CN112333818A - Multi-source fusion indoor positioning system and method based on self-adaptive periodic particle filtering - Google Patents

Multi-source fusion indoor positioning system and method based on self-adaptive periodic particle filtering Download PDF

Info

Publication number
CN112333818A
CN112333818A CN202011162672.4A CN202011162672A CN112333818A CN 112333818 A CN112333818 A CN 112333818A CN 202011162672 A CN202011162672 A CN 202011162672A CN 112333818 A CN112333818 A CN 112333818A
Authority
CN
China
Prior art keywords
positioning
rssi
mobile terminal
fusion
particle
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
Application number
CN202011162672.4A
Other languages
Chinese (zh)
Other versions
CN112333818B (en
Inventor
朱容波
王俊
丁千傲
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
South Central Minzu University
Original Assignee
South Central University for Nationalities
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by South Central University for Nationalities filed Critical South Central University for Nationalities
Priority to CN202011162672.4A priority Critical patent/CN112333818B/en
Publication of CN112333818A publication Critical patent/CN112333818A/en
Application granted granted Critical
Publication of CN112333818B publication Critical patent/CN112333818B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • 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
    • 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
    • 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
    • H04W4/00—Services specially adapted for wireless communication networks; Facilities therefor
    • H04W4/80—Services using short range communication, e.g. near-field communication [NFC], radio-frequency identification [RFID] or low energy communication

Landscapes

  • Engineering & Computer Science (AREA)
  • Radar, Positioning & Navigation (AREA)
  • Remote Sensing (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Signal Processing (AREA)
  • Automation & Control Theory (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Position Fixing By Use Of Radio Waves (AREA)

Abstract

The invention discloses a multisource fusion indoor positioning system and method based on self-adaptive periodic particle filtering, and the method comprises the following steps: step 1: the method comprises the steps that a Bluetooth beacon is laid in an indoor environment, broadcasts from more than three Bluetooth beacon beacons are received through a mobile terminal, and the received Bluetooth beacon ID and the corresponding signal value strength RSSI are uploaded to a server side; step 2: acquiring the numerical value of a built-in sensor of the mobile terminal, determining the stepping direction, the walking direction and the step length, and locally positioning the pedestrian dead reckoning PDR at the mobile terminal; and step 3: converting the received signal value strength RSSI into a distance through a signal strength attenuation model so as to carry out trilateral positioning; and fusing the RSSI positioning result and the PDR positioning result, constructing a self-adaptive periodic particle filter multi-source fusion model, obtaining a final positioning result, and sending the final positioning result to the mobile terminal. The positioning result after the self-adaptive periodic particle filtering fusion is higher in precision, and compared with the traditional particle filtering, the fusion time delay is reduced.

Description

Multi-source fusion indoor positioning system and method based on self-adaptive periodic particle filtering
Technical Field
The invention relates to the field of application of Internet of things, in particular to a multisource fusion indoor positioning system and method based on self-adaptive periodic particle filtering.
Background
Since the conventional positioning signal source GPS cannot be used in an indoor environment, indoor positioning needs to seek other positioning sources. At present, signal sources which can be used for indoor positioning include WiFi, Bluetooth, infrared rays, ultra-wideband, ultrasonic waves, RFID and the like. The complicated and variable indoor environment can cause various problems of non-line-of-sight propagation of positioning signals, multipath effect, shadow fading and the like. Therefore, there are many difficulties in implementing high-precision indoor positioning, and there are problems of environmental limitation and low precision in indoor positioning by a single positioning source. By the multi-source fusion method, positioning information can be enriched, and high-precision and high-reliability indoor positioning results can be provided. Meanwhile, in practical application, the positioning information often requires high real-time performance, especially in a real-time navigation scenario. This makes the positioning delay a key indicator of the performance of another indoor positioning system beyond accuracy.
Generally, the positioning accuracy and the time delay are two mutually restricted indexes. Although the positioning precision is improved by multi-source fusion, the acquisition of the information of the plurality of positioning sources and the processing and fusion of the acquired data can increase the positioning time delay undoubtedly. The particle filter fusion is taken as an example, the probability distribution under the actual scene is restored by a large amount of particle samples, and the non-Gaussian error distribution and the nonlinear system state model can be dealt with, so that the positioning precision can be ensured under the complex indoor environment. However, the positioning time delay of the system is increased by a large amount of particle sampling, so that the indoor positioning system based on the traditional particle filter fusion is difficult to meet the real-time positioning requirement.
At present, the research of indoor positioning algorithm based on multi-source fusion has made a significant breakthrough in the aspect of positioning accuracy. However, many high-precision indoor positioning algorithms are at the cost of time delay, which makes these indoor positioning algorithms difficult to be used in indoor positioning scenes with high real-time requirements. How to reduce the time delay of the multi-source fusion indoor positioning algorithm as much as possible on the premise of ensuring the precision is a problem worthy of exploration. Therefore, it is very important to research an improved particle filter multi-source fusion indoor positioning method with optimized tradeoff between precision and time delay.
Disclosure of Invention
The invention aims to solve the technical problem of providing a multisource fusion indoor positioning system and method based on self-adaptive periodic particle filtering aiming at the defects in the prior art.
The technical scheme adopted by the invention for solving the technical problems is as follows:
the invention provides a multisource fusion indoor positioning system based on self-adaptive periodic particle filtering, which comprises: the system comprises a data source, a mobile terminal and a server side; wherein:
the data source comprises a Bluetooth beacon and a built-in sensor of the mobile terminal; planning and laying a Bluetooth beacon in an indoor environment, receiving broadcasts from more than three Bluetooth beacons through a mobile terminal, and uploading the received Bluetooth beacon ID and the corresponding signal value strength RSSI to a server side;
the mobile terminal determines stepping, walking directions and step lengths by acquiring the numerical value of a sensor built in the mobile terminal, and performs Pedestrian Dead Reckoning (PDR) positioning locally on the mobile terminal;
the server end is used for converting the received signal value strength RSSI into a distance through a signal strength attenuation model so as to carry out trilateral positioning; and fusing the RSSI positioning result and the PDR positioning result, constructing a self-adaptive periodic particle filter multi-source fusion model, obtaining a final positioning result, and sending the final positioning result to the mobile terminal.
Further, the built-in sensor of the mobile terminal of the present invention includes: a gyroscope, an accelerometer and a geomagnetic sensor.
The invention provides a multisource fusion indoor positioning method based on self-adaptive periodic particle filtering, which comprises the following steps:
step 1: constructing a Bluetooth RSSI trilateral positioning model:
planning and laying a Bluetooth beacon in an indoor environment, receiving broadcasts from more than three Bluetooth beacons through a mobile terminal, and uploading the received Bluetooth beacon ID and the corresponding signal value strength RSSI to a server side;
step 2: constructing a PDR positioning model of a built-in sensor of the mobile terminal:
acquiring the numerical value of a built-in sensor of the mobile terminal, determining the stepping direction, the walking direction and the step length, and locally positioning the pedestrian dead reckoning PDR at the mobile terminal;
and step 3: constructing a self-adaptive periodic particle filter multi-source fusion model:
converting the received signal value strength RSSI into a distance through a signal strength attenuation model so as to carry out trilateral positioning; and fusing the RSSI positioning result and the PDR positioning result, constructing a self-adaptive periodic particle filter multi-source fusion model, obtaining a final positioning result, and sending the final positioning result to the mobile terminal.
Further, the specific method of trilateral positioning of the present invention is:
during trilateral positioning calculation, solving and reducing a final positioning error through a multi-AP simultaneous equation, and calculating a distance through a space loss model; the spatial loss model is:
Figure BDA0002744747830000031
wherein d represents the distance between the beacon and the receiving point; d0Represents a reference distance; PL (d) represents the signal strength RSSI at distance d; a represents a path loss exponent; ω represents a mean of 0 and a variance of
Figure BDA0002744747830000032
(ii) a gaussian distribution variable; let diAs the distance between the beacon i and the receiving point, it can be known that:
Figure BDA0002744747830000033
further, the specific formula of the PDR location of the present invention is:
Figure BDA0002744747830000034
in the formula xk、ykThe abscissa and ordinate of step k, lkRepresenting the step size of step k, thetakRepresents the angle between the advancing direction of the kth step and the x-axis, deltax、δyRepresenting errors in the x and y directions.
Further, the adaptive periodic particle filter multi-source fusion model of the present invention includes a method for performing adaptive condition judgment:
constructing an adaptive periodic particle filter, and after the filter receives a PDR positioning result and an RSSI positioning result, adaptively judging whether particle fusion is needed:
in the case of one-dimensional situation, the conditions for adaptive judgment are as follows:
assuming that the PDR positioning error is accumulated to be a negative number, namely the PDR positioning value is gradually lower than the true value along with the advance of the person; setting the RSSI positioning result to be normal fluctuation with the true value as expected and R as variance; if the RSSI positioning fluctuation at the time t is higher than the final positioning value at the time t-1 by a certain amount d, judging that the PDR positioning error is accumulated to a certain degree, and performing particle filter fusion;
in the case of two-dimensional situation, the conditions for adaptive judgment are as follows:
(RSSI.x(t)-final.x(t-1)＞d)&(RSSI.y(t)-final.y(t-1)＞d)
wherein RSSI.x (t) and RSSI.y (t) represent x and y coordinates of the RSSI positioning result at the time t; final.x (t-1) and final.y (t-1) represent the x and y coordinates of the final positioning value at the time t-1; the value of d is determined according to the specific fluctuation condition of the RSSI, and the filtering frequency is controlled by controlling the size of d, so that the final precision of the model is improved by tuning;
if the self-adaptive judgment condition is met, performing particle filter fusion; if the self-adaptive judgment condition is not met, the PDR positioning result is directly used as a final positioning result.
Further, the specific method for multi-source fusion of the adaptive periodic particle filter of the invention comprises the following steps:
step 1: initialization: when the filtering starts, firstly, initializing, assigning the coordinates and weights of n particles with a predetermined number, setting the initial coordinates of the current step as the final coordinates of the previous step and the weights of the particles as 1/n, and entering a prediction step after initialization;
step 2: a prediction step: let random process satisfy Xk＝f(Xk-1)+QkObserved to satisfy Yk＝h(Xk)+RkBayesian filtering requires knowledge of the initial value x0Probability density function f0 +(ii) a The prediction step is infinite integration:
Figure BDA0002744747830000041
let f (x) be the probability density function of x, the theorem of maxima states that n → ∞ time:
Figure BDA0002744747830000042
the concept of weight is introduced:
Figure BDA0002744747830000043
initial weight of each particle is
Figure BDA0002744747830000044
For different xiSetting up different wi；
The position and weight of the particle completely determine the probability distribution function, and also determine the probability density function; if xiIs sampled from f (x), then xiSatisfies the rule of probability distribution; let x be0Obeying normal distribution, the collected sample is x0 (1),x0 (2),…,x0 (n)(ii) a Setting:
Figure BDA0002744747830000045
then:
Figure BDA0002744747830000051
f1 +(x)＝ηfR[y-h(x)]f1 -(x)
if the distribution is normal distribution, directly sampling; and (3) generating particles:
Figure BDA0002744747830000052
for each fQ[x-f(x0 (i))]Consider an inevitable event X ═ f (X)0 (i)) Superposition with a random number Y-N (0, Q); then f1 -(x) Particle X1 -(1),X1 -(2),…,X1 -(n)Comprises the following steps:
X1 -(i)＝f(x0 (i))+v
wherein v is a random number of N (0, Q); at this time x1 -(i)＝f(x0 (i)) The nature of + Q is that the position of the particles is changed and the weight is not changed. Here, the Q of the prediction step is different from that of the conventional particle filter, and the value thereof should be large. For conventional particle filtering in general experiments, the Q value should be as consistent as possible with the variance of the prediction data. In the experiment, due to the existence of PDR positioning error accumulation, even in the conventional particle filtering, the accuracy can be improved a little after the prediction step variance Q is slightly improved, that is, the weight of PDR positioning in fusion is reduced. The self-adaptive periodic particle filtering is fused after multiple steps, and the process is accumulation of normal distribution; and during this period, more obvious error accumulation occursTherefore, the value of Q should be increased appropriately. Let q be the normal fluctuation variance of the original prediction data, c be some small constant (related to the error accumulation speed), and fn be the number of times that the current fusion condition fails to be judged. The theoretical value of Q is:
Q＝(q+c)*(1+fn)
where (q + c) represents the normal variance of a single step and (1+ fn) represents the cumulative number. The actual optimal value is about the value, but the specific value needs to be obtained by parameter optimization through experiments; ending the predicting step;
step 3: and (5) updating.
Further, the specific method of the updating step of the present invention is:
the updating step observes data y1The method comprises the following steps:
Figure BDA0002744747830000053
using the properties of the dirac function yields:
Figure BDA0002744747830000054
setting:
w1 (i)＝fR[y1-h(x1 (i)-)]w0 (i)
at this time, the weight changes, then:
Figure BDA0002744747830000061
the update step does not change the particle position, but changes the weight of the particle;
carrying out weight normalization:
Figure BDA0002744747830000062
because of the single-step particle filtering, the resampling step is omitted. Completing a round of particle filtering, and then returning the result to the mobile terminal as a final positioning result of the current step; and when the filter receives the PDR positioning result and the RSSI positioning result of the next step, starting the next judgment cycle.
The invention has the following beneficial effects: the invention discloses a multisource fusion indoor positioning system and a multisource fusion indoor positioning method based on self-adaptive periodic particle filtering, which comprises the following steps:
(1) and selecting a multi-source positioning source. The indoor environment is complicated, and the indoor positioning source is various. The selection of what positioning source is directly related to the final performance of the system. And comprehensively considering the problems of system precision, hardware overhead, terminal energy consumption, equipment support and the like, and determining to select the Bluetooth Beacon Beacon and the built-in sensor of the mobile terminal as the positioning source of the system. The two are used as indoor multi-source fusion signal sources, and have excellent complementarity in the aspects of error accumulation and interference resistance. And both have mature positioning methods, which is convenient for system realization.
(2) And improving a particle filter fusion algorithm. The particle filter is excellent in problems of arbitrary error distribution and an arbitrary state space model, and highly fits a complex indoor environment. It requires the restoration of the probability distribution through a large number of particle samples, which causes system delay. The particle filter fusion algorithm is improved, the system precision is guaranteed, and meanwhile the self-adaptive filtering strategy capable of reasonably and effectively reducing the particle filter calculation amount is selected, so that the real-time performance of the positioning result is guaranteed.
Simulation experiment results show that compared with the single-source positioning of the two positioning sources, the positioning result fused by the self-adaptive periodic particle filter has higher precision. Compared with the traditional particle filtering, the self-adaptive periodic particle filtering provided by the invention reduces 58% of fusion time delay by reasonably reducing the filtering times on the premise of almost flat precision.
Drawings
The invention will be further described with reference to the accompanying drawings and examples, in which:
FIG. 1 is a system model diagram of an embodiment of the invention;
FIG. 2 is a system flow diagram of an embodiment of the present invention;
FIG. 3 is a schematic diagram of RSSI trilateration according to an embodiment of the present invention;
FIG. 4 illustrates an RSSI trilateration process according to an embodiment of the present invention;
FIG. 5 is a schematic illustration of a PDR location according to an embodiment of the present invention;
FIG. 6 is a PDR location procedure according to an embodiment of the present invention;
FIG. 7 is a flow chart of an adaptive periodic particle filter according to an embodiment of the present invention;
FIG. 8 is a schematic two-dimensional plan positioning of an embodiment of the present invention;
FIG. 9 is a graph of the 3D effect of the analog signal of the embodiment of the present invention;
FIG. 10 is an analog signal coordinate surface projection view of an embodiment of the present invention;
FIG. 11 is a particle count and error value for an embodiment of the present invention;
FIG. 12 is a particle count versus model time for an embodiment of the present invention;
FIG. 13 is a graph of predicted step variance versus resultant error for an embodiment of the present invention;
FIG. 14 is a diagram of a single step effect of an adaptive periodic particle filter according to an embodiment of the present invention;
FIG. 15 is a single step error variation graph of an embodiment of the present invention;
FIG. 16 is a diagram illustrating the effect of conventional particle filtering according to an embodiment of the present invention;
FIG. 17 is a graph illustrating the effect of adaptive periodic particle filtering according to an embodiment of the present invention;
FIG. 18 is a comparison of MAEs of the present invention;
FIG. 19 is a comparison graph of RMSE for an embodiment of the present invention;
FIG. 20 is a comparison graph of model elapsed time for an embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
As shown in fig. 1, the multi-source fusion indoor positioning method based on adaptive periodic particle filtering according to the embodiment of the present invention includes the following steps:
1) construction of Bluetooth RSSI trilateral positioning model
Bluetooth beacon beacons are planned and laid down in an indoor environment. The mobile phone receives the broadcast from more than three beacon beacons, and uploads the received beacon ID and the corresponding signal value strength RSSI to the server side. And the server end converts the received signal value intensity into a distance through a signal intensity attenuation model so as to carry out trilateral positioning. The positioning method has the advantages of low energy consumption, low time delay and higher positioning precision. But is obviously interfered by the outside.
2) PDR positioning model for constructing built-in sensor of mobile phone
The method comprises the steps of determining stepping direction, walking direction and step length by obtaining the numerical value of a built-in sensor (accelerometer, gyroscope and the like) of the mobile phone, and carrying out pedestrian dead reckoning PDR positioning locally at a mobile terminal. The positioning method has the advantages of low energy consumption, low time delay, high positioning precision in short distance and no interference from external environment. However, because each step is to calculate the relative displacement, the long-distance positioning has the problem of error accumulation.
3) Construction of adaptive periodic particle filter multi-source fusion model
By fusing RSSI positioning results, error accumulation of PDR positioning can be eliminated; the anti-interference performance of the PDR positioning also well complements the disadvantage of the RSSI positioning. However, in the conventional particle filter, a large number of particle samples are required to fit the probability density, and a large number of particle weights and resampling are updated, so that the huge calculation amount contained in the steps will generate delay, and the real-time performance of the positioning system is limited. Therefore, an adaptive periodic particle filter algorithm is proposed. The short-distance positioning accuracy of the PDR positioning is relied on, the filtering fusion probability is reduced before obvious error accumulation is generated, and the calculated amount is reduced by reducing the filtering times, so that the time delay is reduced. Along with the gradual accumulation of errors, the probability of fusing the errors with the RSSI positioning results is improved, and only a specific RSSI positioning result is selected for fusing, so that the positioning accuracy is ensured. Therefore, the fusion time delay can be reduced on the premise of ensuring the precision.
Because the indoor environment is closed, complex and variable, the indoor positioning has many problems of non-line-of-sight propagation of positioning signals, multipath effect, shadow fading and the like. When various positioning signal sources are used independently in an indoor environment, certain limitation exists, and the requirement for daily use is difficult to meet. The invention constructs the self-adaptive periodic particle filter, and completes indoor high-precision positioning by fusing mobile terminal PDR positioning and indoor Bluetooth RSSI positioning. Compared with the traditional general particle filter, the particle filter is combined with the characteristics of two positioning signal sources in an indoor environment, and the particle filter fusion is carried out in a self-adaptive periodic mode, so that the time delay generated by a large amount of operations of the particle filter is reduced on the premise of ensuring the positioning accuracy.
The system model is shown in fig. 1. The system model in fig. 1 is composed of a data source, a mobile terminal and a server. The data source consists of a Bluetooth beacon and a mobile terminal built-in sensor. At time t, the plurality of bluetooth beacons inform the mobile terminal of its own ID in the form of broadcast. The mobile terminal binds the ID and the signal value intensity RSS of the broadcast received at the time t to be ID, RSSt. The mobile terminal combines the final positioning result (X, Y) at the moment t-1t-1Acquiring data through a built-in sensor to carry out PDR positioning to obtain a PDR positioning result (x) at the time t1,y1)t. Then { ID, RSS } will betAnd (x)1,y1)tAnd the same is uploaded to the server side. The server combines the map data to carry out RSSI positioning and positions the result (x) of the RSSI positioning2,y2)tAnd PDR location result (x)1,y1)tPerforming particle filter fusion to obtain final positioning result (X, Y) at time ttAnd then returns the result to the mobile terminal. The overall system flow chart is shown in fig. 2.
The formula parameter symbol table in the embodiment of the invention is shown in table 1.
TABLE 1 symbol table
Figure BDA0002744747830000091
Figure BDA0002744747830000101
1) Selection of multiple source positioning sources
There are many signal sources that can be used for indoor positioning, and WiFi, bluetooth, infrared ray, RFID, ultrasonic wave, super bandwidth, built-in sensor of mobile device, etc. are all common signal sources for indoor positioning systems. However, although the infrared, ultrasonic and ultra-wideband positioning systems have high precision, the laying of system equipment and the high price of hardware equipment in actual environments can hinder the practical use and popularization of indoor positioning systems. Therefore, considering practical application scenarios and system cost problems, the most suitable method is not limited to WiFi, Bluetooth and built-in sensors of mobile devices. In the existing scene, the WiFi is laid in advance, and although the positioning facility is seemingly not paid additionally, a plurality of problems are caused. Since its laying is intended to facilitate network communication rather than indoor positioning, there are situations where its actual laying site is different from the indoor positioning needs. If the WiFi node is additionally arranged, the original communication function is more likely to be influenced, and the method is inverted at the end. And if the WiFi is laid under non-integral planning, differences in various aspects such as types and signal strengths of the signal point routers are caused, and unified management is difficult. If WiFi fingerprint positioning is used, continuous consumption of manpower and material resources for construction and later maintenance of a fingerprint database can also exist.
In contrast, although the dedicated beacon laying generates cost on hardware equipment, the bluetooth positioning system has low cost, and the positioning effect is much better than that of WiFi positioning which is not planned to lay in advance. Furthermore, the beacon-based bluetooth positioning system is also low-power-consuming for mobile devices. And the beacon can be used for about three years due to low energy consumption and power supply through a built-in battery, so that the beacon is very convenient to lay in the aspect of beacon without external power supply. The Bluetooth positioning system has the advantages of low price, low energy consumption, convenience in laying, good positioning effect and good mobile terminal support, so that the Bluetooth positioning system based on beacon is selected as one of signal sources. And the trilateral positioning is selected in consideration of the fact that the construction and later maintenance of the fingerprint database are too time-consuming and labor-consuming, and the positioning effect is unlikely to be superior to that of trilateral positioning. Due to the existence of distance calculation errors, three situations shown in fig. 3 can occur in trilateration, and the final positioning errors can be reduced through solution of multiple AP simultaneous equations. Fig. 4 bit RSSI trilateration procedure. The distance can be calculated through a space loss model. The spatial loss model is:
Figure BDA0002744747830000111
wherein d represents the distance between the beacon and the receiving point; d0Represents a reference distance, often taken to be 1 m; PL (d) represents the signal strength RSSI at distance d; a represents the path loss exponent, often depending on the particular environment; ω represents a mean of 0 and a variance of
Figure BDA0002744747830000112
Is measured. Let diAs the distance between the beacon i and the receiving point, it can be known that:
Figure BDA0002744747830000113
the built-in sensor of the mobile phone is carried by the mobile terminal, so that extra hardware cost is not needed for a positioning system, and the problem of obtaining the sensor is not needed to be worried like an RFID label because the sensor is owned by a common user. And the PDR is also receiving wide attention from researchers as a research hotspot of indoor positioning and is gradually mature. The PDR positioning algorithm has the short-distance high-precision positioning characteristic, is excellent in short-distance positioning, but cannot be independently used for long-distance precise positioning due to the fact that the positioning process is iterative and the problem of error accumulation cannot be avoided. If the PDR positioning algorithm is to be used, consideration is needed to eliminate the problem of error accumulation. But one of the positioning sources is to have the excellent characteristics of no environmental interference, no extra cost, no acquisition difficulty and low power consumption.
Figure BDA0002744747830000114
In the formula xk、ykThe abscissa and ordinate of step k, lkRepresenting the step size of step k, thetakRepresents the angle between the advancing direction of the kth step and the x-axis, deltax、δyRepresenting errors in the x and y directions.
In combination with the consideration, the invention selects the beacon-based Bluetooth positioning system and the mobile phone built-in sensor-based PDR positioning system as the signal source of the multisource fusion indoor positioning system.
2) Adaptive periodic particle filtering algorithm
In an indoor environment, the problems of insufficient precision, accumulated errors and the like exist in single-signal-source positioning. Therefore, information from different positioning sources needs to be integrated, and the aim is to achieve positioning information which better meets the real situation, namely multi-source fusion positioning. Multi-source fusion is not a simple weighted superposition of information but should follow certain theoretical rules. Bayes filtering is a filtering theory basis based on probability theory, and particle filtering is a specific implementation mode of Bayes filtering theory. The probability density is approximated by a large number of particles, which is excellent in a nonlinear non-gaussian system. The process is essentially that the posterior probability is solved by adding the prior knowledge and the observation update. The method is suitable for fusing the inaccurate data of the complex number. The method has the advantages that a more accurate filtering value can be obtained after fusion, and the variance of the filtering value is smaller than that of fused data. The advantages described above are well suited to indoor positioning environments. Therefore, the invention selects particle filtering as a multi-positioning source fusion algorithm.
On the premise of ensuring the positioning accuracy, how to reduce the calculation amount of the particle filter as much as possible so as to ensure the real-time performance of the system is a main problem of the research of the invention. Considering that the PDR positioning is accurate in a short distance, the invention considers that particle filter fusion is not needed when the PDR positioning is still accurate, and only the PDR positioning result is taken as a final positioning result. And when the error is accumulated to a certain degree, filtering and fusing a specific RSSI positioning value to eliminate the error accumulation. The continuous filtering process is split into discrete, theoretically resampling steps for particle filtering can be eliminated. Therefore, by reducing the filtering times, the system calculation amount can be effectively reduced, and the time delay is reduced. The algorithmic process is described in detail below.
After the filter obtains the PDR positioning result and the RSSI positioning result of the current step, what is needed is to adaptively determine whether particle filtering is needed at present. I.e. the filter needs an adaptive decision criterion. By combining the fluctuation of RSSI positioning, the self-adaptive judgment condition can be set to judge that particle filter fusion is needed when the positioning distance between the two is greater than a certain value, otherwise, the PDR positioning result is directly adopted. The present invention has been studied under these conditions as follows.
In the one-dimensional case, the PDR positioning error is assumed to accumulate as a negative number, i.e., as the person progresses, the PDR positioning value is gradually lower than the true value. Let the RSSI positioning result be a normal fluctuation with the true value as expected and R as variance. When the RSSI positioning fluctuation at the time t is higher than the final positioning value at the time t-1 by a certain amount d, the positioning error can be judged to be accumulated to a certain degree by the PDR positioning error, and then particle filter fusion can be carried out. The probability of meeting the fusion condition is determined by the variance R of the RSSI positioning fluctuation. The filtering frequency is determined by controlling the size of d. The method is feasible through test verification. In the two-dimensional case, however, relying solely on distance values as a basis for determining whether to filter is not reliable.
As shown in fig. 8, in the two-dimensional plane, it is assumed that the pedestrian advances in the direction as indicated by the dotted line. In the figure, blue squares represent real values at time t, red diamonds represent RSSI positioning results, and yellow circles represent PDR positioning results. Due to the existence of error accumulation, the positioning result of the PDR gradually lags behind the true value, but the advancing direction is unchanged. The RSSI positioning value can randomly appear in a dotted line circle form with the real value as the center of the circle and the fluctuation variance as the radius as the graph, and the probability is higher when the RSSI positioning value is closer to the center of the circle. Then, when the RSSI positioning value is in the red area shown on the right side of fig. 8, the fusion is performed, so that the best fusion effect can be obtained on the premise that the original direction is not changed basically. The adaptive conditions at this time are:
(RSSI.x(t)-final.x(t-1)＞d)&(RSSI.y(t)-final.y(t-1)＞d) (4)
wherein RSSI.x (t) and RSSI.y (t) represent x and y coordinates of the RSSI positioning result at the time t; final.x (t-1), final.y (t-1) represent the x, y coordinates of the final positioning value at time t-1. The value of d should be determined according to the specific fluctuation of RSSI. The filtering frequency can be controlled by controlling the size of d, so that the final precision of the model is improved by tuning.
And (4) directly using the PDR positioning result when the PDR positioning result does not meet the judgment condition. And when the conditions are met, performing particle filtering. At the start of filtering, a simple initialization operation is first performed. And assigning the coordinates and the weights of the n particles with the predetermined number. The initial coordinate of the current step should be the final coordinate of the previous step, and the weight of the particles should be set to 1/n. The prediction step is entered after a simple initialization.
Let random process satisfy Xk＝f(Xk-1)+Qk. Observation satisfies Yk＝h(Xk)+Rk. Bayesian filtering requires knowledge of the initial value x0Probability density function f0 +. The predicted step is an infinite integral shown by equation (6):
Figure BDA0002744747830000131
let f (x) be the probability density function of x, the theorem of maxima states that n → ∞ time:
Figure BDA0002744747830000132
the law of large numbers implies that the probability density can be approximated by a stack of particles. This is the particle filtering. Particle filtering requires a large number of particles to approximate the probability density function. For the cumulative distribution function, more particles are needed where the climb is faster for a better fit curve. In this case, the concept of weight is introduced to make part of the particles have higher weight. This reduces the number of particles required.
Figure BDA0002744747830000133
Corresponding to the initial weight of each particle being
Figure BDA0002744747830000134
And here for different xiSetting up different wi。
Therefore, the particle weight is higher at the position where the curve rapidly climbs, and the fitting can be completed by fewer particles. The position and weight of the particles completely determine the probability distribution function and thus the probability density function. If xiIs sampled from f (x), then xiNaturally satisfies the law of probability distribution. Let x be0Obeying normal distribution, the collected sample is x0 (1),x0 (2),…,x0 (n). Is provided with
Figure BDA0002744747830000141
Then
Figure BDA0002744747830000142
f1 +(x)＝ηfR[y-h(x)]f1 -(x) (10)
In the case of a normal distribution, it can be sampled directly. And (3) generating particles:
Figure BDA0002744747830000143
for each fQ[x-f(x0 (i))]Can be regarded as an inevitable event X ═ f (X)0 (i)) And a random number Y-N (0, Q). Then f1 -(x) Particle X1 -(1),X1 -(2),…,X1 -(n)Comprises the following steps:
X1 -(i)＝f(x0 (i))+v (12)
wherein v is a random number from N (0, Q). At this time x1 -(i)＝f(x0 (i)) The nature of + Q is that the position of the particles is changed and the weight is not changed. Here, the Q of the prediction step is different from that of the conventional particle filter, and the value thereof should be large. For conventional particle filtering in general experiments, the Q value should be as consistent as possible with the variance of the prediction data. In the experiment, due to the existence of PDR positioning error accumulation, even in the conventional particle filtering, the accuracy can be improved a little after the prediction step variance Q is slightly improved, that is, the weight of PDR positioning in fusion is reduced. The self-adaptive periodic particle filtering is fused after multiple steps, and the process is accumulation of normal distribution; and more obvious error accumulation occurs in the period, so the value of Q should be increased by a proper amount. Let q be the normal fluctuation variance of the original prediction data, c be some small constant (related to the error accumulation speed), and fn be the number of times that the current fusion condition fails to be judged. The theoretical value of Q is:
Q＝(q+c)*(1+fn)
where (q + c) represents the normal variance of a single step and (1+ fn) represents the cumulative number. The actual optimum value should be around this value, but the specific value needs to be adjusted by experiment parameters. And ending the predicting step.
The updating step observes data y1The method comprises the following steps:
Figure BDA0002744747830000144
the property of the dirac function is utilized to obtain:
Figure BDA0002744747830000151
setting:
w1 (i)＝fR[y1-h(x1 (i)-)]w0 (i) (15)
at which time the weights change. Then:
Figure BDA0002744747830000152
the update step does not change the particle position but changes the weight of the particle.
Then, weight normalization is carried out:
Figure BDA0002744747830000153
because of the single-step particle filtering, the resampling step is omitted. And completing a round of particle filtering, and then returning the result to the client as the final positioning result of the current step. And when the filter receives the PDR positioning result and the RSSI positioning result of the next step, starting the next judgment cycle. The overall flow of the adaptive periodic particle filter is shown in fig. 7.
3) Algorithm performance analysis
The adaptive periodic particle filter algorithm proposed herein is shown as algorithm 1.
Figure BDA0002744747830000154
Figure BDA0002744747830000161
And (2) calculating according to specific codes of the algorithm, wherein n is the number of particles, m is the number of steps, and the time complexity of the traditional particle filter algorithm for resampling by using SIS is as follows:
3n+m(n+2n+5n+n+n*2n+n+c)+c＝3n+m(2n2+10n + c) + c (18) if other optimizations are usedThe resampling method can reduce the time complexity to:
3n+12mn+c (19)
the time complexity of the adaptive periodic particle filter algorithm proposed herein is:
Figure BDA0002744747830000162
it can be seen that, from the aspect of algorithm time complexity, both of the two finally belong to the order of O (mn), but the coefficient of the method proposed herein is smaller at the same order, and indeed the time consumption can be effectively reduced.
Results and analysis of the experiments
The invention simulates PDR positioning and RSSI trilateral positioning processes in an indoor environment in an MATLAB platform, tests the fusion performance of the traditional particle filter and the self-adaptive periodic particle filter provided by the invention, and compares the model fusion effect by using indexes RMSE root mean square error (formula 19), MAE average absolute error (formula 20) and 3 indexes of the operation duration of a model on the MATlab platform.
Figure BDA0002744747830000163
Figure BDA0002744747830000164
The experimental part is as follows: (1) analyzing the optimal particle number of the model, and selecting the proper particle number by measuring the relationship between the particle number and the error and calculating the time delay; (2) model prediction step variance analysis for verifying a suitable prediction step variance interval; (3) and comparing the fusion effects of the models, and verifying the effectiveness of the self-adaptive periodic particle filter through comparison of three parameter dimensions.
(1) Model optimal particle number analysis
First, analog data needs to be generated. Setting the real positioning value to step by 1 in the x and y directions within 100 unit time of the pedestrian; the RSSI positioning takes a real value as a center, and the fluctuation variance in the x direction and the y direction is 2; PDR positioning error accumulation in x and y directions of each step is 0.1, and fluctuation variance is 0.2. The 3D effect of the analog signal at this time is shown in fig. 8. The projections of which on three coordinate planes are shown in fig. 9.
As can be seen from fig. 9: the RSSI positioning value fluctuates around the true value all the time; as the pedestrian advances, errors of PDR positioning in the directions of the x axis and the y axis are accumulated, but the advancing direction of the pedestrian is basically unchanged, and the method and the device accord with the conception of the invention.
To determine the optimal particle number of the model, it is necessary to know the comparison of indexes of the model under different particle numbers. The three indexes set up here are the MAE mean absolute error, the RMSE root mean square error and the model running time. Because the result of the particle filtering has fluctuation, the invention carries out 10 times of tests on models with different particle numbers, and the average value of each index of the 10 times of test operation results is used as a reference value for drawing so as to ensure the reliability of data. The results are shown in FIGS. 11 and 12. It can be seen from the figure that when the particle number is greater than 200, although the error can be reduced by a small amount even if the particle number is continuously increased, the time consumption of the model is linearly increased with the particle number. Therefore, the number of particles is continuously increased, the cost of the method is judged to be inconsistent with the achievement, and finally 200 particles are used as the optimal test parameters.
(2) Model predictive step variance analysis
And basically consistent with the step of determining the optimal particle number, observing error expression of a filtering result by changing the variance of the prediction step for multiple times, thereby determining an optimal parameter interval. Since the optimum prediction step variance varies from one simulation data to another, but the intervals are approximately the same, the actual model parameters may be set within the ranges. Multiple sets of data were averaged over multiple experiments and the results are shown in fig. 13. It can be seen from the figure that the optimal interval for the predicted step variance is about (1, 1.4).
(3) Model fusion effect display and comparison
Fig. 14 illustrates the fusion process of the adaptive periodic particle filter in a single step. As shown, in steps 1 and 2, error accumulation occurs gradually, and the final positioning value gradually moves away from the true value. And in the step3, after the RSSI positioning is fused, the final positioning value is pulled close to the true value. Step 4 shows that even if the head is pulled over, the error is accumulated and gradually approaches the true value. Fig. 15 shows the error variation of this process in terms of MAE.
Each time, the experiment is performed with the same simulation data, and each parameter of the conventional particle filter model and the adaptive periodic particle filter model is respectively adjusted to be optimal in the above manner, and the final filter curves of the conventional particle filter model and the adaptive periodic particle filter model are shown in fig. 16 and 17.
As the RSSI has stronger positioning fluctuation, the traditional particle filter curve which is continuously fused has more obvious fluctuation. And the self-adaptive periodic particle filter only fuses a specific RSSI positioning value, so that a filter curve is relatively smooth.
The performance of the three metrics of the two models under a plurality of different simulation data is shown in fig. 18, fig. 19 and fig. 20.
It can be seen from the figure that the adaptive periodic particle filter proposed by the present invention is substantially equal to the conventional particle filter in terms of accuracy performance. And the self-adaptive periodic particle filtering is obviously superior to the traditional particle filtering in time consumption indexes.
Experiments successfully verify that the time consumption of a fusion model can be reduced by 58% under the condition of keeping the positioning accuracy basically unchanged when the self-adaptive periodic particle filter algorithm provided by the invention is used for indoor positioning by fusing PDR positioning and RSSI positioning, so that the real-time performance of an indoor positioning system is ensured.
It will be understood that modifications and variations can be made by persons skilled in the art in light of the above teachings and all such modifications and variations are intended to be included within the scope of the invention as defined in the appended claims.

Claims (8)

1. A multi-source fusion indoor positioning system based on adaptive periodic particle filtering is characterized by comprising: the system comprises a data source, a mobile terminal and a server side; wherein:
the data source comprises a Bluetooth beacon and a built-in sensor of the mobile terminal; planning and laying a Bluetooth beacon in an indoor environment, receiving broadcasts from more than three Bluetooth beacons through a mobile terminal, and uploading the received Bluetooth beacon ID and the corresponding signal value strength RSSI to a server side;
the mobile terminal determines stepping, walking directions and step lengths by acquiring the numerical value of a sensor built in the mobile terminal, and performs Pedestrian Dead Reckoning (PDR) positioning locally on the mobile terminal;
the server end is used for converting the received signal value strength RSSI into a distance through a signal strength attenuation model so as to carry out trilateral positioning; and fusing the RSSI positioning result and the PDR positioning result, constructing a self-adaptive periodic particle filter multi-source fusion model, obtaining a final positioning result, and sending the final positioning result to the mobile terminal.
2. The adaptive periodic particle filter-based multi-source fusion indoor positioning system of claim 1, wherein the mobile terminal built-in sensor comprises: a gyroscope, an accelerometer and a geomagnetic sensor.
3. A multisource fusion indoor positioning method based on self-adaptive periodic particle filtering is characterized by comprising the following steps:
step 1: constructing a Bluetooth RSSI trilateral positioning model:
planning and laying a Bluetooth beacon in an indoor environment, receiving broadcasts from more than three Bluetooth beacons through a mobile terminal, and uploading the received Bluetooth beacon ID and the corresponding signal value strength RSSI to a server side;
step 2: constructing a PDR positioning model of a built-in sensor of the mobile terminal:
acquiring the numerical value of a built-in sensor of the mobile terminal, determining the stepping direction, the walking direction and the step length, and locally positioning the pedestrian dead reckoning PDR at the mobile terminal;
and step 3: constructing a self-adaptive periodic particle filter multi-source fusion model:
converting the received signal value strength RSSI into a distance through a signal strength attenuation model so as to carry out trilateral positioning; and fusing the RSSI positioning result and the PDR positioning result, constructing a self-adaptive periodic particle filter multi-source fusion model, obtaining a final positioning result, and sending the final positioning result to the mobile terminal.
4. The multi-source fusion indoor positioning method based on the adaptive periodic particle filtering of claim 3, wherein the specific method of trilateral positioning is as follows:
during trilateral positioning calculation, solving and reducing a final positioning error through a multi-AP simultaneous equation, and calculating a distance through a space loss model; the spatial loss model is:
Figure FDA0002744747820000021
wherein d represents the distance between the beacon and the receiving point; d0Represents a reference distance; PL (d) represents the signal strength RSSI at distance d; a represents a path loss exponent; ω represents a mean of 0 and a variance of
Figure FDA0002744747820000024
(ii) a gaussian distribution variable; let diAs the distance between the beacon i and the receiving point, it can be known that:
Figure FDA0002744747820000022
5. the multi-source fusion indoor positioning method based on adaptive periodic particle filtering according to claim 3, wherein the specific formula of PDR positioning is as follows:
Figure FDA0002744747820000023
in the formula xk、ykThe abscissa and ordinate of step k, lkRepresentsStep size of step k, θkRepresents the angle between the advancing direction of the kth step and the x-axis, deltax、δyRepresenting errors in the x and y directions.
6. The adaptive periodic particle filter-based multi-source fusion indoor positioning method of claim 3, wherein the adaptive periodic particle filter multi-source fusion model comprises a method for performing adaptive condition judgment:
constructing an adaptive periodic particle filter, and after the filter receives a PDR positioning result and an RSSI positioning result, adaptively judging whether particle fusion is needed:
in the case of one-dimensional situation, the conditions for adaptive judgment are as follows:
assuming that the PDR positioning error is accumulated to be a negative number, namely the PDR positioning value is gradually lower than the true value along with the advance of the person; setting the RSSI positioning result to be normal fluctuation with the true value as expected and R as variance; if the RSSI positioning fluctuation at the time t is higher than the final positioning value at the time t-1 by a certain amount d, judging that the PDR positioning error is accumulated to a certain degree, and performing particle filter fusion;
in the case of two-dimensional situation, the conditions for adaptive judgment are as follows:
(RSSI.x(t)-final.x(t-1)＞d)&(RSSI.y(t)-final.y(t-1)＞d)
wherein RSSI.x (t) and RSSI.y (t) represent x and y coordinates of the RSSI positioning result at the time t; final.x (t-1) and final.y (t-1) represent the x and y coordinates of the final positioning value at the time t-1; the value of d is determined according to the specific fluctuation condition of the RSSI, and the filtering frequency is controlled by controlling the size of d, so that the final precision of the model is improved by tuning;
if the self-adaptive judgment condition is met, performing particle filter fusion; if the self-adaptive judgment condition is not met, the PDR positioning result is directly used as a final positioning result.
7. The multisource fusion indoor positioning method based on the adaptive periodic particle filtering of claim 4, wherein the specific method for multisource fusion of the adaptive periodic particle filtering is as follows:
step 1: initialization: when the filtering starts, firstly, initializing, assigning the coordinates and weights of n particles with a predetermined number, setting the initial coordinates of the current step as the final coordinates of the previous step and the weights of the particles as 1/n, and entering a prediction step after initialization;
step 2: a prediction step: let random process satisfy Xk＝f(Xk-1)+QkObserved to satisfy Yk＝h(Xk)+RkBayesian filtering requires knowledge of the initial value x0Probability density function f0 +(ii) a The prediction step is infinite integration:
Figure FDA0002744747820000031
let f (x) be the probability density function of x, the theorem of maxima states that n → ∞ time:
Figure FDA0002744747820000032
the concept of weight is introduced:
Figure FDA0002744747820000033
initial weight of each particle is
Figure FDA0002744747820000034
For different xiSetting up different wi；
The position and weight of the particle completely determine the probability distribution function, and also determine the probability density function; if xiIs sampled from f (x), then xiSatisfies the rule of probability distribution; let x be0Obeying normal distribution, the collected sample is x0 (1),x0 (2),…,x0 (n)(ii) a Setting:
Figure FDA0002744747820000035
then:
Figure FDA0002744747820000041
f1 +(x)＝ηfR[y-h(x)]f1 -(x)
if the distribution is normal distribution, directly sampling; and (3) generating particles:
Figure FDA0002744747820000042
for each fQ[x-f(x0 (i))]Consider an inevitable event X ═ f (X)0 (i)) Superposition with a random number Y-N (0, Q); then f1 -(x) Particle X1 -(1),X1 -(2),…,X1 -(n)Comprises the following steps:
X1 -(i)＝f(x0 (i))+v
wherein v is a random number of N (0, Q); at this time x1 -(i)＝f(x0 (i)) The + Q nature is that the position of the particle is changed and the weight is not changed; the Q of the prediction step is different from that of the traditional particle filter, and the value is large; setting q as the normal fluctuation variance of the original prediction data, c as a certain small constant related to the error accumulation speed, and fn as the number of times of failure judgment of the current fusion condition; the theoretical value of Q is:
Q＝(q+c)*(1+fn)
wherein (q + c) represents the normal variance of a single step, and (1+ fn) represents the cumulative number, and the actual optimal value is near the value, but the specific value is obtained by adjusting parameters through experiments; ending the predicting step;
step 3: and (5) updating.
8. The multi-source fusion indoor positioning method based on the adaptive periodic particle filtering of claim 7, wherein the specific method of the updating step is as follows:
the updating step observes data y1The method comprises the following steps:
Figure FDA0002744747820000043
using the properties of the dirac function yields:
Figure FDA0002744747820000044
setting:
w1 (i)＝fR[y1-h(x1 (i)-)]w0 (i)
at this time, the weight changes, then:
Figure FDA0002744747820000051
the update step does not change the particle position, but changes the weight of the particle;
carrying out weight normalization:
Figure FDA0002744747820000052
because of the single-step particle filtering, the resampling step is omitted. Thus, a round of particle filtering is completed. Then, returning the result to the mobile terminal as the final positioning result of the current step; and when the filter receives the PDR positioning result and the RSSI positioning result of the next step, starting the next judgment cycle.
CN202011162672.4A 2020-10-27 2020-10-27 Multi-source fusion indoor positioning system and method based on self-adaptive periodic particle filtering Active CN112333818B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202011162672.4A CN112333818B (en) 2020-10-27 2020-10-27 Multi-source fusion indoor positioning system and method based on self-adaptive periodic particle filtering

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202011162672.4A CN112333818B (en) 2020-10-27 2020-10-27 Multi-source fusion indoor positioning system and method based on self-adaptive periodic particle filtering

Publications (2)

Publication Number Publication Date
CN112333818A true CN112333818A (en) 2021-02-05
CN112333818B CN112333818B (en) 2021-11-02

Family

ID=74295981

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202011162672.4A Active CN112333818B (en) 2020-10-27 2020-10-27 Multi-source fusion indoor positioning system and method based on self-adaptive periodic particle filtering

Country Status (1)

Country Link
CN (1) CN112333818B (en)

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113093255A (en) * 2021-05-07 2021-07-09 深圳市前海智车科技有限公司 Multi-signal true fusion positioning calculation method, device, equipment and storage medium
CN113938853A (en) * 2021-09-26 2022-01-14 众虎物联网(广州)有限公司 Positioning method, device and system for indoor navigation
CN114302359A (en) * 2021-12-02 2022-04-08 中国船舶重工集团公司第七0九研究所 High-precision indoor positioning method based on WiFi-PDR fusion
TWI764842B (en) * 2021-10-13 2022-05-11 英屬維爾京群島商飛思捷投資股份有限公司 Ranging-type positioning system and method based on crowdsourced calibration
CN114710745A (en) * 2022-04-12 2022-07-05 电子科技大学 Indoor positioning method with Bluetooth and PDR information deeply fused

Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR20100116848A (en) * 2009-04-23 2010-11-02 (주)대우건설 Controlling apparatus for improving stability on maintenance for water purification systems by membrane filters and method thereof
CN104482933A (en) * 2014-12-03 2015-04-01 北京航空航天大学 Method for dead reckoning and WLAN (wireless local area network) integrated positioning based on particle filter
CN108151747A (en) * 2017-12-27 2018-06-12 浙江大学 A kind of indoor locating system and localization method merged using acoustical signal with inertial navigation
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
CN109669201A (en) * 2018-11-13 2019-04-23 斯沃德教育科技股份有限公司 It is a kind of based on wearable indoor and outdoor location algorithm
CN110602647A (en) * 2019-09-11 2019-12-20 江南大学 Indoor fusion positioning method based on extended Kalman filtering and particle filtering
CN111698774A (en) * 2020-06-23 2020-09-22 北京麦钉艾特科技有限公司 Indoor positioning method and device based on multi-source information fusion
CN111735458A (en) * 2020-08-04 2020-10-02 西南石油大学 Navigation and positioning method of petrochemical inspection robot based on GPS, 5G and vision

Patent Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR20100116848A (en) * 2009-04-23 2010-11-02 (주)대우건설 Controlling apparatus for improving stability on maintenance for water purification systems by membrane filters and method thereof
CN104482933A (en) * 2014-12-03 2015-04-01 北京航空航天大学 Method for dead reckoning and WLAN (wireless local area network) integrated positioning based on particle filter
CN108151747A (en) * 2017-12-27 2018-06-12 浙江大学 A kind of indoor locating system and localization method merged using acoustical signal with inertial navigation
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
CN109669201A (en) * 2018-11-13 2019-04-23 斯沃德教育科技股份有限公司 It is a kind of based on wearable indoor and outdoor location algorithm
CN110602647A (en) * 2019-09-11 2019-12-20 江南大学 Indoor fusion positioning method based on extended Kalman filtering and particle filtering
CN111698774A (en) * 2020-06-23 2020-09-22 北京麦钉艾特科技有限公司 Indoor positioning method and device based on multi-source information fusion
CN111735458A (en) * 2020-08-04 2020-10-02 西南石油大学 Navigation and positioning method of petrochemical inspection robot based on GPS, 5G and vision

Cited By (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113093255A (en) * 2021-05-07 2021-07-09 深圳市前海智车科技有限公司 Multi-signal true fusion positioning calculation method, device, equipment and storage medium
CN113093255B (en) * 2021-05-07 2024-05-07 深圳市前海智车科技有限公司 Multi-signal true fusion positioning calculation method, device, equipment and storage medium
CN113938853A (en) * 2021-09-26 2022-01-14 众虎物联网(广州)有限公司 Positioning method, device and system for indoor navigation
CN113938853B (en) * 2021-09-26 2023-05-12 众虎物联网(广州)有限公司 Positioning method, device and system for indoor navigation
TWI764842B (en) * 2021-10-13 2022-05-11 英屬維爾京群島商飛思捷投資股份有限公司 Ranging-type positioning system and method based on crowdsourced calibration
CN114302359A (en) * 2021-12-02 2022-04-08 中国船舶重工集团公司第七0九研究所 High-precision indoor positioning method based on WiFi-PDR fusion
CN114302359B (en) * 2021-12-02 2023-08-01 中国船舶重工集团公司第七0九研究所 WiFi-PDR fusion-based high-precision indoor positioning method
CN114710745A (en) * 2022-04-12 2022-07-05 电子科技大学 Indoor positioning method with Bluetooth and PDR information deeply fused
CN114710745B (en) * 2022-04-12 2023-04-18 电子科技大学 Indoor positioning method with Bluetooth and PDR information deeply fused

Also Published As

Publication number Publication date
CN112333818B (en) 2021-11-02

Similar Documents

Publication Publication Date Title
CN112333818B (en) Multi-source fusion indoor positioning system and method based on self-adaptive periodic particle filtering
Zhu et al. Adapted error map based mobile robot UWB indoor positioning
Jourdan et al. Optimal sensor placement for agent localization
CN102752855B (en) Indoor personnel positioning system and method based on path rule and prediction
CN102395192B (en) Method and device for locating wireless sensor terminal
CN104482933B (en) A kind of method based on particle filter reckoning and WLAN integrated positioning
CN103997717A (en) Real-time indoor positioning system and method
CN108919177B (en) Positioning map construction method based on virtual information source estimation and track correction
CN110933599B (en) Self-adaptive positioning method fusing UWB and WIFI fingerprints
CN105372628A (en) Wi-Fi-based indoor positioning navigation method
WO2010106530A2 (en) A location and tracking system
CN111970633A (en) Indoor positioning method based on WiFi, Bluetooth and pedestrian dead reckoning fusion
Graefenstein et al. Robust method for outdoor localization of a mobile robot using received signal strength in low power wireless networks
Khan et al. Indoor Wi-Fi positioning algorithm based on combination of location fingerprint and unscented Kalman filter
Xu et al. An improved indoor localization method for mobile robot based on WiFi fingerprint and AMCL
You et al. Indoor positioning system with cellular network assistance based on received signal strength indication of beacon
Xia et al. An improved PSO localization algorithm for UWB sensor networks
Cheng et al. UWB/INS Fusion Positioning Algorithm based on Generalized Probability Data Association for Indoor Vehicle
Li et al. Improvement of triangle centroid localization algorithm based on PIT criterion (ITCL-PIT) for WSNs
Zhao et al. A UWB/Bluetooth fusion algorithm for indoor localization
CN114521014B (en) Method for improving positioning precision in UWB positioning process
Zhou et al. An improved MCB localization algorithm based on weighted RSSI and motion prediction
Liu et al. An UWB/PDR fusion algorithm based on improved square root unscented Kalman filter
Souli et al. Adaptive frequency band selection for accurate and fast positioning utilizing SOPs
Zhao et al. Adaptive range-based nonlinear filters for wireless indoor positioning system using dynamic Gaussian model

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