CN112729301A - Indoor positioning method based on multi-source data fusion - Google Patents
Indoor positioning method based on multi-source data fusion Download PDFInfo
- Publication number
- CN112729301A CN112729301A CN202011434444.8A CN202011434444A CN112729301A CN 112729301 A CN112729301 A CN 112729301A CN 202011434444 A CN202011434444 A CN 202011434444A CN 112729301 A CN112729301 A CN 112729301A
- Authority
- CN
- China
- Prior art keywords
- data
- wireless signal
- obtaining
- time
- indoor positioning
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 63
- 230000004927 fusion Effects 0.000 title claims abstract description 38
- 239000002245 particle Substances 0.000 claims description 55
- 238000013528 artificial neural network Methods 0.000 claims description 20
- 230000002457 bidirectional effect Effects 0.000 claims description 6
- 230000015654 memory Effects 0.000 claims description 6
- 238000013507 mapping Methods 0.000 claims description 3
- 238000006073 displacement reaction Methods 0.000 abstract description 4
- 238000005516 engineering process Methods 0.000 description 8
- 238000005259 measurement Methods 0.000 description 8
- 238000012549 training Methods 0.000 description 7
- 238000004364 calculation method Methods 0.000 description 5
- 230000005484 gravity Effects 0.000 description 3
- 238000012952 Resampling Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000001914 filtration Methods 0.000 description 2
- 238000004519 manufacturing process Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 230000000007 visual effect Effects 0.000 description 2
- 238000007476 Maximum Likelihood Methods 0.000 description 1
- 230000003044 adaptive effect Effects 0.000 description 1
- 230000002860 competitive effect Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000007787 long-term memory Effects 0.000 description 1
- 230000007774 longterm Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000006403 short-term memory Effects 0.000 description 1
- 230000000087 stabilizing effect Effects 0.000 description 1
- 239000000126 substance Substances 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01C—MEASURING DISTANCES, LEVELS OR BEARINGS; SURVEYING; NAVIGATION; GYROSCOPIC INSTRUMENTS; PHOTOGRAMMETRY OR VIDEOGRAMMETRY
- G01C21/00—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00
- G01C21/20—Instruments for performing navigational calculations
- G01C21/206—Instruments for performing navigational calculations specially adapted for indoor navigation
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01C—MEASURING DISTANCES, LEVELS OR BEARINGS; SURVEYING; NAVIGATION; GYROSCOPIC INSTRUMENTS; PHOTOGRAMMETRY OR VIDEOGRAMMETRY
- G01C21/00—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00
- G01C21/10—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00 by using measurements of speed or acceleration
- G01C21/12—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00 by using measurements of speed or acceleration executed aboard the object being navigated; Dead reckoning
- G01C21/16—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00 by using measurements of speed or acceleration executed aboard the object being navigated; Dead reckoning by integrating acceleration or speed, i.e. inertial navigation
- G01C21/165—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00 by using measurements of speed or acceleration executed aboard the object being navigated; Dead reckoning by integrating acceleration or speed, i.e. inertial navigation combined with non-inertial navigation instruments
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
- G01S11/00—Systems for determining distance or velocity not using reflection or reradiation
- G01S11/02—Systems for determining distance or velocity not using reflection or reradiation using radio waves
- G01S11/06—Systems for determining distance or velocity not using reflection or reradiation using radio waves using intensity measurements
Abstract
The invention discloses an indoor positioning method based on multi-source data fusion, which comprises the following steps: obtaining inertial navigation data, and obtaining motion data according to the inertial navigation data, wherein the motion data is a position distance change value and a course change value; receiving a wireless signal, and obtaining a first position according to the strength of the wireless signal, wherein the wireless signal is generated by an indoor wireless signal transmitter; and obtaining a second position according to the motion data and the first position, wherein the accuracy of the second position is higher than that of the first position. Compared with the traditional inertial navigation method, the method has the advantages that the predicted displacement vector contains more accurate motion information under different terminal use modes, and the indoor positioning precision is effectively improved.
Description
Technical Field
The invention relates to the technical field of indoor positioning, in particular to an indoor positioning method based on multi-source data fusion.
Background
In recent years, the demand for location-based services has rapidly increased, making indoor positioning attractive to a lot of interest. The popularization of smart phones and the advantages of various built-in sensors, support of abundant radio frequency signals, convenience in carrying and the like make the indoor positioning technology based on smart phones become a great research hotspot.
The implementation of inertial navigation and positioning on smartphones has been a major research task, with the task of navigation and positioning using measurements provided by inertial measurement units. However, the small and cheap inertia measurement unit provided in the smart phone makes the inertial navigation based on the smart phone suffer from high sensor noise and bias to generate system drift, which cannot well meet the indoor positioning requirement. In order to obtain more competitive performance, the data-driven inertial navigation technology utilizes time sequence observation data collected by an intelligent mobile phone inertial measurement unit to be associated with a real motion track acquired by an optical motion capture system or a visual inertial odometer, returns motion parameters (speed, course and the like) back and forth, and converts inertial tracking into a sequence learning problem, so that the robustness of the inertial navigation technology is superior to that of the traditional inertial navigation technology (strapdown inertial navigation, pedestrian dead reckoning and the like) under different mobile phone use modes (handholding, swinging, calling, pocket and the like).
However, when the starting point is unknown, inertial navigation can only acquire relative coordinates, not absolute coordinates in the global coordinate system, and as the inertial tracking process continues, the system will inevitably suffer from drift. Positioning technologies (WiFi positioning, Bluetooth positioning, geomagnetic positioning and the like) based on other positioning signal sources which may be arranged indoors are proposed to deal with the challenges of inertial navigation, and although the difficulty of inertial navigation is solved, the positioning errors are still large due to the multipath effect and the influence of obstacles in the indoor environment. Therefore, the positioning method that only relies on a single sensor cannot well meet the requirement of indoor positioning.
Accordingly, the prior art is yet to be improved and developed.
Disclosure of Invention
In view of the above-mentioned shortcomings of the prior art, the present invention provides an indoor positioning method based on multi-source data fusion, aiming to improve the accuracy of indoor positioning.
The technical scheme of the invention is as follows:
an indoor positioning method based on multi-source data fusion, comprising the following steps:
obtaining inertial navigation data, and obtaining motion data according to the inertial navigation data, wherein the motion data is a position distance change value and a course change value;
receiving a wireless signal, and obtaining a first position according to the strength of the wireless signal, wherein the wireless signal is generated by an indoor wireless signal transmitter;
and obtaining a second position according to the motion data and the first position, wherein the accuracy of the second position is higher than that of the first position.
The indoor positioning method based on multi-source data fusion, wherein the inertial navigation data comprise accelerometer data, gyroscope data and gravimeter data, and the motion data of the terminal device are obtained according to the inertial navigation data, and the method comprises the following steps:
mapping the accelerometer data and the gyroscope data to a coordinate system where the gravimeter data is located to obtain updated accelerometer data and updated gyroscope data;
inputting the updated accelerometer data and the updated gyroscope data into a trained deep neural network, and obtaining motion data according to the trained deep neural network.
The indoor positioning method based on the multi-source data fusion is characterized in that the trained deep neural network comprises two bidirectional long-short term memory layers, two dropout layers and a full connection layer; and a dropout layer is arranged behind each bidirectional long-short term memory layer.
The indoor positioning method based on the multi-source data fusion is characterized in that the single acquisition time of the inertial navigation data is 0.5s-2.5 s.
The indoor positioning method based on multi-source data fusion, wherein the wireless signals are at least three different wireless signals, and when the intensities of the at least three different wireless signals are all greater than a preset threshold value, a first position is obtained according to the intensities of the wireless signals, and the method comprises the following steps:
and obtaining a first position according to the positions of the at least three wireless signal transmitters and the wireless signal strengths of the at least three wireless signal transmitters.
The indoor positioning method based on multi-source data fusion, wherein the wireless signal is at least one wireless signal, and when the intensity of the at least one wireless signal is less than or equal to a preset threshold, the obtaining of the first position according to the intensity of the wireless signal includes:
and obtaining a first position according to a pre-made radio map and the strength of the wireless signal.
The indoor positioning method based on multi-source data fusion, wherein the method for manufacturing the pre-manufactured radio map comprises the following steps:
receiving the intensity of at least two wireless signals of a wireless signal transmitter at a preset position in a room;
and obtaining a radio map according to the strength of the at least two wireless signals and the preset position.
The indoor positioning method based on multi-source data fusion is characterized in that the wireless signals are Bluetooth signals, the wireless signal transmitters are Bluetooth beacons, and the intensity of the wireless signals is as follows:
wherein λ represents the distance between the bluetooth beacon and the terminal device, RSS (λ) represents the signal strength of the bluetooth beacon, λ0Denotes the reference distance, RSS (lambda)0) Indicating the reference distance lambda0Signal strength of the Bluetooth beacon, η represents the path loss exponent, XσDenotes zero mean Gaussian variance as σ2The noise of (2).
The indoor positioning method based on multi-source data fusion, wherein obtaining a second position according to the motion data and the first position, comprises:
obtaining a second position through a particle filter fusion model according to the motion data and the first position
The indoor positioning method based on multi-source data fusion is characterized in that the particle filter fusion model is as follows:
wherein, XtAbscissa, Y, representing terminal device at time ttThe ordinate of the terminal device at time t is indicated,
the abscissa of the ith particle at time t-1 is shown,
denotes the ordinate, l, of the ith particle at time t-1tIndicating the change in the position distance, ψ, acquired at time ttIndicating the heading taken at time t,
representing the weight of the ith particle at the time t, wherein N represents the number of the particles;
and determining the weight of the ith particle at the time t according to the position of the ith particle at the time t-1 and the first position at the time t.
Has the advantages that: the indoor positioning method based on multi-source data fusion provided by the invention comprises the following steps: obtaining inertial navigation data, and obtaining motion data according to the inertial navigation data, wherein the motion data is a position distance change value and a course change value; receiving a wireless signal, and obtaining a first position according to the strength of the wireless signal, wherein the wireless signal is generated by an indoor wireless signal transmitter; and obtaining a second position according to the motion data and the first position, wherein the accuracy of the second position is higher than that of the first position. Compared with the traditional inertial navigation method, the method has the advantages that the predicted displacement vector contains more accurate motion information under different terminal use modes, and the indoor positioning precision is effectively improved.
Drawings
Fig. 1 is a schematic flow chart of an indoor positioning method based on multi-source data fusion according to the present invention.
Fig. 2 (a-c) are schematic diagrams of trilateration of an indoor positioning method based on multi-source data fusion according to the present invention.
Detailed Description
The invention provides an indoor positioning method based on multi-source data fusion, and the invention is further described in detail below in order to make the purpose, technical scheme and effect of the invention clearer and clearer. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
Referring to fig. 1, the indoor positioning method based on multi-source data fusion of the present invention includes:
and S10, acquiring inertial navigation data, and acquiring motion data of the terminal equipment according to the inertial navigation data.
Specifically, the motion data refers to a position distance change value and a heading change value of the terminal device, the position distance change value refers to a change of a moving distance of the terminal device within a short time (e.g., 1s, 2s), and the heading change value refers to a direction of the terminal device within the short time. The terminal equipment acquires inertial navigation data, predicts a position distance change value and a course change value of the terminal equipment through the inertial navigation data, and is fused with other positioning methods, such as signal source positioning data of Bluetooth, WiFi, geomagnetism and the like. Further, the terminal device includes, but is not limited to, a smart phone, a smart watch, or a smart bracelet.
In one implementation, the inertial navigation data includes accelerometer data, gyroscope data, and gravimeter data; step S10 includes:
s11, mapping the accelerometer data and the gyroscope data to a coordinate system where the gravimeter data are located to obtain updated accelerometer data and updated gyroscope data;
and S12, inputting the updated accelerometer data and the updated gyroscope data into a trained deep neural network, and obtaining motion data according to the trained deep neural network.
In particular, the arbitrary heading of the terminal device makes it difficult for inertial navigation to infer reliable motion parameters using only raw accelerometer and gyroscope data, so stable sensor data will make the regression task of data-driven inertial navigation easier. By introducing the gravity meter measurement, the roll and pitch ambiguity of the mobile phone is eliminated, the y axis of the terminal equipment is aligned with the weight bearing course, namely, the accelerometer data and the gyroscope data are aligned (mapped) on a coordinate system determined by the gravity data, and the accelerometer data and the gyroscope data are represented on a fixed coordinate system.
The invention learns the relation between inertial navigation data and motion parameters through Deep Neural Network (DNN) to carry out regression of position distance and course. The training data of the deep neural network is formed into time sequence observation data recorded by the inertial measurement unit which is marked with a real motion parameter label. The acquisition process of the training data comprises the following steps: firstly, acquiring a real motion track parameter of the terminal equipment by adopting a visual inertial odometer system, wherein the real motion track parameter has no displacement in a vertical course (the data of a gravimeter is 0), acquiring original sensor data of an accelerometer and a gyroscope corresponding to the real motion track parameter by adopting an inertial measurement unit, and adding time marks to the real motion track parameter and the original sensor data. And then, converting the posture data of the real motion track parameters into stable position distance and heading characteristics, and synchronizing the aligned accelerometer and gyroscope data into a time stamp corresponding to the stable distance and the heading characteristics through linear interpolation.
In the process of training the network, the initial deep neural network obtains predicted motion data (a distance change value and a course change value) corresponding to the original sensor data according to the original sensor data, then an adaptive moment estimation optimizer is adopted to minimize the mean square error between the predicted motion data and real motion data (a position distance and a course characteristic corresponding to the original sensor data), parameters of the initial deep neural network are corrected, and the step of obtaining the distance change value and the course change value corresponding to the original sensor data according to the original sensor data is continuously executed until the training condition of the initial deep neural network meets a preset condition, so that the trained deep neural network is obtained.
Obtaining corresponding position distance change values and heading change values through the trained deep neural network by aligning accelerometer data and gyroscope data within a finite time window, the process being represented as:
where Δ l represents a positional distance variation value, Δ ψ represents a heading variation value, aiRepresenting accelerometer data, wiRepresenting gyroscope data, DNN representing a trained deep neural network, and n representing a finite time window. The limited time window refers to a time period during which inertial navigation data is acquired a single time, for example: and if the time from the starting point to the end point of the terminal equipment is 2s, acquiring inertial navigation data in the 2s, and predicting a position distance change value and a course change value in the 2s through the trained deep neural network, wherein the single acquisition time of the inertial navigation data is 2 s. Preferably, the single acquisition time of the inertial navigation data is 0.5s-2.5s, and too short acquisition time increases the calculation load, and too long acquisition time causes the positioning accuracy to be reduced.
The method not only utilizes the deep neural network with great potential in the aspect of model-free generalization to regress the motion characteristics of the terminal equipment, but also effectively combines the method for stabilizing the data of the inertial measurement unit by using gravity, thereby ensuring that the network is more stable without losing accuracy, and increasing the adaptability of the method to users and use modes.
In one embodiment, the trained deep neural network includes two bidirectional long-short term memory layers, two dropout layers, and one fully connected layer.
In particular, in consideration of the accuracy and practicality of the network, the long-term dependency can be better utilized by adopting two bidirectional long-term and short-term memory layers as core building blocks. Meanwhile, in order to alleviate the overfitting problem, a dropout layer with input elements randomly set to zero is placed after each long-short term memory layer. A fully connected layer is added at the end of all layers to perform regression of the position distance and heading change values. It should be noted that the accuracy of the motion data predicted by the data-driven inertial navigation directly affects the final positioning performance, so that the neural network framework can be adjusted according to the actual training data.
And S20, receiving a wireless signal, and obtaining a first position according to the strength of the wireless signal, wherein the wireless signal is generated by an indoor wireless signal transmitter.
Specifically, the strength of the wireless signal received by the terminal device may reflect the distance between the terminal device and the wireless signal transmitter, so that the location of the terminal device may be known through the strength of the wireless signal received by the terminal device.
In one implementation, the wireless signals are at least three different wireless signals, and when the strengths of the at least three different wireless signals are greater than a preset threshold, step S20 includes: and obtaining a first position according to the positions of the at least three wireless signal transmitters and the wireless signal strengths of the at least three wireless signal transmitters.
Specifically, when the strengths of at least three different wireless signals are greater than a preset threshold, the terminal device may directly perform position calculation by using a trilateration method, where the trilateration method needs to obtain the at least three different wireless signals to perform position calculation. Referring to fig. 2 (a-c), trilateration uses the geometry of a circle to calculate the location of a terminal device, and if the location of a wireless signal transmitter is known, the location of the terminal device is (x, y), and the location of the wireless signal transmitter is (x, y)i，yi) A system of N equations can be obtained:
(x-xi)2+(y-yi)2＝λi，i＝1，2，...，N
wherein λ isiIs the distance of the wireless signal transmitter to the terminal device. By solving the system of equations, a first position (x, y) of the terminal device can be obtained. Under ideal conditions (see fig. 2(a)), the N equations may obtain a unique solution to the system of equations, i.e., a common intersection of the N circles, but in the case where the N circles, although intersecting, do not intersect at the same point (see fig. 2(b)), trilateration (least squares or maximum likelihood estimation) based on the intersection of lines may be used to find an approximate solution to the system of equations, thereby obtaining the first position. However, although trilateration can estimate position in non-ideal situations, it does not solve the situation in fig. 2(c) well because the distance reflected by the received signal strength is small and often unreliable.
In another implementation, when the strength of the at least one wireless signal is less than or equal to a preset threshold, step S20 includes: and obtaining a first position according to a pre-made radio map and the strength of the wireless signal.
Specifically, when the strength of at least one wireless signal is less than or equal to a preset threshold, the terminal device may perform position calculation by using a position fingerprinting method. The location fingerprinting method is to compare the strength of the wireless signal received by the terminal equipment with a radio map for positioning.
The fingerprint positioning method comprises two stages: an off-line training phase and an on-line positioning phase. The method comprises the following steps of manufacturing a radio map in an off-line training stage, selecting a group of preset positions as reference points, determining the number of the preset positions according to an area needing positioning, and dividing the positioning area into a plurality of n × n grid points, such as: 2 x 2 or 1 x 1, each grid point is a reference point, i.e. a preset position. For example, the positioning region is 10 × 1m2Can be every 2 x 1m2And (4) taking grid points, and taking 5 grid points in total, namely dividing the positioning area into 5 preset positions. Receiving the signal strength from all detected wireless signal transmitters and transmitter ID (hereinafter ID) at each reference point (since each wireless is referred to asThe information of the signal transmitter may be acquired many times, so that the intensity of the wireless signal corresponding to each ID is the average signal intensity), the intensity of the wireless signal received by each reference point is called a fingerprint, and the fingerprints of all the reference points form a radio map. And reading the manufactured radio map in the online positioning stage, and comparing the strength of the wireless signals received from the real time with the strength of the wireless signals in the radio map to obtain the first position. Further, the KNN algorithm (K nearest neighbor algorithm) is used to compare the received wireless signal strength with the radio map, the first n (e.g. n is 5, n may be the maximum number of all IDs received in the online phase, but n is usually 3-7 in consideration of the calculation amount) wireless signal transmitters with the strongest wireless signal strength in the online phase are selected, and the square of the difference between the signal strength of the n wireless signal transmitters stored in each reference point in the radio map and the signal strength of the n wireless signal transmitters received in the online phase is calculated, that is, the square corresponds to the distance from each reference point. And finally, selecting a proper k value (for example, k is 3, and the range of k is usually 1 to 5), finding the coordinates of the nearest k reference points according to the magnitude relation of the distance, and taking the centers of the coordinates of the k reference points to estimate the first position.
And a proper positioning method is selected by setting a threshold value, so that the positioning result can reach reasonable accuracy. If the strength of the received three wireless signals is greater than the threshold value, calculating the first position by using a trilateration method, otherwise, estimating the first position by using a position fingerprint method.
In one implementation, the wireless signal is a bluetooth signal, the wireless signal transmitter is a bluetooth beacon, and the strength of the wireless signal is:
wherein λ represents the distance between the bluetooth beacon and the terminal device, RSS (λ) represents the signal strength of the bluetooth beacon, λ0Denotes the reference distance, RSS (lambda)0) Indicating the reference distance lambda0Bluetooth beacon ofEta represents the path loss exponent, XσDenotes zero mean Gaussian variance as σ2The noise of (2).
S30, obtaining a second position according to the motion data and the first position; the accuracy of the second position is higher than the accuracy of the first position.
Specifically, the inertial navigation technology in step S10 is fused with the wireless positioning technology in step S20, so that the advantages of each technology can be exerted, and the disadvantages thereof can be suppressed.
Further, step S30 includes: obtaining a second position through a particle filter fusion model according to the motion data and the first position
In particular, the particle filtering is based on a sequential monte carlo framework that utilizes a set of weighted random particles to represent a posteriori densities of unknown locations in a dynamic state estimation framework. The particle filter fusion model adopts the first position of the terminal equipment estimated by a wireless positioning method as observation data, and models the motion of the equipment by using motion data obtained by a data-driven inertial navigation method. The specific modeling method is as follows:
(1) particle initialization: let the set of particles be H ═ { X ═ Xi1, 2., N }, where N is the number of particles. Each particle has a three-dimensional joint probability distribution (i.e., X)i＝(xi，yi，θi) Wherein (x)i，yi) Denotes the 2D position, θ, of the ith particleiIs the direction of the ith particle. When the time t is 0, the initial position is obtained by the method of step S20 using a gaussian distribution
Randomly selecting particles around an initial position and obtaining initial position information of the particles, an initial weight of each particle
Is 1/N.
(2) Particle motion model: starting from the initial time, the position of the particle is continuously changed along with the movement of the terminal device, and according to the position of the particle at the time t-1 (the last time of the time t) and the position distance change value and the course obtained at the time t (namely the position distance change value and the course change value in the time period from t-1 to t, the difference between t and t-1 is the single obtaining time of the inertial navigation data), the movement of the terminal device is modeled and expressed as:
wherein the content of the first and second substances,
the abscissa representing the ith particle at time t,
the ordinate of the ith particle at time t is shown,
the abscissa of the ith particle at time t-1 is shown,
denotes the ordinate, l, of the ith particle at the time immediately preceding ttIndicating the value of the change in the position distance, ψ, obtained at time ttIndicating the heading taken at time t. The position distance variation value and the heading at the time t obtained by the data-driven inertial navigation estimation accurately contain the motion information of the corresponding particles.
(3) Particle updating and resampling: when the positions of the terminal equipment at the time t-1 and the time t are different, the weights of all the particles need to be updated. And performing particle updating according to the proximity degree of the position of the particle at the time t-1 and the first position at the time t, wherein the position of the particle at the time t-1 is the position of the particle at the time t-1 acquired when the second position at the time t-1 is calculated. Particle resampling is then performed to retain the more weighted particles and to discard the less weighted particles.
(4) Target terminal device position estimation: in order to make the posterior probability expression smoother and to show the superiority of particle filtering, the weighted sum of the particles is used to estimate the second position of the terminal device, which is expressed as:
wherein, XtAbscissa, Y, representing terminal device at time ttThe ordinate of the terminal device at time t is indicated,
represents the weight of the ith particle at time t, and N represents the number of particles.
And (4) combining the formulas in the steps (2) and (4) to obtain a final particle filter fusion model as follows:
wherein, XtAbscissa, Y, representing terminal device at time ttThe ordinate of the terminal device at time t is indicated,
the abscissa of the ith particle at time t-1 is shown,
denotes the ordinate, l, of the ith particle at time t-1tIndicating the value of the change in the position distance, ψ, obtained at time ttIndicating the heading taken at time t,
indicating the ith particle at time tWeight, N denotes the number of particles.
In summary, the indoor positioning method based on multi-source data fusion provided by the invention comprises the following steps: obtaining inertial navigation data, and obtaining motion data according to the inertial navigation data, wherein the motion data is a position distance change value and a course change value; receiving a wireless signal, and obtaining a first position according to the strength of the wireless signal, wherein the wireless signal is generated by an indoor wireless signal transmitter; and obtaining a second position according to the motion data and the first position, wherein the accuracy of the second position is higher than that of the first position. Compared with the traditional inertial navigation method, the method has the advantages that the predicted displacement vector contains more accurate motion information under different terminal use modes, and the indoor positioning precision is effectively improved.
It is to be understood that the invention is not limited to the examples described above, but that modifications and variations may be effected thereto by those of ordinary skill in the art in light of the foregoing description, and that all such modifications and variations are intended to be within the scope of the invention as defined by the appended claims.
Claims (10)
1. An indoor positioning method based on multi-source data fusion is characterized by comprising the following steps:
obtaining inertial navigation data, and obtaining motion data according to the inertial navigation data, wherein the motion data is a position distance change value and a course change value;
receiving a wireless signal, and obtaining a first position according to the strength of the wireless signal, wherein the wireless signal is generated by an indoor wireless signal transmitter;
obtaining a second position according to the motion data and the first position; the accuracy of the second position is higher than the accuracy of the first position.
2. The indoor positioning method based on multi-source data fusion of claim 1, wherein the inertial navigation data comprises accelerometer data, gyroscope data and gravimeter data, and the obtaining motion data of the terminal device according to the inertial navigation data comprises:
mapping the accelerometer data and the gyroscope data to a coordinate system where the gravimeter data is located to obtain updated accelerometer data and updated gyroscope data;
inputting the updated accelerometer data and the updated gyroscope data into a trained deep neural network, and obtaining motion data according to the trained deep neural network.
3. The indoor positioning method based on multi-source data fusion of claim 2, wherein the trained deep neural network comprises two bidirectional long-short term memory layers, two dropout layers and one full connection layer; and a dropout layer is arranged behind each bidirectional long-short term memory layer.
4. The indoor positioning method based on multi-source data fusion of any one of claims 1 to 3, wherein the single acquisition time of the inertial navigation data is 0.5s to 2.5 s.
5. The indoor positioning method based on multi-source data fusion of claim 1, wherein the wireless signals are at least three different wireless signals, and when the intensities of the at least three different wireless signals are all greater than a preset threshold, the obtaining of the first position according to the intensities of the wireless signals comprises:
and obtaining a first position according to the positions of the at least three wireless signal transmitters and the wireless signal strengths of the at least three wireless signal transmitters.
6. The indoor positioning method based on multi-source data fusion of claim 1, wherein the wireless signal is at least one wireless signal, and when the intensity of the at least one wireless signal is less than or equal to a preset threshold, the obtaining of the first position according to the intensity of the wireless signal comprises:
and obtaining a first position according to a pre-made radio map and the strength of the wireless signal.
7. The indoor positioning method based on multi-source data fusion of claim 6, wherein the pre-fabricated radiomap is fabricated by a method comprising:
receiving the intensity of at least two wireless signals of a wireless signal transmitter at a preset position in a room;
and obtaining a radio map according to the strength of the at least two wireless signals and the preset position.
8. The indoor positioning method based on multi-source data fusion of claim 1, wherein the wireless signal is a bluetooth signal, the wireless signal transmitter is a bluetooth beacon, and the intensity of the wireless signal is:
wherein λ represents the distance between the bluetooth beacon and the terminal device, RSS (λ) represents the signal strength of the bluetooth beacon, λ0Denotes the reference distance, RSS (lambda)0) Indicating the reference distance lambda0Signal strength of the Bluetooth beacon, η represents the path loss exponent, XσDenotes zero mean Gaussian variance as σ2The noise of (2).
9. The indoor positioning method based on multi-source data fusion of claim 1, wherein the obtaining a second position according to the motion data and the first position comprises:
and obtaining a second position through a particle filter fusion model according to the motion data and the first position.
10. The indoor positioning method based on multi-source data fusion of claim 9, wherein the particle filter fusion model is:
wherein, XtAbscissa, Y, representing terminal device at time ttThe ordinate of the terminal device at time t is indicated,
the abscissa of the ith particle at time t-1 is shown,
denotes the ordinate, l, of the ith particle at time t-1tIndicating the change in the position distance, ψ, acquired at time ttIndicating the heading taken at time t,
representing the weight of the ith particle at the time t, wherein N represents the number of the particles;
and determining the weight of the ith particle at the time t according to the position of the ith particle at the time t-1 and the first position at the time t.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011434444.8A CN112729301A (en) | 2020-12-10 | 2020-12-10 | Indoor positioning method based on multi-source data fusion |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011434444.8A CN112729301A (en) | 2020-12-10 | 2020-12-10 | Indoor positioning method based on multi-source data fusion |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112729301A true CN112729301A (en) | 2021-04-30 |
Family
ID=75598713
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011434444.8A Pending CN112729301A (en) | 2020-12-10 | 2020-12-10 | Indoor positioning method based on multi-source data fusion |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112729301A (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113546394A (en) * | 2021-06-21 | 2021-10-26 | 国家体育总局体育科学研究所 | Vector velocity measuring device |
| CN114001736A (en) * | 2021-11-09 | 2022-02-01 | Oppo广东移动通信有限公司 | Positioning method, positioning device, storage medium and electronic equipment |
| WO2023071615A1 (en) * | 2021-10-26 | 2023-05-04 | 上海瑾盛通信科技有限公司 | Positioning method and apparatus, and terminal and storage medium |
| CN117308925A (en) * | 2023-11-29 | 2023-12-29 | 中国人民解放军国防科技大学 | Navigation method, device, equipment and medium for spectral map inertial navigation combination |
| WO2024082214A1 (en) * | 2022-10-20 | 2024-04-25 | Telefonaktiebolaget Lm Ericsson (Publ) | Improved target positioning by using multiple terminal devices |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104519571A (en) * | 2014-12-26 | 2015-04-15 | 北京工业大学 | Indoor positioning method based on RSS (Received Signal Strength) |
| CN106197406A (en) * | 2016-06-20 | 2016-12-07 | 天津大学 | A kind of based on inertial navigation with the fusion method of RSSI wireless location |
| KR101739390B1 (en) * | 2015-12-11 | 2017-05-24 | 국방과학연구소 | Method for improving the accuracy of self-alignment about the inertial navigation system through gravitational error compensation |
| CN107990900A (en) * | 2017-11-24 | 2018-05-04 | 江苏信息职业技术学院 | A kind of particle filter design methods of pedestrian's indoor positioning data |
| CN109409431A (en) * | 2018-10-29 | 2019-03-01 | 吉林大学 | Multisensor attitude data fusion method and system neural network based |
| CN109506653A (en) * | 2018-11-12 | 2019-03-22 | 上海理工大学 | Based on the indoor positioning pedestrian's method for improving particle filter under NLOS environment |
| CN109883416A (en) * | 2019-01-23 | 2019-06-14 | 中国科学院遥感与数字地球研究所 | A kind of localization method and device of the positioning of combination visible light communication and inertial navigation positioning |
| CN110207692A (en) * | 2019-05-13 | 2019-09-06 | 南京航空航天大学 | A kind of inertia pre-integration pedestrian navigation method of map auxiliary |
| CN110536245A (en) * | 2019-08-28 | 2019-12-03 | 杭州电子科技大学 | A kind of indoor wireless positioning method and system based on deep learning |
| US20200355503A1 (en) * | 2018-01-10 | 2020-11-12 | Oxford University Innovation Limited | Determining the location of a mobile device |
-
2020
- 2020-12-10 CN CN202011434444.8A patent/CN112729301A/en active Pending
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104519571A (en) * | 2014-12-26 | 2015-04-15 | 北京工业大学 | Indoor positioning method based on RSS (Received Signal Strength) |
| KR101739390B1 (en) * | 2015-12-11 | 2017-05-24 | 국방과학연구소 | Method for improving the accuracy of self-alignment about the inertial navigation system through gravitational error compensation |
| CN106197406A (en) * | 2016-06-20 | 2016-12-07 | 天津大学 | A kind of based on inertial navigation with the fusion method of RSSI wireless location |
| CN107990900A (en) * | 2017-11-24 | 2018-05-04 | 江苏信息职业技术学院 | A kind of particle filter design methods of pedestrian's indoor positioning data |
| US20200355503A1 (en) * | 2018-01-10 | 2020-11-12 | Oxford University Innovation Limited | Determining the location of a mobile device |
| CN109409431A (en) * | 2018-10-29 | 2019-03-01 | 吉林大学 | Multisensor attitude data fusion method and system neural network based |
| CN109506653A (en) * | 2018-11-12 | 2019-03-22 | 上海理工大学 | Based on the indoor positioning pedestrian's method for improving particle filter under NLOS environment |
| CN109883416A (en) * | 2019-01-23 | 2019-06-14 | 中国科学院遥感与数字地球研究所 | A kind of localization method and device of the positioning of combination visible light communication and inertial navigation positioning |
| CN110207692A (en) * | 2019-05-13 | 2019-09-06 | 南京航空航天大学 | A kind of inertia pre-integration pedestrian navigation method of map auxiliary |
| CN110536245A (en) * | 2019-08-28 | 2019-12-03 | 杭州电子科技大学 | A kind of indoor wireless positioning method and system based on deep learning |
Non-Patent Citations (3)
| Title |
|---|
| 丁其宽: "基于位置指纹的无线室内定位技术研究", 中国优秀硕士学位论文全文数据库信息科技辑, no. 12, pages 7 - 16 * |
| 张兴凡: "重力模型误差对惯性导航系统精度的影", 光学与光电技术, vol. 16, no. 4, pages 86 - 91 * |
| 曾庆化 等: "重力辅助和模拟零速修正的手机航向修正方法", 中国惯性技术学报, vol. 26, no. 3, pages 289 - 294 * |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113546394A (en) * | 2021-06-21 | 2021-10-26 | 国家体育总局体育科学研究所 | Vector velocity measuring device |
| WO2023071615A1 (en) * | 2021-10-26 | 2023-05-04 | 上海瑾盛通信科技有限公司 | Positioning method and apparatus, and terminal and storage medium |
| CN114001736A (en) * | 2021-11-09 | 2022-02-01 | Oppo广东移动通信有限公司 | Positioning method, positioning device, storage medium and electronic equipment |
| WO2023082797A1 (en) * | 2021-11-09 | 2023-05-19 | Oppo广东移动通信有限公司 | Positioning method, positioning apparatus, storage medium, and electronic device |
| WO2024082214A1 (en) * | 2022-10-20 | 2024-04-25 | Telefonaktiebolaget Lm Ericsson (Publ) | Improved target positioning by using multiple terminal devices |
| CN117308925A (en) * | 2023-11-29 | 2023-12-29 | 中国人民解放军国防科技大学 | Navigation method, device, equipment and medium for spectral map inertial navigation combination |
| CN117308925B (en) * | 2023-11-29 | 2024-02-09 | 中国人民解放军国防科技大学 | Navigation method, device, equipment and medium for spectral map inertial navigation combination |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110118549B (en) | Multi-source information fusion positioning method and device | |
| CN110856112B (en) | Crowd-sourcing perception multi-source information fusion indoor positioning method and system | |
| CN112729301A (en) | Indoor positioning method based on multi-source data fusion | |
| CN105588566B (en) | A kind of indoor locating system merged based on bluetooth with MEMS and method | |
| CN105424030B (en) | Fusion navigation device and method based on wireless fingerprint and MEMS sensor | |
| Zhao et al. | Crowdsourcing and multisource fusion-based fingerprint sensing in smartphone localization | |
| Pei et al. | Optimal heading estimation based multidimensional particle filter for pedestrian indoor positioning | |
| Liu et al. | Fusing similarity-based sequence and dead reckoning for indoor positioning without training | |
| CN109413578B (en) | Indoor positioning method based on fusion of WIFI and PDR | |
| CN103561462A (en) | Indoor positioning system and method totally based on smart mobile terminal platform | |
| US10557711B2 (en) | Apparatus for inferring pedestrian position based on pedestrian movement detection, and method therefor | |
| CN107014375B (en) | Indoor positioning system and method with ultra-low deployment | |
| EP2951533A1 (en) | Method and system for varying step length estimation using nonlinear system identification | |
| CN106840163A (en) | A kind of indoor orientation method and system | |
| CN111829516B (en) | Autonomous pedestrian positioning method based on smart phone | |
| You et al. | Hybrid indoor positioning system for pedestrians with swinging arms based on smartphone IMU and RSSI of BLE | |
| Yu et al. | Precise 3D indoor localization and trajectory optimization based on sparse Wi-Fi FTM anchors and built-in sensors | |
| Yu et al. | Autonomous 3D indoor localization based on crowdsourced Wi-Fi fingerprinting and MEMS sensors | |
| CN109164411A (en) | A kind of personnel positioning method based on multi-data fusion | |
| CN113566820B (en) | Fused pedestrian positioning method based on position fingerprint and PDR algorithm | |
| Sun et al. | Indoor localization using mind evolutionary algorithm-based geomagnetic positioning and smartphone IMU sensors | |
| KR101264306B1 (en) | Apparatus of tracking user indoor using user motion model learning and recording media therefor | |
| CN106028450B (en) | Indoor positioning method and equipment | |
| US11029415B2 (en) | Systems and methods for estimating initial heading at start-up of navigation | |
| CN106197418B (en) | A kind of indoor orientation method merged based on the fingerprint technique of sliding window with 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 |