Locating method

Abstract

To provide a locating method that enables the precision of identifying locations of objects to be located to be improved.SOLUTION: A radio wave receiver installed in a vehicle receives a radio wave transmitted from each radio wave transmitter (ST1). Using a received radio wave correction amount, the radio wave intensity is attenuated and amplified with respect to the intensity of the received radio wave, and the radio wave intensity is artificially increased or decreased (ST2). The calculated intensity of the radio wave from each radio wave transmitter is compared with a fingerprint to identify a grid having the most similar pattern of radio wave intensity as a tentative current position (ST3). An area that contains the identified grid number is identified (ST4). A position of the vehicle is identified using a machine learning model created for the identified area (ST5).SELECTED DRAWING: Figure 3

Description

[Technical Field]

[0001] The present invention relates to a location determination method, and more particularly to an improvement for increasing the accuracy of determining the location of a location-determining object. [Background technology]

[0002] Conventionally, GPS (Global Positioning System) and the like have been used as a technology for determining the position of an object (e.g., a vehicle) outdoors. However, a fingerprint-based location determination method using Bluetooth is known as a technology for determining a position indoors (e.g., determining the position of a logistics truck in a logistics warehouse) where GPS and the like cannot be used (see Patent Document 1).

[0003] A technology for identifying a location using the fingerprint method involves installing multiple radio wave transmitters (e.g., Wi-Fi access points) within a target space, dividing the target space into multiple grids, and measuring the radio wave strength (RSSI) from each radio wave transmitter at each point in each grid to create an RSSI fingerprint (RSSI map).Then, by comparing the radio wave strength (individual radio wave strength identified for each radio wave transmitter) from each radio wave transmitter received by a radio wave receiver mounted on an object (object to be located) with the fingerprint, the grid of the fingerprint with the most similar pattern of radio wave strength received from each radio wave transmitter is identified (positioned) as the grid where the object to be located is currently located. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] Japanese Patent Application Laid-Open No. 2004-500566 Summary of the Invention [Problem to be solved by the invention]

[0005] However, conventional fingerprint-based location identification technology was limited to the accuracy of determining which grid the location of the object to be located falls on, and it was not possible to identify the location with any greater accuracy.

[0006] The present invention has been made in view of the above points, and an object of the present invention is to provide a position specifying method that can improve the accuracy of specifying the position of an object to be located. [Means for solving the problem]

[0007] The solution of the present invention for achieving the above object is based on a location identification method for identifying the location of a location-identifying object using a fingerprint, and the location identification method includes the steps of: individually creating a machine learning model for each of a plurality of pre-divided areas to predict coordinate values ​​based on the strength of radio waves received from a plurality of radio wave transmitters; identifying an area among the plurality of areas in which the location-identifying object is located based on the strength of radio waves received by a radio wave receiver provided on the location-identifying object; and identifying the coordinate values ​​of the location-identifying object within the identified area using the machine learning model created for the identified area.

[0008] By this specification, the area in which the location-specific object is located is identified based on the strength of the radio waves received by a radio wave receiver provided on the location-specific object, and then the coordinate values ​​of the location-specific object within that area (the position of the location-specific object within the area) are identified using a machine learning model created for that area. Therefore, compared to identifying the position of the location-specific object using a fingerprint method without dividing the space into multiple areas (identifying the coordinate values ​​of the location-specific object using a single model targeting the entire space), the accuracy of identifying the position of the location-specific object can be improved by targeting a narrowed area and identifying the coordinate values ​​(position) of the location-specific object using a machine learning model dedicated to that area.

[0009] In addition, in the step of identifying the area, the grid in which the object to be located is located is identified based on fingerprints created for a plurality of grids partitioned in each of the plurality of areas, and the area to which the identified grid belongs is identified as the area in which the object to be located is located.

[0010] This allows for highly accurate narrowing down of the area when selecting a machine learning model, suppressing erroneous recognition of the area and improving the accuracy of identifying the position of the object to be located.

[0011] In addition, in the step of identifying the area, the similarity of the pattern of the strength of the radio waves received from each radio wave transmitter is calculated for each grid of the fingerprint, and the grid in which the object to be located is located is determined by performing a weighted average process of the similarity, and the area containing the grid is identified as the area in which the object to be located is located.

[0012] This makes it possible to effectively utilize existing fingerprint-based location identification technology to identify the area in which the location-identification object is located, thereby effectively identifying the area in which the location-identification object is located.

[0013] In addition, in the step of identifying the area, the strength of the radio waves received from the radio wave transmitter is corrected to the strength of the radio waves that would be received if the external factors were not present, using a received radio wave correction amount based on external factors that affect the radio wave reception state of the radio wave receiver, and then the similarity is calculated.

[0014] This makes it possible to eliminate adverse effects on the accuracy of position identification caused by external factors, and to improve the accuracy of identifying the position of the object to be positioned. [Effects of the Invention]

[0015] In this invention, a machine learning model is individually created for each of a plurality of pre-divided areas, the area in which the location-specific object is located is identified based on the strength of the radio waves received by the radio wave receiver, and the coordinate values ​​of the location-specific object are identified using the machine learning model created for this identified area. Therefore, compared to identifying the position of the location-specific object using a fingerprint method without dividing the space into a plurality of areas, the accuracy of identifying the position of the location-specific object can be improved by targeting a narrowed-down area and identifying the coordinate values ​​of the location-specific object using a machine learning model dedicated to that area. [Brief explanation of the drawings]

[0016] [Figure 1] FIG. 2 is a plan view showing each grid partitioned within a logistics warehouse according to an embodiment. [Figure 2] FIG. 2 is a plan view showing each area partitioned within a logistics warehouse. [Figure 3] FIG. 10 is a flowchart showing a procedure for identifying a location. [Figure 4] 1 and shows an example of the current position of a vehicle in a logistics warehouse. [Figure 5] 1 and illustrates an example of a vehicle position tentatively identified in the area identification process using virtual lines. [Figure 6] FIG. 10 is a plan view of area A showing an example of the position of the vehicle identified in the position identification process. DETAILED DESCRIPTION OF THE INVENTION

[0017] Hereinafter, an embodiment of the present invention will be described with reference to the drawings. In this embodiment, the present invention will be described as being applied to a location identification method for identifying the location of a vehicle (such as a logistics truck) traveling within a logistics warehouse.

[0018] -Explanation of each grid and area- The logistics warehouse is divided into multiple grids and multiple areas. FIG. 1 is a plan view showing each grid divided into areas within the logistics warehouse 1. FIG. 2 is a plan view showing each area divided into areas within the logistics warehouse 1. For example, the length of the logistics warehouse 1 in the X direction is 100 m, and the length in the Y direction is 20 m. The size of the logistics warehouse 1 is not limited to this.

[0019] As shown in FIG. 1, in this embodiment, the logistics warehouse 1 is divided into a total of 80 grids, 20 in the X direction and 4 in the Y direction. In the following description, for convenience, a number (grid number) is assigned to each grid. The numbers 1 to 80 written on each grid in FIG. 1 are grid numbers assigned to each grid. In this embodiment, the dimensions of each grid in the X direction and the Y direction are approximately the same, but these dimensions do not necessarily have to be the same. Furthermore, the shape of each grid in a plan view does not necessarily have to be rectangular.

[0020] As shown in FIG. 2, in this embodiment, the logistics warehouse 1 is divided into three areas. In this embodiment, the area including the grids with grid numbers 1 to 8, 21 to 28, 41 to 48, and 61 to 68 (32 grids) will be referred to as Area A. Furthermore, the area including the grids with grid numbers 9 to 16, 29 to 35, 49 to 56, and 69 to 76 (32 grids) will be referred to as Area B. Furthermore, the area including the grids with grid numbers 17 to 20, 37 to 40, 57 to 60, and 77 to 80 (16 grids) will be referred to as Area C. The letters A to C written in each area in FIG. 2 indicate the area assigned to each area. The number of areas and the number of grids included in each of Areas A to C are not limited to these. Furthermore, the shape of each area in a plan view does not necessarily have to be rectangular.

[0021] Radio wave transmitters 2A to 2L are installed at multiple locations within the logistics warehouse 1. In this embodiment, one radio wave transmitter 2A (2B to 2L) is attached to each of the pillars 3, 3, ... erected within the logistics warehouse 1, and radio waves are transmitted from each of the radio wave transmitters 2A to 2L into the space within the logistics warehouse 1. In this embodiment, the strength of the radio waves transmitted from each of the radio wave transmitters 2A to 2L is set to be the same. However, the strength of the radio waves transmitted from each of the radio wave transmitters 2A to 2L may differ from one another.

[0022] In the logistics warehouse 1, for each grid (for each grid with grid numbers 1 to 80), the strength of the radio waves from each of the radio wave transmitters 2A to 2L is measured at each point, and a radio wave strength fingerprint is created that indicates the strength of the individual radio waves identified for each of the radio wave transmitters 2A to 2L at each of the points. As is well known, this is done to identify the grid on which the vehicle V is located, among the grids (for each of the grid numbers 1 to 80) of the fingerprint, based on the similarity of the pattern of the strength of the radio waves received from each of the radio wave transmitters 2A to 2L, by comparing the strength of the radio waves from each of the radio wave transmitters 2A to 2L received by a radio wave receiver mounted on the vehicle V (see FIG. 4).

[0023] In this embodiment, an individual machine learning model is created for each of areas A to C. That is, for each of areas A to C, training data (training data on the relationship between coordinate values ​​within a range corresponding to each of areas A to C and the strength of radio waves received from each of the radio wave transmitters 2A to 2L) is acquired and learned (learning the strength of radio waves received from each of the radio wave transmitters 2A to 2L), and a machine learning model that predicts coordinate values ​​from the strength of radio waves received from each of the radio wave transmitters 2A to 2L is created. That is, a machine learning model that is dedicated to each of areas A, B, and C and is different from each other is created. Note that this dedicated machine learning model for each area may be created by acquiring and learning training data including the strength of radio waves received from multiple radio wave transmitters located within the respective area (e.g., radio wave transmitters 2A, 2B, 2C, 2G, 2H, and 2I in the case of area A), or may be created by acquiring and learning training data including the strength of radio waves received from all of the radio wave transmitters 2A to 2L.

[0024] This machine learning model is designed to not only identify the grid contained in the relevant area based on the strength of the radio waves received from each of the radio wave transmitters 2A to 2L, but also to predict the position of vehicle V within that grid (more specifically, the position within the grid of the radio wave receiver installed on vehicle V).

[0025] The vehicle V is also equipped with a radio wave receiver (not shown) capable of receiving radio waves transmitted from each of the radio wave transmitters 2A to 2L. The radio wave reception state of this radio wave receiver is affected by the location of the loading platform or luggage on the vehicle V. For example, if the radio wave receiver is installed near the loading platform of the vehicle V, the effect of the loading platform impeding radio wave reception will vary depending on the relative positions of the radio wave receiver and the loading platform and the size of the loading platform (including the state of luggage). In other words, if a loading platform or luggage is located in the space between the radio wave receiver installed in the vehicle V and each of the radio wave transmitters 2A to 2L, radio wave reception will be impeded. However, the state of impeding radio wave reception at the radio wave receiver differs for each of the radio wave transmitters 2A to 2L, such as in combinations where the loading platform or luggage is located between the radio wave receiver and each of the radio wave transmitters 2A to 2L, combinations where the loading platform or luggage is not located, and combinations where the loading platform or luggage is located in only a portion of the space.

[0026] Vehicle V stores, as individual information, information related to the influence of external factors such as the loading platform on the radio wave reception state as a received radio wave correction amount corresponding to each radio wave received from each radio wave transmitter 2A-2L. This received radio wave correction amount is used to correct the strength of the radio waves actually received from each radio wave transmitter 2A-2L (the strength of the radio waves affected by external factors) to the strength of the radio waves that would be received if external factors such as the loading platform were not present. The greater the influence of external factors on the radio wave reception, the greater the correction amount is set. Furthermore, this received radio wave correction amount is determined in advance for each vehicle (for each vehicle type and each state of luggage loaded in the vehicle) through experiments and simulations. Specifically, this received radio wave correction amount is determined by calculating the amount of attenuation and amplification of the radio wave strength due to diffraction and reflection of the radio waves caused by external factors such as the loading platform.

[0027] It is assumed that the load status of the cargo on the loading platform of the vehicle V will change as cargo is loaded and unloaded at the logistics warehouse 1. For this reason, it is preferable to change the received radio wave correction amount in accordance with the loading and unloading of cargo. In this case, information on the load status in accordance with the loading and unloading of cargo may be acquired by an input operation by the driver of the vehicle V, or the vehicle V may be equipped with sensors capable of acquiring information on the load status of cargo, and the received radio wave correction amount may be changed based on the information acquired by these sensors.

[0028] The various pieces of information, such as the above-mentioned fingerprint, machine learning model, and received radio wave correction amount, may be stored in a position identification device (not shown) mounted on the vehicle V, or may be stored in a management server (not shown) that manages the traveling positions of one or more vehicles V traveling within the logistics warehouse 1. Also, some of the various pieces of information may be stored in the position identification device mounted on the vehicle V, and other information may be stored in the management server, so that the various pieces of information can be shared by mutual communication between the vehicle V and the management server.

[0029] -Location identification processing- Next, the location specification process in this embodiment will be described.

[0030] 3 is a flowchart showing the procedure for identifying the location. Before the location is identified, the machine learning models are created for each of Area A, Area B, and Area C (this corresponds to the step of creating a machine learning model for each of the pre-divided areas based on the strength of radio waves received from each of the radio wave transmitters in the present invention), and the received radio wave correction amount for the vehicle V is calculated.

[0031] Here, an example will be described in which the actual position of vehicle V is located in the lower left portion of grid number 27 in FIG. 4 and this position is identified with high accuracy.

[0032] First, in step ST1, the radio wave receiver mounted on the vehicle V receives radio waves transmitted from each of the radio wave transmitters 2A to 2L. As a result, the strength of the radio waves received from each of the radio wave transmitters 2A to 2L is measured for each of the radio wave transmitters 2A to 2L.

[0033] Then, in step ST2, the received radio wave correction amount is used to attenuate or amplify the radio wave strength received in step ST1, thereby artificially increasing or decreasing the radio wave strength. In other words, the radio wave strength is calculated by excluding the influence of external factors such as the loading platform on the radio wave reception state, and is used to identify the location of vehicle V without being influenced by external factors.

[0034] In step ST3, the strength of the radio waves from each radio wave transmitter 2A to 2L calculated in step ST2 (the strength of the radio waves calculated using the received radio wave correction amount) is compared with the fingerprint created for the entire logistics warehouse 1, and the grid (grid numbers 1 to 80) of the fingerprint that has the most similar pattern of the strength of the radio waves received from each radio wave transmitter 2A to 2L is identified as the tentative current location.

[0035] In this case, the similarity is calculated for each grid in order of the grid with the highest similarity in the radio wave strength pattern, such as the first candidate grid, the second candidate grid, and the third candidate grid, and a weighted average process is performed on these grids according to the similarity to identify the tentative current position of vehicle V.

[0036] Figure 5 shows the case where the position of vehicle V identified in this manner is the lower right part of the figure at grid number 26. For example, if the first candidate grid is grid number 27, the second candidate grid is grid number 26, and the third candidate grid is grid number 5, the tentative current position of vehicle V may be identified as shown in Figure 5. In Figure 5, vehicle V is shown by a virtual line because the identified position of vehicle V is a tentative position.

[0037] In step ST4, the area that includes the grid number identified in step ST3 (number 26 in the above process) is identified from among areas A to C. In the above process, the tentative current position was identified as grid number 26, so area A is identified here.

[0038] Since the above processing is performed in steps ST3 and ST4, these processings constitute area identification processing, which corresponds to the "step of identifying the area among multiple areas in which the object to be located is located based on the strength of the radio waves received by the radio wave receiver provided on the object to be located" in this invention.

[0039] In step ST5, the position of vehicle V is identified using a machine learning model created for the area identified by the area identification process. In this position identification process, as in step ST2 described above, the received radio wave correction amount is used to attenuate and amplify the strength of the received radio waves, thereby artificially increasing or decreasing the radio wave strength and calculating the radio wave strength that eliminates the influence of external factors on the radio wave reception state, such as the loading platform. Note that information on the radio wave strength calculated in step ST2 may be stored and used as is.

[0040] The radio wave intensity calculated in this manner is applied to a machine learning model (a machine learning model created for the area identified by the area identification process), thereby identifying the current position of vehicle V within this area. The processing in step ST5 corresponds to the "step of identifying the coordinate values ​​of a position-identifying object within the identified area using a machine learning model created for the identified area" in this invention.

[0041] Since area A has been identified in the above process, the machine learning model created for area A is used to identify the position of vehicle V. Figure 6 shows the case where the position of vehicle V identified in this way for area A is in the lower left part of the figure at grid number 27.

[0042] -Effects of the embodiment- As described above, in this embodiment, the area in which the vehicle V is located is identified from the strength of radio waves received by a radio wave receiver mounted on the vehicle V, and then the coordinate values ​​of the vehicle V within that area are identified using a machine learning model created for that area. In other words, compared to identifying the coordinate values ​​of the vehicle V using a fingerprint method targeting an entire space including multiple areas (identifying the coordinate values ​​of the vehicle V using a single model targeting the entire space), the accuracy of identifying the position of the vehicle V can be improved by targeting a narrowed down area and identifying the coordinate values ​​of the vehicle V using a machine learning model dedicated to that area.

[0043] Furthermore, in this embodiment, the grid in which the vehicle V is located is identified based on fingerprints created for a plurality of grids partitioned in each of a plurality of areas, and the area to which the identified grid belongs is identified as the area in which the vehicle V is located. This makes it possible to suppress erroneous recognition of the area in which the vehicle V is located, thereby improving the accuracy of identifying the location of the vehicle V.

[0044] Furthermore, in this embodiment, the similarity of the intensity patterns of the radio waves received from each of the radio wave transmitters 2A to 2L is calculated for each grid of the fingerprint, and a weighted average process of the similarity is performed to determine the grid where the vehicle V is located, and the area containing the grid is identified as the area where the vehicle V is located. Therefore, it is possible to effectively use existing fingerprint-based location identification technology to identify the area where the vehicle V is located, and the area where the vehicle V is located can be identified effectively.

[0045] Furthermore, in this embodiment, the strength of the radio waves received from the radio wave transmitter is corrected to the strength of the radio waves that would be received if the external factors were not present, using a received radio wave correction amount based on external factors that affect the radio wave reception state of the radio wave receiver, and then the similarity is calculated. This makes it possible to eliminate adverse effects on the accuracy of location identification caused by external factors, and improve the accuracy of identifying the location of the vehicle V.

[0046] -Other embodiments- The present invention is not limited to the above-described embodiments, and all modifications and applications within the scope of the claims and equivalents thereto are possible.

[0047] For example, in the above embodiment, the present invention has been described as being applied as a position identification method for identifying the position of a vehicle V traveling within a logistics warehouse 1. The present invention is not limited to this, and can be applied to various cases where the position of a position-identifying object is identified indoors.

[0048] In the above embodiment, the object to be located is a logistics truck traveling within the logistics warehouse 1. However, the present invention is not limited to this and can also be applied to a case where the object to be located is a forklift traveling within the logistics warehouse 1. In this case, the reception state of the radio wave at the radio wave receiver is affected by the size of the package being transported by the forklift and the elevation position of the forks. Therefore, it is preferable to change the received radio wave correction amount based on information about the size of the package and the elevation position of the forks. In this case, the information about the size of the package and the elevation position of the forks may be acquired by an input operation by the forklift driver, or the forklift may be equipped with sensors capable of acquiring information about the size of the package and the elevation position of the forks, and the received radio wave correction amount may be changed based on the information acquired by the sensors.

[0049] In the above embodiment, the areas dividing the logistics warehouse 1 are area A and area B, which have the same surface area, and area C, which is smaller than area A and area B. However, the present invention is not limited to this, and all areas may have the same surface area, or all areas may have different surfaces.

[0050] In the above embodiment, the arrangement of the radio wave transmitters 2A to 2L installed in the logistics warehouse 1 is symmetrical with respect to the center position in the X direction of the logistics warehouse 1, and is symmetrical with respect to the center position in the Y direction of the logistics warehouse 1. The present invention is not limited to this, and the arrangement of the radio wave transmitters 2A to 2L in each direction may be asymmetrical. [Industrial Applicability]

[0051] The present invention is applicable to a position specifying method for specifying the position of a logistics truck traveling within a logistics warehouse. [Explanation of symbols]

[0052] 1 Logistics warehouse 2A~2L radio wave transmitter 1~80 grid Areas A to C V Vehicle (location-specific object)

Claims

1. A location determination method for determining the location of a location-determining object using a fingerprint, comprising: A step of individually creating a machine learning model for predicting coordinate values ​​for each of a plurality of pre-divided areas based on the strength of radio waves received from each of a plurality of radio wave transmitters; identifying an area in which the object to be located is located among the plurality of areas based on the intensity of radio waves received by a radio wave receiver provided in the object to be located; and identifying coordinate values ​​of the object to be located within the identified area using the machine learning model created for the identified area.

2. 2. The method of claim 1, A location identification method characterized in that in the area identification step, the grid in which the location-identification object is located is identified based on fingerprints created for multiple grids divided in each of the multiple areas, and the area to which the identified grid belongs is identified as the area in which the location-identification object is located.

3. 3. The location identification method according to claim 2, A location identification method characterized in that in the step of identifying the area, the similarity of the pattern of radio wave intensity received from each radio wave transmitter is calculated for each grid of the fingerprint, a weighted average process is performed to determine the grid in which the object to be located is located, and the area containing the grid is identified as the area in which the object to be located is located.

4. 4. The method of claim 3, A location identification method characterized in that in the step of identifying the area, the strength of the radio waves received from the radio wave transmitter is corrected to the strength of the radio waves that would be received if the external factors were not present, using a received radio wave correction amount based on external factors that affect the radio wave reception status of the radio wave receiver, and then the similarity is calculated.

Images