Method and apparatus for utilizing multi-map combination for fingerprinting and deep learning positioning

Abstract

A method utilizing a multi-map combination for fingerprinting and deep learning positioning according to an embodiment of the present invention comprises: a) a step in which a first positioning device creates a plurality of fingerprinting maps that are matched with a plurality of grids and divided into grid units based on RSSI values ​​of BLE beacons installed in a plurality of grids; b) a step in which the first positioning device constructs a fingerprinting multi-map by aggregating a plurality of fingerprinting maps according to environmental variables affecting positioning performance, and measures the location of a user located within an area within the plurality of grids based on a fingerprinting map reconstructed by reflecting weights of environmental variable information to the fingerprinting maps according to environmental variables; c) a step in which a second positioning device assigns colors according to the RSSI values ​​of the BLE beacons and then creates a plurality of deep learning maps, which are images in a form that match the plurality based on the RSSI values ​​to which the colors are assigned. and d) a step of the second positioning device assembling a plurality of deep learning maps for each environmental variable to construct a deep learning multi-map, and reconstructing the reconstructed deep learning map or deep learning positioning result data for each environmental variable by reflecting the weights of environmental variable information received from the first positioning device to measure the location of the user; may be included.

Description

Technology Field

[0001] The present invention relates to a method and apparatus utilizing a multi-map combination in fingerprinting and deep learning positioning, and more specifically, to a method and apparatus utilizing a multi-map combination in fingerprinting and deep learning positioning that can measure the location of a user using a fingerprinting and deep learning multi-map. Background Technology

[0002] The Smart Gate Free Payment System is a technology that allows passengers carrying smart devices, such as smartphones, to pay fares in a walk-through manner without physical contact, such as boarding and alighting tags, when using public transportation like subways.

[0003] As shown in FIG. 1 (a), this smart gate-free payment system installs a number of BLE (Bluetooth Low Energy) beacons on a grid (10) installed on the floor of a place where passengers pass, and positions passengers using RSSI (Received Signal Strength Indicator)-based triangulation or fingerprinting techniques, thereby enabling automatic payment of public transportation fares in a walk-through manner.

[0004] Additionally, in a smart gate free payment system, it is assumed that the passenger's positioning location is made up of a grid (10), and accordingly, determining the passenger's location on the grid (10) can be defined as positioning.

[0005] Among RSSI-based triangulation and fingerprinting techniques, triangulation has the problem of reduced positioning accuracy due to large variations in RSSI caused by the characteristics of radio waves, and as a result of this problem with triangulation, fingerprinting techniques are mainly used.

[0006] In order to use the fingerprinting method, as shown in (b) of FIG. 1, a plurality of wireless routers (AP 1 to 4) acquire the RSSI values ​​of BLE beacons at each grid (10) for a certain period of time, and a fingerprinting map can be produced using the average value, median value, maximum value, etc. calculated based on this.

[0007] To implement this fingerprinting technique, the work is typically divided into offline and online stages. In the offline stage, a fingerprinting map is created in advance using wireless communication signal strength, and in the online stage, positioning is performed by comparing the current received signal strength with the fingerprinting map.

[0008] Meanwhile, the grid (10) in which BLE beacons are installed is composed of a plurality of grids (10) forming a matrix structure, and each grid (10) is labeled. For example, among the plurality of grids (10) in which BLE beacons are installed, 10 grids (10) arranged in one column can be labeled (6-2 to 10-3) as shown in FIG. 2.

[0009] In this way, when four fingerprinting maps produced on different dates are applied to each labeled grid (10), the appearance of the fingerprinting map for each grid (10) is the same as the graph shown in FIG. 2, and the change in the appearance of the fingerprinting map for each grid (10) is confirmed based on the fact that the graph for each grid (10) is different.

[0010] Changes in such fingerprinting maps are caused by factors such as changes in the surrounding environment over time, which implies a degradation in position estimation performance; therefore, when the degree of change is significant, it is necessary to create a new fingerprinting map.

[0011] Therefore, there is a need for measures to minimize the degradation of positioning performance of the fingerprinting map even if the fingerprinting map changes over time. Prior art literature

[0012] Republic of Korea Published Patent Application No. 10-2017-0078116 (Published July 7, 2017) The problem to be solved

[0013] Accordingly, the present invention has been devised to solve the above-mentioned problems, and the objective of the present invention is to provide a method and apparatus utilizing a multi-map combination in fingerprinting and deep learning positioning, which can minimize performance degradation of the fingerprinting map caused by environmental variables by constructing a fingerprinting multi-map based on a plurality of fingerprinting maps and reflecting the weight of the environmental variable in the fingerprinting map suitable for the environmental variable among the fingerprinting multi-maps, and can measure the location of a user through the fingerprinting map with minimized performance degradation.

[0014] In addition, the objective of the present invention is to provide a method and apparatus for utilizing a multi-map combination in fingerprinting and deep learning positioning, which can minimize performance degradation of a deep learning map caused by environmental variables by constructing a deep learning multi-map based on a deep learning map that is an input value of a CNN or RNN model and reflecting the weights of environmental variables in a deep learning map suitable for environmental variables among the deep learning multi-maps, and can measure the location of a user through a deep learning map with minimized performance degradation.

[0015] However, the technical problems to be solved by the present invention are not limited to those mentioned above, and other technical problems not mentioned will be clearly understood by those skilled in the art to which the present invention belongs from the description below. means of solving the problem

[0016] A method for utilizing a multi-map combination in fingerprinting and deep learning positioning according to an embodiment of the present invention as a technical method for achieving the above-mentioned purpose comprises: a) a step in which a first positioning device creates a plurality of fingerprinting maps that are matched with a plurality of grids and divided into grid units based on the RSSI values ​​of BLE beacons installed in a plurality of grids; b) a step in which the first positioning device constructs a fingerprinting multi-map by aggregating a plurality of fingerprinting maps according to environmental variables affecting positioning performance, and measures the location of a user located within an area within the plurality of grids based on a fingerprinting map reconstructed by reflecting the weight of environmental variable information in the fingerprinting map according to the environmental variable; c) a step in which a second positioning device assigns a color according to the RSSI values ​​of the BLE beacons and then creates a plurality of deep learning maps, which are images in a form that match the plurality based on the color-assigned RSSI values. and d) a step of the second positioning device assembling a plurality of deep learning maps for each environmental variable to construct a deep learning multi-map, and reconstructing the reconstructed deep learning map or deep learning positioning result data for each environmental variable by reflecting the weights of environmental variable information received from the first positioning device to measure the location of the user; may be included.

[0017] Meanwhile, a device utilizing a multi-map combination for fingerprinting and deep learning positioning according to an embodiment of the present invention, which is a technical means for performing a method utilizing a multi-map combination for the above-mentioned fingerprinting and deep learning positioning, comprises: a first positioning device that creates a plurality of fingerprinting maps that are matched with the plurality of grids and divided into grid units based on the RSSI values ​​of BLE beacons installed in the plurality of grids, constructs a fingerprinting multi-map by aggregating the plurality of fingerprinting maps according to environmental variables affecting positioning performance, and measures the location of a user located within an area of ​​the plurality of grids based on a fingerprinting map reconstructed by reflecting the weight of environmental variable information in the fingerprinting maps according to the environmental variables; The second positioning device may include: a deep learning map that is an image matching the plurality of elements based on the colored RSSI values ​​of the BLE beacons, and a deep learning multi-map that is constructed by combining the plurality of deep learning maps according to environmental variables, and a deep learning map or deep learning positioning result data according to environmental variables that is reconstructed by reflecting the weights of environmental variable information transmitted from the first positioning device to measure the location of the user. Effects of the invention

[0018] According to one embodiment of the present invention, by reflecting the weights of environmental variables in a fingerprinting and deep learning map for measuring a user's location, a fingerprinting and deep learning map robust to environmental variables can be produced.

[0019] According to one embodiment of the present invention, by producing fingerprinting and deep learning maps that are robust to environmental variables, the performance degradation of the fingerprinting map and deep learning map is minimized, and the location of the user can be measured as accurately as possible using the fingerprinting map or deep learning map with minimized performance degradation.

[0020] According to one embodiment of the present invention, a smart gate-free payment system that enables automatic payment of public transportation fares in a walk-through manner based on accurately measuring the user's location can be easily implemented in an indoor location.

[0021] However, the effects obtainable from the present invention are not limited to those mentioned above, and other unmentioned effects will be clearly understood by those skilled in the art from the description below. Brief explanation of the drawing

[0022] Figure 1 is a diagram illustrating the installation method of a BLE beacon and the method of creating a fingerprinting map. Figure 2 is a graph showing the change in appearance of multiple labeled grids and fingerprinting maps produced on different dates. FIG. 3 is a block diagram schematically illustrating the components of a device utilizing a multi-map combination for fingerprinting positioning according to an embodiment of the present invention. Figure 4 is a drawing illustrating a fingerprinting multi-map produced through the fingerprinting multi-map construction unit shown in Figure 3. Figure 5 is a diagram for explaining the positioning method of the first positioning unit shown in Figure 3. Figure 6 is a diagram illustrating the process of a method utilizing a multi-map combination for fingerprinting positioning based on an environment weight map-based positioning method. Figure 7 is a diagram illustrating the detailed process of the weight calculation step shown in Figure 6. Figure 8 is a diagram illustrating the process of a method utilizing a multi-map combination for fingerprinting positioning based on a distance sum-based positioning method. Figure 9 is a diagram illustrating the process of a method utilizing a multi-map combination for fingerprinting positioning based on a judgment result aggregation positioning method. FIG. 10 is a block diagram schematically illustrating the components of a device utilizing a multi-map combination for deep learning positioning according to another embodiment of the present invention. FIG. 11 is a drawing illustrating an example of a grid image input to the second positioning unit shown in FIG. 10. Figure 12 is a diagram illustrating the process of a method utilizing a multi-map combination for deep learning positioning based on deep learning positioning results. Figure 13 is a diagram illustrating the process of a method utilizing a multi-map combination for map-based deep learning positioning. Figure 14 is a graph showing the positioning performance results of the fingerprinting map. Specific details for implementing the invention

[0023] Hereinafter, embodiments of the present invention are described in detail with reference to the attached drawings so that those skilled in the art can easily implement the present invention. However, since the description of the present invention is merely an example for structural or functional explanation, the scope of the present invention should not be interpreted as being limited by the embodiments described in the text. That is, since the embodiments are subject to various modifications and may take various forms, the scope of the present invention should be understood to include equivalents capable of realizing the technical concept. Furthermore, the objectives or effects presented in the present invention do not imply that a specific embodiment must include all of them or only such effects; therefore, the scope of the present invention should not be understood as being limited by them.

[0024] The meaning of the terms described in this invention should be understood as follows.

[0025] Terms such as "first" and "second" are intended to distinguish one component from another, and the scope of rights shall not be limited by these terms. For example, the first component may be named the second component, and similarly, the second component may be named the first component. When a component is referred to as being "connected" to another component, it should be understood that it may be directly connected to that other component, or that there may be other components in between. Conversely, when a component is referred to as being "directly connected" to another component, it should be understood that there are no other components in between. Meanwhile, other expressions describing the relationship between components, such as "between" and "exactly between," or "adjacent to" and "directly adjacent to," shall be interpreted in the same manner.

[0026] A singular expression should be understood to include a plural expression unless the context clearly indicates otherwise, and terms such as "include" or "have" are intended to specify the existence of the set-up features, numbers, steps, actions, components, parts, or combinations thereof, and should be understood not to preclude the existence or addition of one or more other features, numbers, steps, actions, components, parts, or combinations thereof.

[0027] Unless otherwise defined, all terms used herein have the same meaning as generally understood by those skilled in the art to which this invention pertains. Terms defined in commonly used dictionaries should be interpreted as having meanings consistent with the context of the relevant technology and should not be interpreted as having an ideal or overly formal meaning unless explicitly defined in this invention.

[0029] Device and method utilizing multi-map combination for fingerprinting positioning

[0030] FIG. 3 is a block diagram schematically illustrating the components of a device utilizing a multi-map combination for fingerprinting positioning according to an embodiment of the present invention.

[0031] Referring to FIG. 3, a device (100) utilizing a multi-map combination for fingerprinting positioning according to one embodiment of the present invention includes a fingerprinting multi-map construction unit (110) for constructing a fingerprinting multi-map and a first positioning unit (120) for measuring the location of a user using the fingerprinting multi-map.

[0032] In the present invention, the device (100) utilizing a multi-map combination for fingerprinting positioning is preferably understood as a first positioning device (100) that positions a user based on a fingerprinting multi-map.

[0033] The fingerprinting multi-map construction unit (110) can produce a fingerprinting map based on modeling that divides specific areas of an indoor place into grid units, and this step may be an offline step.

[0034] Here, a specific area of ​​the indoor space may be a place where a smart gate-free payment system is implemented to allow public transportation fares to be automatically paid via a walk-through method.

[0035] That is, in the present invention, the user refers to a passenger who pays public transportation fares in a walk-through manner through a customer service (smart gate free payment system) provided by a device (100) that utilizes a multi-map combination for fingerprinting positioning.

[0036] Additionally, in a specific area of ​​an indoor space, a plurality of grids (10) shown in FIG. 1 may be arranged to match each grid of the fingerprinting map for the production of the fingerprinting map, and a BLE beacon (not shown) may be installed in each grid (10).

[0037] And in a specific area of ​​the indoor location, a wireless router (AP) providing a Wi-Fi channel to transmit the RSSI value of the BLE beacon acquired in real time to the fingerprinting multi-map construction unit (110) can be placed at a grid unit length (e.g., 0.5 to 2 m).

[0038] In the fingerprinting multi-map construction unit (110), the fingerprinting map to be produced based on the RSSI value of the BLE beacon changes shape depending on environmental variables, and there is a problem that positioning performance is degraded due to this change in shape.

[0039] Here, environmental variables that degrade positioning performance may include the usage status of Wi-Fi channels provided by wireless routers, the usage status of Bluetooth devices, the situation of trains entering the platform, and the flow of people at the station gates.

[0040] In the present invention, environmental variables will be described as being limited to the above variables for the sake of convenience of explanation, but in addition to the above variables, variables that affect the fingerprinting map may be added.

[0041] In the present invention, the fingerprinting multi-map construction unit (110) can construct a fingerprinting multi-map based on producing a plurality of fingerprinting maps in order to improve the degradation of the positioning performance of the fingerprinting map.

[0042] In addition, the fingerprinting multi-map construction unit (110) constructs a fingerprinting multi-map by combining multiple fingerprinting maps according to environmental variables, and an example of the fingerprinting multi-map constructed in this way is as shown in FIG. 4.

[0043] Figure 4 is a drawing illustrating a fingerprinting multi-map produced through the fingerprinting multi-map construction unit shown in Figure 3.

[0044] Referring to FIG. 4, the fingerprinting multi-map of the present invention may include a first fingerprinting multi-map (111) that reflects radio interference of a Wi-Fi channel provided by a wireless router.

[0045] In the present invention, a plurality of wireless routers provide different Wi-Fi channels, and it is assumed that a smart device (e.g., a smartphone) equipped with a BLE beacon and Wi-Fi function installed at each grid (10) uses the plurality of Wi-Fi channels in an overlapping manner.

[0046] In addition, it is desirable for smart devices to have multiple Wi-Fi channels to use them in an overlapping manner.

[0047] And it is advisable for users located in specific areas of an indoor space to have a smart device.

[0048] In the present invention, when multiple smart devices use multiple Wi-Fi channels in overlap, radio interference problems occur, and such radio interference causes a decrease in the positioning performance of the fingerprinting map.

[0049] In the present invention, the cause of radio interference of the Wi-Fi channel differs before and after the fingerprinting multi-map construction.

[0050] As a specific example, before the fingerprinting multi-map is built, radio interference of the Wi-Fi channel may occur when the BLE beacon transmits the RSSI value to the fingerprinting multi-map building unit (110) through multiple wireless routers, and after the fingerprinting multi-map is built, it may occur when a smart device located in a specific area of ​​an indoor place that matches the fingerprinting map transmits the RSSI value to the first positioning unit (120) through multiple wireless routers.

[0051] At this time, multiple wireless routers can transmit identification information (IP address) of the wireless routers to the fingerprinting multi-map construction unit (110) so that the fingerprinting multi-map construction unit (110) can identify the Wi-Fi channel that transmitted the RSSI value.

[0052] In the present invention, the fingerprinting multi-map construction unit (110) creates a fingerprinting map based on the RSSI value of a BLE beacon received through a plurality of wireless routers, and by repeating the process of matching the created fingerprinting map with a Wi-Fi channel that receives the RSSI value from the BLE beacon, the fingerprinting map can be created when there is interference between multiple Wi-Fi channels, thereby enabling the construction of a first fingerprinting multi-map (111).

[0053] That is, the first fingerprinting multi-map (111) can be composed of fingerprinting maps for Wi-Fi channel interference equal to the number of Wi-Fi channels used to transmit the RSSI value of the BLE beacon to the fingerprinting multi-map construction unit (110), and it is preferable that there be multiple fingerprinting maps for Wi-Fi channel interference.

[0054] As a specific example, the first fingerprinting multi-map (111) can be configured as fingerprinting maps (111-1 to 111-n) when the fingerprinting multi-map construction unit (110) obtains RSSI values ​​from a BLE beacon through Wi-Fi channels 1 to n of a wireless router.

[0055] Referring to FIG. 4, the fingerprinting multi-map of the present invention may include a second fingerprinting multi-map (112) that reflects Bluetooth congestion resulting from the use of a Bluetooth device (e.g., earphones, headphones, smartwatch, etc.).

[0056] In the present invention, Bluetooth congestion can be calculated based on BLE beacons installed at each grid (10) scanning Bluetooth devices.

[0057] As a specific example, a BLE beacon can obtain identification information of a Bluetooth device by scanning the Bluetooth device using the Bluetooth function of the Bluetooth device, and can transmit the identification information of the Bluetooth device to a fingerprinting multi-map construction unit (110) through a Wi-Fi channel provided by a wireless router, and the fingerprinting multi-map construction unit (110) can calculate the number of Bluetooth devices for each grid (10) through the identification information of the Bluetooth device received from the BLE beacon, and calculate the Bluetooth congestion level by dividing the sum of the number of Bluetooth devices in the fingerprinting map by the area of ​​the fingerprinting map.

[0058] In the present invention, the fingerprinting multi-map construction unit (110) can construct a second fingerprinting multi-map (112) by repeating the process of matching the calculated Bluetooth congestion of the fingerprinting map with the produced fingerprinting map to produce a plurality of Bluetooth congestion fingerprinting maps.

[0059] Additionally, the fingerprinting multi-map construction unit (110) can match the Bluetooth congestion level to the fingerprinting map as is, but preferably, the Bluetooth congestion level can be converted into multiple Bluetooth congestion levels.

[0060] As a specific example, if the Bluetooth congestion level is a value included in the first congestion level among multiple Bluetooth congestion levels, it can be converted to the first congestion level.

[0061] That is, each Bluetooth congestion fingerprinting map that constructs the second fingerprinting multi-map (112) may be a fingerprinting map matched with at least one of the multiple Bluetooth congestion levels pre-set in the fingerprinting multi-map constructing unit (110).

[0062] In the present invention, the second fingerprinting multi-map (112) may be composed of Bluetooth congestion fingerprinting maps equal to the number of Bluetooth congestion levels, and it is preferable that the plurality of Bluetooth congestion fingerprinting maps be multiple.

[0063] As a specific example, the second fingerprinting multi-map (112) may be composed of first to n Bluetooth congestion fingerprinting maps (112-1 to 112-n) that match the first to n Bluetooth congestion levels when the fingerprinting multi-map constructor (110) classifies the Bluetooth congestion level into first to n Bluetooth congestion levels.

[0064] Referring to FIG. 4, the fingerprinting multi-map of the present invention may include a third fingerprinting multi-map (113) that reflects train radio interference resulting from the train's entry into the platform.

[0065] Positioning errors occur in fingerprinting maps due to train propagation, such as inrush current generated when a train enters a platform.

[0066] In the present invention, the fingerprinting multi-map construction unit (110) can construct a third fingerprinting multi-map (113) by receiving RSSI values ​​acquired by a BLE beacon and, when a train enters a platform at the same time that the fingerprinting map is being produced, measuring the train entry location corresponding to the location information of the train that has entered the platform, and repeating the process of matching the train entry location with the fingerprinting map to produce a plurality of train entry location fingerprinting maps.

[0067] At this time, the fingerprinting multi-map construction unit (110) can predict the train entry location by using the train schedule transmitted from the station management system (not shown) to predict the train entry location at the same time as the production of the train entry location fingerprinting map.

[0068] In the present invention, the train schedule may include time information for when the train enters the platform according to the schedule (timetable).

[0069] In the present invention, the third fingerprinting multi-map (113) may be composed of a number of train entry position fingerprinting maps equal to the number of times the fingerprinting multi-map constructor (110) calculates the degree of interference of train radio waves based on the train entry position, and it is preferable that the train entry position fingerprinting map be composed of a plurality of maps.

[0070] As a specific example, the third fingerprinting multi-map (113) may be composed of the first-n train entry position fingerprinting map (113-1 to 113-n) when the fingerprinting multi-map constructor (110) calculates the degree of interference of train radio waves n times based on the train entry position or a virtual train entry position.

[0071] Referring to FIG. 4, the fingerprinting multi-map of the present invention may include a fourth fingerprinting multi-map (114) that reflects the gate congestion of history.

[0072] Fingerprinting maps experience positioning errors due to the dense flow of people in the station.

[0073] In the present invention, the fingerprinting multi-map construction unit (110) receives the RSSI value acquired by the BLE beacon, calculates the congestion level of the station at the same time as the fingerprinting map is being produced, and then repeats the process of matching the congestion level of the station with the fingerprinting map to produce a plurality of congestion level fingerprinting maps, thereby constructing a fourth fingerprinting multi-map (114).

[0074] At this time, the fingerprinting multi-map construction unit (110) receives an image of the area around the gate of the station from a shooting means (e.g., a vision sensor) installed around the gate of the station to calculate the congestion level of the station, and calculates the flow of people based on determining the number of people passing through the gate of the station at the same time as the congestion fingerprinting map is produced based on the image of the area around the gate of the station, and can calculate the congestion level of the station based on the calculated flow of people.

[0075] That is, the congestion level of the station is calculated based on the number of people passing through the station gates at the same time when the congestion fingerprinting map is produced, and accordingly, in the present invention, it is preferable to understand it as the congestion level of the station gates.

[0076] In the present invention, when it is impossible for the fingerprinting map construction unit (110) to receive images of the area around the gate of the station from the vision sensor, it can predict the number of people at the time when the congestion fingerprinting map is produced through the time-based number of people data transmitted from the station management system, and calculate the congestion level of the station using the predicted number of people.

[0077] In the present invention, the fourth fingerprinting multi-map (114) may be composed of congestion fingerprinting maps as many times as the number of times the fingerprinting multi-map construction unit (110) calculates the gate congestion of the station using the flow of people, and it is preferable that the congestion fingerprinting map be composed of a plurality of maps.

[0078] As a specific example, the fourth fingerprinting multi-map (114) may be composed of the first-n congestion fingerprinting map (114-1 to 114-n) when the fingerprinting multi-map construction unit (110) calculates the gate congestion of the history n times.

[0079] In the present invention, the fingerprinting multi-map construction unit (110) constructs the first, second, third, and fourth fingerprinting multi-maps (111 to 114) as described above, and the fingerprinting multi-maps may be added or changed if an environment variable matching the fingerprinting map is added or changed. However, for convenience of explanation, the fingerprinting multi-maps in the present invention will be described as the first, second, third, and fourth fingerprinting multi-maps (111 to 114).

[0080] In the present invention, the fingerprinting multi-map construction unit (110) can store the first, second, third, and fourth fingerprinting multi-maps (111 to 114) on a server (not shown).

[0081] Additionally, the fingerprinting multi-map construction unit (110) performs synchronization with the server whenever the first, second, third, and fourth fingerprinting multi-maps (111~114) are updated, thereby allowing the first, second, third, and fourth fingerprinting multi-maps (111~114) stored on the server to be updated.

[0082] Figure 5 is a diagram for explaining the positioning method of the first positioning unit shown in Figure 3.

[0083] Referring to FIG. 5, the first positioning unit (120) receives the first, second, third, and fourth fingerprinting multi-maps (111 to 114) through a server and can store the first, second, third, and fourth fingerprinting multi-maps (111 to 114) in a storage device.

[0084] In the present invention, the first positioning unit (120) can measure the location of a user located within a plurality of grids (10), which are specific areas of an indoor place that match the fingerprinting map, and this step may be an online step.

[0085] In addition, the first positioning unit (120) can scan one or more smart devices located in a specific area of ​​an indoor location using the Wi-Fi function of a smart device (e.g., a smartphone) equipped by the user.

[0086] As a specific example, the first positioning unit (120) can extract Wi-Fi scan information (121) included in the identification information of the smart device when the smart device uses (or connects to) a Wi-Fi channel provided by a plurality of wireless routers through a Wi-Fi function.

[0087] In the present invention, Wi-Fi scan information (121) may be information for identifying the Wi-Fi channel used by the smart device and may be stored in the storage device of the first positioning unit (120).

[0088] And the first positioning unit (120) can scan a Bluetooth device using the Bluetooth function of a Bluetooth device equipped by the user.

[0089] As a specific example, when the first positioning unit (120) receives the identification information of a Bluetooth device obtained by scanning a Bluetooth device through a BLE beacon via the Wi-Fi channel of a wireless router, it can extract the Bluetooth scan information (122) included in the identification information of the Bluetooth device.

[0090] In the present invention, Bluetooth scan information (122) may be information for identifying a Bluetooth device in use within an area of ​​a specific indoor location that matches a fingerprint map, and may be stored in a storage device of the first positioning unit (120).

[0091] In the present invention, the first positioning unit (120) receives a train schedule from the station management system in the same way as the fingerprinting multi-map construction unit (110), and can extract time information (123) from the train schedule.

[0092] In the present invention, the time information (123) may be information about the time when a train enters the platform according to the train schedule, and may be stored in the storage device of the first positioning unit (120).

[0093] In the present invention, the first positioning unit (120) can calculate the gate congestion (124) of the station in the same manner as the fingerprinting multi-map construction unit (110), and the gate congestion (124) of the station can be stored in the storage device of the first positioning unit (120).

[0094] In the present invention, Wi-Fi scan information (121), Bluetooth scan information (122), time information (123), and the gate congestion level (124) of the station may be environmental variable information stored in real time in the storage device of the first positioning unit (120).

[0095] Accordingly, the following environmental variable information is preferably understood as Wi-Fi scan information (121), Bluetooth scan information (122), time information (123), and the gate congestion level (124) of the station.

[0096] In the present invention, the first positioning unit (120) can measure the user's location based on reflecting it in the fingerprinting multi-map (111-114).

[0097] In the present invention, various positioning methods (S120) for the first positioning unit (120) to measure the user's position are as shown in FIGS. 6 to 9.

[0098] FIG. 6 is a diagram illustrating the process of a method utilizing a multi-map combination in fingerprinting positioning based on an environment weight map-based positioning method, FIG. 7 is a diagram illustrating the detailed process of the weight calculation step illustrated in FIG. 6, FIG. 8 is a diagram illustrating the process of a method utilizing a multi-map combination in fingerprinting positioning based on a distance sum-based positioning method, and FIG. 9 is a diagram illustrating the process of a method utilizing a multi-map combination in fingerprinting positioning based on a judgment result aggregation positioning method.

[0099] Referring to FIG. 6, the first positioning unit (120) can measure the location of a user based on correcting (or reconstructing) fingerprinting multi-maps (111~114) provided from a server in an environment weight map-based positioning method (S121).

[0100] First, the first positioning unit (120) receives fingerprinting multi-maps (111~114) through a server, and can select (or extract) fingerprinting maps (111-n, 112-n, 113-n, 114-n) for each environment variable to reflect environment variable information (121~124) from the fingerprinting multi-maps (111~114) (S121a).

[0101] In the above step (S121a), the fingerprinting map (111-n) at the time of Wi-Fi channel interference when constructing the first fingerprinting multi-map (111) may be a fingerprinting map suitable for reflecting Wi-Fi scan information (121).

[0102] In the above step (S121a), the Bluetooth congestion fingerprinting map (112-n) that constructs the second fingerprinting multi-map (112) may be a fingerprinting map suitable for reflecting Bluetooth scan information (122).

[0103] In the above step (S121a), the train entry position fingerprinting map (113-n) that constructs the third fingerprinting multi-map (113) may be a fingerprinting map suitable for reflecting time information (123).

[0104] In the above step (S121a), the congestion fingerprinting map (114-n) for constructing the fourth fingerprinting multi-map (114) may be a fingerprinting map suitable for reflecting the gate congestion (124) of the history.

[0105] Afterwards, the first positioning unit (120) can calculate the weights of the environmental variable information (121~124) (S121b), and the detailed process of the weight calculation step (S121b) is as shown in FIG. 7.

[0106] Referring to FIG. 7, the first positioning unit (120) calculates the standard deviation relative to the average of the fingerprinting maps for each selected environmental variable using environmental variable information (111~114) (S1211), calculates the sum of the standard deviations by summing the calculated standard deviations of the fingerprinting maps (S1212), and can calculate the weight of the environmental variable information based on dividing the standard deviations of the fingerprinting maps for each environmental variable by the sum of the standard deviations (S1213).

[0107] That is, in the above step (S1213), the weight can be calculated in proportion to the standard deviation, and for each environmental variable information (121~124), an appropriate level of interference is set in preparation for a situation where interference is abnormally (exceptionally) large, and when this is exceeded, the weight is calculated as '0' to achieve a reduction in performance degradation.

[0108] After the above weight calculation step (S121b), the first positioning unit (120) can correct (or reconstruct) the fingerprinting maps for each environment variable (111-n, 112-n, 113-n, 114-n) into respective environment weight-based fingerprinting maps by multiplying the weights by the environment variable fingerprinting maps (S121c) (S121d).

[0109] In the above step (S121c), the weight is a positive number greater than '0', and it is most preferable that the weight be '1'.

[0110] After the above step (S121d), the first positioning unit (120) calculates the location of a user located within a specific indoor location area using the RSSI value of a smart device received through a plurality of wireless routers, and can calculate the Euclidean distance between the user location and each environment weight-based fingerprinting map (S121e).

[0111] After the above step (S121e), the first positioning unit (120) can measure the user's location based on the Euclidean distance from the user's location calculated in the environment weighted fingerprinting map among each environment weighted fingerprinting map, where the Euclidean distance from the user's location is calculated as the minimum value (S121f).

[0112] Referring to FIG. 8, the first positioning unit (120) can measure the location of a user based on correcting (or reconstructing) fingerprinting multi-maps (111~114) provided from a server in a positioning method (S122) based on a distance sum-based positioning method.

[0113] First, the first positioning unit (120) receives fingerprinting multi-maps (111~114) through a server, and can select (or extract) fingerprinting maps (111-n, 112-n, 113-n, 114-n) to which environment variable information (121~124) is reflected from the fingerprinting multi-maps (111~114) (S122a).

[0114] After the above step (S122a), the first positioning unit (120) calculates the location of a user located within a specific indoor location area using the RSSI value of a smart device transmitted through a plurality of wireless routers, and can calculate the Euclidean distance between the user location and the fingerprinting map (111-n, 112-n, 113-n, 114-n) for each environment variable (S122b).

[0115] After the above step (S122a), the first positioning unit (120) calculates the sum of Euclidean distances between the user location and the fingerprinting maps (111-n, 112-n, 113-n, 114-n) for each environment variable, and can measure the user's location based on the Euclidean distance with the user location calculated through the environment variable fingerprinting map that has the smallest Euclidean distance value among the environment variable fingerprinting maps (111-n, 112-n, 113-n, 114-n) (S122c).

[0116] In the above step (S122c), the first positioning unit (120) can calculate the Euclidean distance value by multiplying the weights calculated based on the environment variable information (121~124). Since the method of calculating the weights is the same as the weight calculation step (S121b) of the environment weight map-based positioning method described above, the explanation thereof will be omitted for convenience.

[0117] Referring to FIG. 9, the first positioning unit (120) can measure the location of a user based on correcting (or reconstructing) the fingerprinting multi-maps (111~114) provided by the server in a positioning method (S123) based on a judgment result aggregation positioning method.

[0118] First, the first positioning unit (120) receives fingerprinting multi-maps (111~114) through a server, and can select (or extract) fingerprinting maps (111-n, 112-n, 113-n, 114-n) suitable for environment variable information (121~124) from the fingerprinting multi-maps (111~114) (S123a).

[0119] After the above step (S123a), the first positioning unit (120) calculates the location of a user located within a specific indoor location area using the RSSI value of a smart device transmitted through a plurality of wireless routers, and can calculate the Euclidean distance between the user location and the fingerprinting map (111-n, 112-n, 113-n, 114-n) for each environmental variable (S123b).

[0120] After the above step (S123b), the first positioning unit (120) measures the user's location based on the Euclidean distance calculated as the minimum value among the Euclidean distances between the calculated user location and the fingerprinting maps (111-n, 112-n, 113-n, 114-n) for each environment variable (S123c), and can determine the final user's location by integrating these user positioning results and determining the point most frequently positioned as the user's location (S123d).

[0121] In the above step (S123d), if the first positioning unit (120) needs to determine the location of the end user using two or more fingerprinting maps that have the same minimum Euclidean distance from the user location, it can determine the location of the end user using the fingerprinting map that has a larger number of received RSSI values ​​of the BLE beacon.

[0122] In the present invention, the first positioning unit (120) can measure the location of a user using at least one of the following methods: the environment weight map-based positioning method shown in FIG. 6, the distance sum-based positioning method shown in FIG. 8, and the judgment result aggregation positioning method shown in FIG. 9.

[0124] Device and method utilizing multi-map combination for deep learning positioning

[0125] FIG. 10 is a block diagram schematically illustrating the components of a device utilizing a multi-map combination for deep learning positioning according to another embodiment of the present invention, and FIG. 11 is a diagram illustrating an example of a grid image input to a second positioning unit shown in FIG. 10.

[0126] Referring to FIG. 10, a device (200) utilizing a multi-map combination for deep learning positioning according to another embodiment of the present invention includes an image conversion unit (210) and a second positioning unit (220).

[0127] A device (200) utilizing a multi-map combination for such deep learning positioning is preferably understood as a second positioning device (200) that positions a user based on a map which is an input value of a deep learning model.

[0128] That is, it is preferable to understand that the device utilizing a multi-tamp combination for fingerprinting and deep learning positioning of the present invention includes a fingerprinting multi-map construction unit (110) and a first positioning unit (120), an image conversion unit (210), and a positioning unit (220).

[0129] Additionally, the second positioning device (200) receives environmental variable information (121~124) from the first positioning device (100) described above in order to measure the user's location.

[0130] That is, it is preferable for the second positioning device (200) to operate together with the first positioning device (100) to receive environmental variable information (121~124), and accordingly, the present invention can measure the user's location by utilizing a multi-map combination for fingerprinting and deep learning positioning.

[0131] In the present invention, the image conversion unit (210) receives the RSSI value of a BLE beacon and can generate an image based on the received RSSI value of the BLE beacon.

[0132] As a specific example, the image conversion unit (210) can assign a color according to the RSSI value of the BLE beacon and convert (or generate) the RSSI value of the colored BLE beacon into an image that matches a plurality of grids (10) placed in a specific area of ​​an indoor place as shown in FIG. 11.

[0133] In the present invention, the image generated from the image conversion unit (210) is preferably understood as a deep learning map for measuring the user's location, and the deep learning map can be transmitted to the second positioning unit (220) after being produced from the image conversion unit (210) so as to be used as an input value for the second positioning unit (220).

[0134] Additionally, when the deep learning map is produced from the image conversion unit (210) for a certain period of time, it can be constructed as a deep learning multi-map (215) composed of multiple deep learning maps.

[0135] In the present invention, the deep learning multi-map (215) can be constructed through a plurality of deep learning maps suitable for reflecting environment variable information (121~124).

[0136] In the present invention, the second positioning unit (220) can be loaded after a CNN model, which is one type of deep learning model, has been trained and verified.

[0137] That is, the deep learning multi-map (215) can be used as an input value for the CNN model mounted on the second positioning unit (220).

[0138] In the present invention, the second positioning unit (220) is not limited to having a CNN model installed, but may be installed after an RNN model has been trained and verified.

[0139] At this time, the image conversion unit (210) can construct a deep learning multi-map (215) for a certain period of time as time series data that is the input value of the RNN model and then transmit it to the second positioning unit (220).

[0140] In the present invention, the second positioning unit (220) can measure the user's location by reflecting the weights of the environment variable information (121~124) in the CNN model or RNN model into which the input value is input.

[0141] In the present invention, various positioning methods (S220) for the second positioning unit (220) to measure the user's location are as shown in FIGS. 12 and FIGS. 13.

[0142] FIG. 12 is a diagram illustrating the process of a method utilizing a multi-map combination for deep learning positioning based on deep learning positioning results, and FIG. 13 is a diagram illustrating the process of a method utilizing a multi-map combination for map-based deep learning positioning.

[0143] Referring to FIG. 12, the second positioning unit (220) receives a deep learning multi-map (215) from the image conversion unit (210) in the deep learning positioning method (S221) based on the deep learning positioning result, selects (or extracts) a deep learning map (221-n, 222-n, 223-n, 224-n) for each environment variable to reflect environment variable information (121~124) from the deep learning multi-map (215), and can measure the user's location using the deep learning map (221-n, 222-n, 223-n, 224-n) for each environment variable (S221a).

[0144] In the above step (S221a), the second positioning unit (220) obtains deep learning positioning result data (225a, 225b, 225c, 225d) by reflecting environment variable information (121~124) in the deep learning maps (221-n, 222-n, 223-n, 224-n) by environment variable.

[0145] Afterwards, the second positioning unit (220) receives the weights of the environmental variable information (121~124) calculated through the detailed process of the weight calculation step (S121b) described above from the first positioning device (100), and can distinguish the weights of the environmental variable information (121~124) to be reflected in each deep learning positioning result data (225a, 225b, 225c, 225d) based on the environmental variable information (121~124) reflected in the deep learning positioning result data (225a, 225b, 225c, 225d) (S221b).

[0146] After the above step (S221b), the second positioning unit (220) can reconstruct the deep learning positioning result data (225a, 225b, 225c, 225d) by multiplying the weights of environmental variable information (121~124) that are desirable to be reflected in the deep learning positioning result data (225a, 225b, 225c, 225d) for each environmental variable (S221c), and finally measure the user's location based on the reconstructed deep learning positioning result data (225a, 225b, 225c, 225d) for each environmental variable (S221d).

[0147] In the above step (S221d), it is desirable that the final measured user location differs from the user location measured using each deep learning map (221-n, 222-n, 223-n, 224-n) as the weights of the environment variable information (121~124) are reflected.

[0148] Referring to FIG. 13, in a map-based deep learning positioning method (S222), the second positioning unit (220) receives a deep learning multi-map (215) from an image conversion unit (210), selects (or extracts) a deep learning map (221-n, 222-n, 223-n, 224-n) for each environment variable to reflect environment variable information (121~124) from the deep learning multi-map (215), reflects the environment variable information (121~124), receives the weights of the environment variable information (121~124) calculated through the detailed process of the weight calculation step (S121b) described above from the first positioning device (100), and the environment variable reflected in the deep learning map (221-n, 222-n, 223-n, 224-n) Based on the information (121~124), the weights of the environment variable information (121~124) to be reflected in the deep learning map (221-n, 222-n, 223-n, 224-n) for each environment variable can be distinguished (S222a).

[0149] After the above step (S222a), the second positioning unit (220) can reconstruct the deep learning map (221-n, 222-n, 223-n, 224-n) by multiplying the weights of the environmental variable information (121~124) that is desirable to be reflected in the deep learning map (221-n, 222-n, 223-n, 224-n) for each environmental variable (S222b), and measure the user's location based on the reconstructed deep learning map (221-n, 222-n, 223-n, 224-n) for each environmental variable (S222c).

[0150] In the present invention, the second positioning unit (220) can measure the location of a user based on a deep learning positioning method based on deep learning positioning results shown in FIG. 12 or a map-based deep learning positioning method shown in FIG. 13.

[0152] Variant example

[0153] The device (100) utilizing a multi-map combination for fingerprinting positioning of the present invention may be modified into a device based on a modified example described below.

[0154] Variation Example 1)

[0155] Fingerprinting caps are significantly affected by surrounding users (people) of the BLE beacon, allowing for the detailed application of the influence of surrounding users.

[0156] In Variation Example 1, the fingerprinting multi-map construction unit (110) analyzes the influence of the number of users adjacent to a plurality of grids (10) that match the fingerprinting map, the distance and direction between each user, and the movement speed of each user, and can produce a fingerprinting map by reflecting these influences in the fingerprinting map production stage.

[0157] In Variation Example 1, when the first positioning unit (120) is linked with a system that recognizes the surroundings to measure the user's location, it can show optimized performance for user positioning by reflecting the influence of surrounding users.

[0158] Variation Example 2)

[0159] Depending on the model, smart devices (e.g., smartphones) measure different RSSI values ​​even in the same environment. These differences in RSSI values ​​occur because the sensitivity of the transmitting and receiving ends, antennas, and digital signal processing performance differ for each smart device.

[0160] In Variation Example 2, the fingerprinting multi-map construction unit (110) creates a fingerprinting map for each model of the smart device, and the first positioning unit (120) can measure the location of the user based on the fingerprinting map suitable for the model of the smart device.

[0161] At this time, creating fingerprint maps for all smart devices presents problems in terms of time and cost, so fingerprint maps are created for representative smart devices.

[0162] In addition, when the first positioning unit (120) measures the user's location based on a smart device for which a fingerprinting map has not been produced, it can measure the user's location using a combination of weights of the produced fingerprinting map.

[0163] Variation Example 3)

[0164] When mechanical equipment equipped with motors (e.g., air conditioners, generators, etc.) within the station is in operation, electromagnetic waves are generated due to the rotation of the motor, and radio interference caused by electromagnetic waves may be reflected when creating the fingerprinting map.

[0165] In Variation Example 3, the fingerprinting multi-map construction unit (110) measures the degree of radio interference in advance according to the location of mechanical equipment within the station and the operating status of mechanical equipment (weak, medium, strong), and can produce a fingerprinting map by reflecting the degree of radio interference.

[0166] In addition, the fingerprinting multi-map construction unit (110) creates the most suitable fingerprinting map according to the operating status of the mechanical equipment in conjunction with the history management system, and the first positioning unit (120) can measure the location of a user located in a specific area of ​​an indoor place based on this fingerprinting map.

[0167] Variation Example 4)

[0168] The first positioning unit (120) measured the user's location using the Euclidean distance between the user's location and the fingerprinting map, but in Variant Example 4, the user's location can be measured using not only the Euclidean distance but also the maximum value of the RSSI value, the K neighbor node utilization technique, the extended Kalman filter, etc.

[0170] Comparative example

[0171] In the following comparative example, an experiment was conducted to compare the positioning performance results of a fingerprinting map that does not reflect environmental variable information (121-124) and a fingerprinting map that reflects environmental variable information (121-124), and the positioning performance results of each fingerprinting map according to the comparative experiment are as shown in FIG. 14.

[0172] Figure 14 is a graph showing the positioning performance results of the fingerprinting map.

[0173] Referring to FIG. 14, the positioning performance of the fingerprinting map immediately after being produced by the fingerprinting multi-map construction unit (110) can be calculated as shown in the purple graph of FIG. 14, and the positioning performance of the fingerprinting map 5 days after the time when the positioning performance was calculated can be calculated as shown in the blue graph of FIG. 14.

[0174] In other words, the fingerprinting map resulted in a significant decrease in positioning performance as time passed after the comparative example was produced.

[0175] Meanwhile, the positioning performance of the fingerprinting map reflecting environmental variable information (121–124) through the environmental weight map-based positioning method shown in Fig. 6 can be calculated as shown in the red graph of Fig. 14, the positioning performance of the fingerprinting map reflecting environmental variable information (121–124) through the distance sum-based positioning method shown in Fig. 8 can be calculated as shown in the green graph of Fig. 14, and the positioning performance of the fingerprinting map reflecting environmental variable information (121–124) through the judgment result aggregation positioning method shown in Fig. 9 can be calculated as shown in the orange graph of Fig. 14.

[0176] In this way, it was found that the fingerprinting map reflecting environmental variable information (121–124) can maintain positioning performance similar to the fingerprinting map immediately after production (purple graph in Fig. 14) based on the reflection of environmental variables.

[0178] As described above, the detailed description of the preferred embodiments of the present invention disclosed is provided to enable those skilled in the art to implement and practice the present invention. Although the present invention has been described with reference to preferred embodiments, those skilled in the art will understand that various modifications and changes can be made to the present invention without departing from the scope of the invention. For example, those skilled in the art may utilize each configuration described in the embodiments described above in combination with one another. Accordingly, the present invention is not intended to be limited to the embodiments shown herein, but to be given the broadest scope consistent with the principles and novel features disclosed herein.

[0179] The present invention may be embodied in other specific forms without departing from the technical spirit and essential features of the invention. Accordingly, the above detailed description should not be interpreted restrictively in all respects but should be considered exemplary. The scope of the invention shall be determined by a reasonable interpretation of the appended claims, and all modifications within the equivalent scope of the invention are included within the scope of the invention. The invention is not intended to be limited to the embodiments shown herein, but to be given the broadest possible scope consistent with the principles and novel features disclosed herein. Furthermore, embodiments may be constructed by combining claims that are not explicitly related in the claims, or included as new claims through amendments made after filing. Explanation of the symbols

[0180] 10: Grid, 100: First positioning device, 110: Fingerprinting Multi-Map Construction Unit, 111: 1st Fingerprinting Multi-Map, 112: 2nd Fingerprinting Multi-Map, 113: 3rd Fingerprinting Multi-Map, 114: 4th fingerprinting multi-map, 120: 1st positioning unit, 121: Wi-Fi scan information, 122: Bluetooth scan information, 123: Time Information, 124: History Gate Congestion, 200: Second positioning device, 210: Image conversion unit, 215: Deep learning map, 220: Second positioning unit.

Claims

Claim 1 A method for utilizing a multi-map combination in fingerprinting and deep learning positioning, performed by a device utilizing a multi-map combination in fingerprinting and deep learning positioning, comprising: a) a step in which a first positioning device creates a plurality of fingerprinting maps that are matched with a plurality of grids and divided into grid units based on the RSSI values ​​of BLE beacons installed in a plurality of grids; b) a step in which the first positioning device constructs a fingerprinting multi-map by aggregating a plurality of fingerprinting maps according to environmental variables affecting positioning performance, and measures the location of a user located within an area within the plurality of grids based on a fingerprinting map reconstructed by reflecting the weights of environmental variable information in the fingerprinting maps according to environmental variables; c) a step in which a second positioning device assigns colors according to the RSSI values ​​of the BLE beacons and then creates a plurality of deep learning maps, which are images in a form that match the plurality based on the color-assigned RSSI values. and d) a step of the second positioning device assembling a plurality of deep learning maps for each environmental variable to construct a deep learning multi-map, and reconstructing the reconstructed deep learning map or deep learning positioning result data for each environmental variable by reflecting the weights of environmental variable information transmitted from the first positioning device to measure the location of the user; wherein the environmental variables include the usage status of Wi-Fi channels provided by wireless routers arranged at each grid unit length in the plurality of grids, the usage status of Bluetooth devices, the train entry status into the platform, and the flow of people at the station gates, characterized in that the method of utilizing a multi-map combination in fingerprinting and deep learning positioning. Claim 2 delete Claim 3 A method utilizing a multi-map combination for fingerprinting and deep learning positioning according to claim 1, wherein the environmental variable information includes Wi-Fi scan information for identifying a Wi-Fi channel used by a smart device equipped by the user, Bluetooth scan information for identifying a Bluetooth device in use within a plurality of grid areas, time information regarding the time when a train enters the platform on a train schedule received from a station management system, and the congestion level of the station calculated based on the flow of people at the station gate. Claim 4 In claim 1, the step b) comprises: b-1) the first positioning device extracting an environment variable-specific fingerprinting map to reflect the environment variable information from the fingerprinting multi-map; b-2) the first positioning device calculating the weight of the environment variable information; b-3) the first positioning device multiplying the environment variable-specific fingerprinting map by the weight of the environment variable information to reconstruct the environment variable-specific fingerprinting map into an environment weight-based fingerprinting map; b-4) the first positioning device calculating the user's location using the RSSI value of the smart device equipped by the user through wireless routers arranged at each grid unit length in the plurality of grids, and calculating the Euclidean distance between the user location and each environment weight-based fingerprinting map; and b-5) a step in which the first positioning device measures the location of the user based on the Euclidean distance to the user location calculated in the environment weighted fingerprinting map among the respective environment weighted fingerprinting maps, wherein the Euclidean distance to the user location is calculated as a minimum value; a method utilizing a multi-map combination in fingerprinting and deep learning positioning, characterized by including: Claim 5 A method for utilizing a multi-map combination in fingerprinting and deep learning positioning, wherein, in claim 4, step b-2) comprises: b-2-1) a step in which the first positioning device calculates the standard deviation relative to the average value of the fingerprinting map for each environment variable using the environment variable information; b-2-2) a step in which the first positioning device calculates the sum of the standard deviations of the fingerprinting map for each environment variable calculated in step b-2-1) to obtain the sum of the standard deviations; and b-2-3) a step in which the first positioning device calculates the weight of the environment variable information based on dividing the standard deviation of the fingerprinting map for each environment variable by the sum of the standard deviations. Claim 6 In claim 4, the above step b-3) is characterized by the weight of the environment variable information being a positive number greater than '0', and being '1' among the positive numbers, in a method utilizing a multi-map combination for fingerprinting and deep learning positioning. Claim 7 A method for utilizing a multi-map combination in fingerprinting and deep learning positioning according to claim 1, wherein step b) comprises: b-1) a step in which the first positioning device extracts an environment variable-specific fingerprinting map to which the environment variable information is reflected from the fingerprinting multi-map; b-2) a step in which the first positioning device calculates the location of the user using the RSSI value of the smart device equipped by the user through wireless routers arranged at grid unit lengths in the plurality of grids, and calculates the Euclidean distance between the user location and the environment variable-specific fingerprinting map; and b-3) a step in which the first positioning device calculates the sum of the Euclidean distances between the user location and the environment variable-specific fingerprinting map, and measures the user location based on the Euclidean distance with the user location calculated through the environment variable fingerprinting map with the smallest Euclidean distance value among the environment variable-specific fingerprinting maps. Claim 8 In claim 7, the above step b-3) is characterized by the first positioning device multiplying the fingerprinting map for each environment variable by the weight of the environment variable information to calculate the sum of the Euclidean distances between the user location and the fingerprinting map for each environment variable, in a method utilizing a multi-map combination for fingerprinting and deep learning positioning. Claim 9 A method utilizing a multi-map combination for fingerprinting and deep learning positioning, characterized in that, in claim 8, the weight of the environment variable information is a positive number greater than '0', and among the positive numbers, it is '1'. Claim 10 A method for utilizing a multi-map combination in fingerprinting and deep learning positioning, wherein, in claim 1, step b) comprises: b-1) a step in which the first positioning device extracts an environment variable-specific fingerprinting map to which the environment variable information is reflected from the fingerprinting multi-map; b-2) a step in which the first positioning device calculates the location of the user using the RSSI value of the smart device equipped by the user through wireless routers arranged at grid unit lengths in the plurality of grids, and calculates the Euclidean distance between the user location and the environment variable-specific fingerprinting map; b-3) a step in which the first positioning device measures the location of the user based on the Euclidean distance calculated as the minimum value among the Euclidean distances between the user location and the environment variable-specific fingerprinting map calculated in step b-2); and b-4) a step in which the first positioning device integrates the user positioning results of step b-3) and determines the point most frequently positioned as the user's location as the final user's location. Claim 11 A method utilizing a multi-map combination in fingerprinting and deep learning positioning, characterized in that, in claim 1, the deep learning multi-map is an input value of a CNN model or an RNN model mounted on the second positioning device. Claim 12 In claim 11, the above step d) comprises: d-1) a step in which the second positioning device extracts a deep learning map for each environment variable to which the environment variable information is to be reflected from the deep learning multi-map, and obtains deep learning positioning result data for each environment variable by reflecting the environment variable information in the deep learning map for each environment variable; d-2) a step in which the second positioning device receives weights of environment variable information for the deep learning map for each environment variable from the first positioning device, and distinguishes the weights of environment variable information to be reflected in the deep learning positioning result data for each environment variable based on the environment variable information reflected in the deep learning positioning result data for each environment variable; and d-3) a step in which the second positioning device multiplies the weight of the environment variable information separated in step d-2) to the deep learning positioning result data for each environment variable to reconstruct the deep learning positioning result data for each environment variable, and measures the location of a user based on the reconstructed deep learning positioning result data for each environment variable; characterized by including a method utilizing a multi-map combination in fingerprinting and deep learning positioning. Claim 13 In claim 11, the step d) comprises: d-1) a step in which the second positioning device extracts a deep learning map for each environment variable to be reflected from the deep learning multi-map, and then reflects the environment variable information in the deep learning map for each environment variable; d-2) a step in which the second positioning device receives weights of environment variable information for the deep learning map for each environment variable from the first positioning device, and distinguishes the weights of environment variable information to be reflected in the deep learning map for each environment variable based on the environment variable information reflected in the deep learning map for each environment variable; and d-3) a step in which the second positioning device multiplies the weights of environment variable information distinguished in step d-2) to the deep learning map for each environment variable to reconstruct the deep learning map for each environment variable, and measures the location of a user based on the reconstructed deep learning map for each environment variable; characterized in that the method of utilizing a multi-map combination in fingerprinting and deep learning positioning is further characterized by including: Claim 14 A method utilizing a multi-map combination in fingerprinting and deep learning positioning according to claim 1, characterized by measuring the location of the user using steps a) and b), or measuring the location of the user using steps a), b), c), and d). Claim 15 A first positioning device utilizing a multi-map combination for fingerprinting and deep learning positioning, comprising: generating multiple fingerprinting maps that are matched with multiple grids and divided into grid units based on RSSI values ​​of BLE beacons installed in multiple grids; constructing a fingerprinting multi-map by aggregating multiple fingerprinting maps according to environmental variables affecting positioning performance; and measuring the location of a user located within an area of ​​the multiple grids based on a fingerprinting map reconstructed by reflecting weights of environmental variable information in the fingerprinting maps according to environmental variables. A device utilizing a multi-map combination for fingerprinting and deep learning positioning, comprising: a second positioning device that measures the location of the user by reconstructing the deep learning map or deep learning positioning result data by environmental variables, by reflecting the weights of environmental variable information transmitted from the first positioning device; wherein the environmental variables include the usage status of Wi-Fi channels provided by wireless routers placed at each grid unit length in the plurality of grids, the usage status of Bluetooth devices, the train entry status into the platform, and the flow of people at the station gate.

Images