning result through a fingerprint classification model and a related algorithm and feeding back the positioning result to the user.
The core technology for solving the problems comprises a multi-mode fingerprint, a segmented combined fingerprint mapping method and a multi-mode fingerprint classification model.
(1) A multi-modal fingerprint. In order to improve the information content of the fingerprint, the invention combines the geomagnetism, the Wi-Fi and the direction value into a multi-mode fingerprint. The characteristic that Wi-Fi can provide coarse-grained positioning is utilized, the characteristic is complementary with geomagnetism and direction, and the problem that the geomagnetism fingerprint is similar in different places in an open environment is avoided.
(2) And (4) a segmented combined fingerprint mapping method. In order to ensure that the pedestrian track and the fingerprint information are mutually mapped in a refined manner and simultaneously reduce the fingerprint acquisition cost, the invention provides a segmentation recombination idea, which is characterized in that a main path of an indoor map is divided by adopting a grid in a shape of a Chinese character 'mi', and a continuous segmented path in eight main directions is used for approximately representing the walking track of a pedestrian. The invention provides a fixed-length fingerprint segmentation method based on sequence extreme points, which takes the extreme points as marks for segmentation and alignment of segmented fingerprint sequences, improves segmentation granularity of segmented fingerprints and restores complex and diverse tracks while solving the problem of matching dislocation of the segmented fingerprints.
(3) A multi-modal fingerprint classification model. In order to more accurately identify the multi-modal fingerprint formed by combining a plurality of segmented fingerprints according to time sequence, the invention provides a multi-modal fingerprint classification model based on a Convolutional Neural Network (CNN) and a long-short term memory network (LSTM). Because the fingerprint feature extraction in-process, no matter be the biggest pooling or average pooling can all reduce or fuzzy information, the cavity convolution can expand the receptive field when not losing fingerprint information, consequently uses the cavity convolution to carry out the feature extraction to every section fingerprint of walking in-process, through the information of LSTM transmission a plurality of segmentation fingerprints on the chronogenesis, the model that the training generated can carry out accurate matching to diversified fingerprint sequence, promotes the positioning accuracy in two-dimensional space.
Drawings
FIG. 1 is a diagram of a multi-modal data fusion indoor positioning system based on neural network;
FIG. 2 is a diagram illustrating a division of a main path of an open space according to the present invention;
FIG. 3 is an exemplary diagram of a fixed-length fingerprint segmentation based on sequence extremum points according to the present invention;
FIG. 4 is an exemplary diagram of a walking trajectory and corresponding segmented fingerprints according to the present invention;
FIG. 5 is a diagram of a multi-modal fingerprint classification model according to the present invention;
Detailed Description
The invention is further described with reference to the accompanying drawings, and the system structure is shown in fig. 1, and mainly includes four parts, namely, multi-modal data acquisition, multi-modal fingerprint processing, multi-modal fingerprint classification model training, and real-time position estimation. It should be noted that the description of the embodiments is only for the purpose of facilitating understanding of the present invention, and does not limit the present invention.
1. Multi-modal data collection
Due to the fact that walking paths in the open space are complex and various, fingerprint sequences of all possible paths are difficult to acquire in a field acquisition mode, and the difficulty in establishing the mapping relation between complex tracks and the fingerprint sequences is greatly increased. In order to ensure that the acquired fingerprint data set can cover the whole indoor space, provide better positioning accuracy and not cause too high cost of acquired data, the invention divides the map by adopting a grid in the shape of a Chinese character 'mi', and continuously segmented paths in eight main directions are used for approximately representing all paths in the space. The whole map is covered by the grid in the shape of a Chinese character 'mi', and the grid is regarded as a whole, wherein each straight line is a main path. Fingerprint collection is carried out on the main paths, and all positions and paths where pedestrians walk in a two-dimensional space are covered as far as possible.
2. Multimodal fingerprint processing
The invention provides a curve difference matrix, which replaces absolute numerical values with differences among sampling points to describe fingerprintsThe sequences form the shape of curves, thereby enabling the neural network to better perceive the high-dimensional spatial features of the data. Compared with the normalization processing, the degree of differentiation between the sequences is reserved. Converting the segmented fingerprints into a curve difference matrix, and assuming that the segmented geomagnetic fingerprint sequence is Sxyz＝{g1,g2,…,gnThe curve difference matrix D is shown in equation 1:
after conversion, a two-dimensional image with richer contents is formed, different types of curves correspond to different curve difference value matrix diagrams, and the image category is clear. The horizontal and vertical coordinates of the matrix all represent the coordinates of sampling points of the sequence, and the difference value between the sampling points has high-dimensional characteristics in space, so that the condition of classifying by using an image processing mode is met, and the feature extraction can be performed by adopting a convolutional neural network to realize the classification of the geomagnetic fingerprint sequence.
The mobile phone can acquire the name of a Wi-Fi AP nearby the mobile phone through a built-in Wi-Fi module, and also can acquire an MAC address and an RSS value. The MAC address is the unique identification of the Wi-Fi AP, so that the condition of duplicate names of the APs can be effectively avoided. The RSS values indicate signal strength, with negative values, with larger values indicating less signal attenuation. The RSS value is the main basis for estimating the distance between the mobile phone and the Wi-Fi AP. The single AP information can be expressed as w ═ MAC (media access control, RSS), the mobile phone position determination needs to be carried out according to the fingerprint information of multiple APs, fingerprints acquired by pedestrians in the walking process form a Wi-Fi fingerprint sequence, and finally the position in the actual environment is associated with the Wi-Fi fingerprint sequence to form a Wi-Fi fingerprint model. The Wi-Fi fingerprint sequence can therefore be expressed as:
where m represents the total number of Wi-Fi APs that can be scanned and n represents the number of samples of the fingerprint sequence. Wi-Fi is sampled in the same place, the collected RSS value fluctuates to a certain degree, and sampling values of different models of mobile phones in the same place are different to a certain degree. If the model is directly trained by using the collected data, the model is difficult to converge due to the fact that the value is large and difficult and deviation exists, so that the Wi-Fi RSS value needs to be normalized, as shown in formula 3:
and dimension is cancelled, influence caused by data fluctuation and equipment difference is reduced, and model convergence is accelerated.
The fixed-length geomagnetic fingerprint segmentation method based on the sequence extreme points firstly needs to mark all effective extreme points of a fingerprint sequence before segmentation. Different extreme points exist in intervals with different sizes of the acquired fingerprint sequence, and the heights and the widths of waveforms corresponding to the different extreme points are different, wherein the different extreme points contain some pseudo wave peaks (wave troughs) formed by the outside, and the pseudo wave peaks cannot be used as mark points for sequence alignment due to large variation and poor reliability. In order to eliminate these false peaks and select the effective extreme points, the spacing G between peaks and the peak H need to be limited. The fingerprint sequence shows fine fluctuation due to the fact that the magnetic field has fine fluctuation and is interfered by external natural or human factors. To mitigate the interference of these noises, a sliding average filtering is first performed on the fingerprint sequence. In addition, in order to ensure that stable wave crests and wave troughs are selected, the wave crest value needs to meet the requirement of Hmax>Mu, trough value needs to satisfy Hmin<μ, where μ is the mean of the magnetic fingerprints of the main path. In order to reduce the redundant amount of fingerprints, the distance between wave peaks is specified to satisfy G under the inspiration of overlapped segmentation of a DCGIL algorithm>SSN/2. After the parameter conditions are met, all qualified extreme points can be marked by traversing the whole fingerprint sequence. After the screening rule and the segmentation length of the effective extreme point of the geomagnetic fingerprint sequence are determined, the geomagnetic fingerprint sequence can be segmented. The core idea of the segmentation method is to traverse the geomagnetic fingerprint sequence of a section of main path to mark all effective extreme points, and then traverse every two phases in the traveling directionAnd adjacent extreme points, namely the number of segments between two extreme points after the sampling number between the two extreme points is divided by SSN (single sign language) rounding, the segments are sequentially segmented from the two extreme points to the center, and finally the segments are segmented according to the number of segments meeting the distance between the head extreme point and the tail extreme point of the sequence, and all the segmented sequence segments are marked with position labels. According to the segmentation method, the path is carved by smaller segments, so that the walking length required by positioning is balanced, the positioning cost is reduced, the dislocation problem during sequence matching is solved, and the overall positioning accuracy is improved, as shown in fig. 3.
After the geomagnetic fingerprint sequences of all main paths in all directions are segmented by the method, all segmented geomagnetic fingerprint sequences in the indoor space are obtained. The Wi-Fi data collected on one section of main path are also sequences, and after sampling frequency conversion, the Wi-Fi fingerprint sequences are segmented by the same length to obtain Wi-Fi fingerprints of each segment. After each segmented geomagnetic fingerprint sequence is converted into a corresponding curve difference matrix D, the corresponding normalized Wi-Fi fingerprint W and the discretized traveling direction value O jointly form a multi-mode segmented fingerprint MSFP (D, W, O). After a plurality of adjacent multi-mode segmented fingerprints are spliced in sequence, a multi-mode fingerprint MFP (MSFP) of a walking path is formed1,…,MSFPn}。
In an open indoor space, due to the fact that tracks of pedestrians are complex and diverse and the track lengths are not fixed, the track fingerprints can be formed by serially connecting various different segmented fingerprints, and diversified fingerprint sequences with different lengths (the types and the number of MSFPs forming the MFP are not constant) need to be prepared to serve as a data set for training a fingerprint classification model. If the sampling cost is very high by the actual walking of people, a Random Way Point (RWP) model is adopted to automatically simulate the walking process, and various possible two-dimensional tracks are generated. As shown in fig. 4, the dotted line in the figure is an example of a walking trajectory simulated by RWP, the line segment (MSFP) is an example of a fingerprint path segment on the walking path, and the fork is an extreme point for alignment.
3. Multi-modal fingerprint classification model training
The specific structure of the multi-modal fingerprint classification model is shown in fig. 5, and the model is mainly formed by combining cavity convolution and LSTM, and the input of the model is divided into a curve difference matrix of a segmented geomagnetic fingerprint sequence, a Wi-Fi fingerprint sequence W and a direction sequence O. And generating a one-dimensional tensor after the characteristics of the segmented geomagnetic fingerprint sequence are extracted by cavity convolution, splicing W and O, sending the W and O together as fusion data of the segmented fingerprint into an LSTM, updating the state of a neuron after the input of the LSTM is processed by each gate of the LSTM, and finally outputting the confidence coefficient of the current segmented fingerprint sequence at each position coordinate after a full connection layer (FC). Conv denotes the convolutional layer, and the hole convolution is still used, and the expansion ratio is the sawtooth structure of 1, 2, 5, 1, 2. The time step of the LSTM is 10, and if the length of the input sequence is insufficient, the input sequence is complemented with 0, and in order to reduce the influence of noise addition on the model, a Masking layer is added in front of the LSTM, and the time step complemented with 0 is filtered out. All Conv are followed by a BN to prevent gradient disappearance and speed up training. ReLU is used as the activation function except for the last layer. Since the multi-classification task is adopted, the activation function of the last layer FC adopts a Softmax function. A Keras deep learning framework is selected as a model framework, TensorFlow is used at the rear end, an Adam optimization algorithm is selected by an optimization method of a neural network, and a catagorical _ cross is selected as a loss function.
4. Real-time position estimation
When real-time positioning is carried out in a real scene, in order to ensure that the segmented fingerprint sequences sent into the classification model do not have the dislocation problem, the real-time fingerprint sequences need to be segmented in advance by a fixed-length fingerprint segmentation method based on sequence extreme points, effective segmented fingerprints are screened out and combined according to a time sequence, and then position points are predicted. Firstly, the geomagnetic fingerprint sequence and the acceleration sequence are subjected to moving average filtering processing, and noise in original data is filtered. And then mapping the direction change points in the acquired direction sequence to geomagnetic sequences, Wi-Fi sequences and acceleration sequences, and dividing the sequences into subsequences of each direction respectively. Sequentially processing fingerprint sequences in different directions, finding out all extreme points meeting the requirements of the distance G and the height H in the current fingerprint sequence, calculating whether the distance from the starting point to the first extreme point is greater than FSL, if so, segmenting the fingerprint of the length of FSL before the first extreme point, converting the geomagnetic fingerprint into a curve difference matrix, normalizing the Wi-Fi fingerprint, and adding a direction value to obtain MSFP. And then, taking all the extreme points as base points in sequence, carrying out fingerprint segmentation of the FSL length forward, stopping the segmentation of the current extreme point if the segmentation sequence contains the next extreme point, and converting and processing the data after the segmentation. The last segment requires the calculation of the longitudinal difference ld from the end of the whole sequence. After the circulation is finished, all the obtained MSFPs are spliced into the MPF according to the time sequence. And finally, the MPF is sent into a multi-mode fingerprint classification model, all category confidence coefficients are obtained after prediction is carried out by the model, and the final positioning result of the positioning algorithm is calculated according to the position of the highest confidence coefficient and the ld.
The user usage scenario of the invention is as follows:
in some large buildings, the walking path is complex, and pedestrians often get lost of position in the space, and need to obtain the location-based service in real time, that is, know the position of the pedestrian or other people or objects in the space. In the scene, the method has better effect, and has higher positioning accuracy, lower required equipment deployment cost and lower positioning walking cost compared with the existing method.
Claims (5)
1. A multimode data fusion indoor positioning system based on a neural network is characterized in that: the accuracy and efficiency of real-time positioning are improved under the condition of lower positioning cost through the model constructed by the neural network. The system comprises the mobile terminal and the server terminal, and after a user initiates a positioning request through the mobile terminal, the server terminal can quickly return a positioning result, so that good positioning experience is provided for the user.
The mobile terminal is divided into a fingerprint acquisition module and a real-time positioning module, a multi-mode fingerprint sequence of a main path in indoor space is acquired through the fingerprint acquisition module of the mobile terminal before a positioning system is used, then a segmented combined fingerprint mapping method is adopted to segment the multi-mode fingerprint of the main path, and then all fingerprints are processed respectively and are used as a data set after effective combination. And training a multi-mode fingerprint classification model based on a neural network at a server by using the data set, and establishing a mapping relation between the multi-mode fingerprint and the position coordinate. After a user sends a positioning request through the mobile terminal real-time positioning module, the server terminal firstly processes various sensing data acquired in real time, calculates the walking distance of the user, processes the multi-mode fingerprint and sends the processed multi-mode fingerprint to the multi-mode fingerprint classification model, predicts the most possible positioning position point, and finally returns the result to the mobile terminal of the user after position correction.
2. The multimodal fingerprint model of claim 1 wherein: and combining the geomagnetism, the Wi-Fi and the direction value into a multi-modal fingerprint. The characteristic that Wi-Fi can provide coarse-grained positioning is utilized, the characteristic is complementary with geomagnetism and direction, and the problem that the geomagnetism fingerprint is similar in different places in an open environment is avoided.
3. The segmented combined fingerprint mapping method according to claim 1, wherein: the method is characterized in that a grid in a shape of a Chinese character 'mi' is adopted to divide a main path of an indoor map, and continuous effective segmented paths in eight main directions are used for approximately representing the walking track of a pedestrian. The fingerprint sequence is segmented by adopting a fixed-length fingerprint segmentation method based on sequence extreme points, and then segmented fingerprints are combined into various tracks by using a random waypoint model.
4. The fixed-length fingerprint segmentation method based on the sequence extreme points as claimed in claim 3, wherein: the extreme point is used as a mark for segmenting and aligning the segmented fingerprint sequence, so that the problem of matching dislocation of the segmented fingerprints is solved, the segmentation granularity of the segmented fingerprints is improved, and complex and diverse tracks are restored.
5. The multi-modal neural network-based fingerprint classification model as claimed in claim 1, wherein: because the fingerprint characteristic extraction in-process, no matter be the biggest pooling or average pooling can all reduce or fuzzy information, the hole convolution can enlarge the receptive field when not losing fingerprint information, consequently uses the hole convolution to carry out the characteristic extraction to every section fingerprint of walking in-process, through the information of LSTM transmission a plurality of segmentation fingerprints on the chronogenesis, the model that the training generated can carry out accurate matching to diversified fingerprint sequence, promotes the positioning accuracy in two-dimensional space.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011466281.1A CN114630266B (en) | 2020-12-13 | 2020-12-13 | Multi-mode data fusion indoor positioning system based on neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011466281.1A CN114630266B (en) | 2020-12-13 | 2020-12-13 | Multi-mode data fusion indoor positioning system based on neural network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114630266A true CN114630266A (en) | 2022-06-14 |
| CN114630266B CN114630266B (en) | 2024-06-28 |
Family
ID=81897231
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011466281.1A Active CN114630266B (en) | 2020-12-13 | 2020-12-13 | Multi-mode data fusion indoor positioning system based on neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114630266B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116381753A (en) * | 2023-06-01 | 2023-07-04 | 北京航空航天大学 | Neural network assisted navigation method of GNSS/INS integrated navigation system during GNSS interruption |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20160183116A1 (en) * | 2014-12-18 | 2016-06-23 | Alibaba Group Holding Limited | Method and apparatus of positioning mobile terminal based on geomagnetism |
| CN110856112A (en) * | 2019-11-14 | 2020-02-28 | 深圳先进技术研究院 | Crowd-sourcing perception multi-source information fusion indoor positioning method and system |
| CN111149141A (en) * | 2017-09-04 | 2020-05-12 | Nng软件开发和商业有限责任公司 | Method and apparatus for collecting and using sensor data from a vehicle |
-
2020
- 2020-12-13 CN CN202011466281.1A patent/CN114630266B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20160183116A1 (en) * | 2014-12-18 | 2016-06-23 | Alibaba Group Holding Limited | Method and apparatus of positioning mobile terminal based on geomagnetism |
| CN111149141A (en) * | 2017-09-04 | 2020-05-12 | Nng软件开发和商业有限责任公司 | Method and apparatus for collecting and using sensor data from a vehicle |
| CN110856112A (en) * | 2019-11-14 | 2020-02-28 | 深圳先进技术研究院 | Crowd-sourcing perception multi-source information fusion indoor positioning method and system |
Non-Patent Citations (4)
| Title |
|---|
| DOCUMENT 5D/TEMP/536: "The technical feasibility of IMT in the bands above 6 GHz", INTERNATIONAL TELECOMMUNICATION UNION, 24 February 2015 (2015-02-24) * |
| 刘诗佳: "基于WPF和Silverlight的动态应用程序研究与实现", CNKI优秀硕士学位论文全文库, 15 January 2013 (2013-01-15) * |
| 杨彦兵: "电动助力车用新型智能充电器控制算法研究及系统实现", CNKI优秀硕士学位论文全文库, 15 January 2014 (2014-01-15) * |
| 王照远;高井祥;李增科;常梦阳;: "一种融合WiFi/地磁/PDR的室内定位方法", 测绘通报, no. 06, 25 June 2020 (2020-06-25) * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116381753A (en) * | 2023-06-01 | 2023-07-04 | 北京航空航天大学 | Neural network assisted navigation method of GNSS/INS integrated navigation system during GNSS interruption |
| CN116381753B (en) * | 2023-06-01 | 2023-08-15 | 北京航空航天大学 | Neural network assisted navigation method of GNSS/INS integrated navigation system during GNSS interruption |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114630266B (en) | 2024-06-28 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110837778B (en) | Traffic police command gesture recognition method based on skeleton joint point sequence | |
| CN110675418B (en) | Target track optimization method based on DS evidence theory | |
| CN104915636B (en) | Remote sensing image road recognition methods based on multistage frame significant characteristics | |
| CN103162682B (en) | Based on the indoor path navigation method of mixed reality | |
| CN114419152B (en) | Target detection and tracking method and system based on multi-dimensional point cloud characteristics | |
| CN110856112A (en) | Crowd-sourcing perception multi-source information fusion indoor positioning method and system | |
| CN105636201B (en) | Indoor orientation method based on sparse signal fingerprint database | |
| CN110073362A (en) | System and method for lane markings detection | |
| CN107182036A (en) | The adaptive location fingerprint positioning method merged based on multidimensional characteristic | |
| CN109116298B (en) | Positioning method, storage medium and positioning system | |
| CN112580479A (en) | Geomagnetic indoor positioning system based on cavity convolution neural network | |
| CN115808170B (en) | Indoor real-time positioning method integrating Bluetooth and video analysis | |
| US11724721B2 (en) | Method and apparatus for detecting pedestrian | |
| CN114241464A (en) | Cross-view image real-time matching geographic positioning method and system based on deep learning | |
| Feng et al. | Visual Map Construction Using RGB‐D Sensors for Image‐Based Localization in Indoor Environments | |
| CN110470295A (en) | A kind of indoor walking navigation and method based on AR positioning | |
| CN110390706A (en) | A kind of method and apparatus of object detection | |
| CN108629295A (en) | Corner terrestrial reference identification model training method, the recognition methods of corner terrestrial reference and device | |
| CN113596989A (en) | Indoor positioning method and system for intelligent workshop | |
| CN115511210A (en) | Community life circle defining method and device based on multi-source big data | |
| CN114743128B (en) | Multi-mode northeast tiger re-identification method and device based on heterogeneous neural network | |
| He et al. | Online semantic-assisted topological map building with LiDAR in large-scale outdoor environments: Toward robust place recognition | |
| CN114630266B (en) | Multi-mode data fusion indoor positioning system based on neural network | |
| CN114155489A (en) | Multi-device cooperative unmanned aerial vehicle flyer detection method, device and storage medium | |
| CN114615468A (en) | Intelligent positioning method based on security monitoring |
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 |