US-20260127395-A1 - Wireless Tag Direction Identification

US20260127395A1US 20260127395 A1US20260127395 A1US 20260127395A1US-20260127395-A1

Abstract

A method of locating a wireless tag via a computing device is described. At least one rotation sensor is used to determine a plurality of angle data of the computing device in the horizontal plane during motion of the computing device. A plurality of signal strength data of the wireless tag is processed during the motion of the computing device. A plurality of data points are created by correlating ones of the plurality of angle data with ones of the plurality of signal strength data. A non-linear regression machine learning model is trained with the plurality of data points. A visual indicator of a direction of the wireless tag is generated based on the training. A computer readable medium storing the instructions is also described.

Inventors

Bradley Abelman

Assignees

ZEBRA TECHNOLOGIES CORPORATION

Dates

Publication Date: 20260507
Application Date: 20250822

Claims (20)

1 . A method of locating a wireless tag via a computing device, the method comprising: using at least one rotation sensor to determine a plurality of angle data of the computing device in the horizontal plane during motion of the computing device; processing a plurality of signal strength data of the wireless tag during the motion of the computing device; creating a plurality of data points by correlating ones of the plurality of angle data with ones of the plurality of signal strength data; training a non-linear regression machine learning model with the plurality of data points; generating, at the computing device, a visual indicator of a direction of the wireless tag based on the training; and displaying the visual indicator of the direction of the wireless tag at the computing device.
2 . The method of claim 1 , wherein training the non-linear regression machine learning model comprises training a non-linear signal strength estimation function to determine coefficients that best fit the non-linear signal strength estimation function to the plurality of data points to define a representative curve.
3 . The method of claim 2 , wherein the training comprises gradient descent training.
4 . The method of claim 3 , further comprising determining a maximum value in the defined curve, the maximum value representing a direction of the wireless tag, wherein generating the visual indicator is based on the maximum value in the defined curve.
5 . The method of claim 4 , wherein the gradient descent training comprises: determining a compute error in the training of the non-linear signal strength estimation function; determining whether the compute error is within a predetermined range; and defining the representative curve as a curve that satisfies the non-linear signal strength estimation function when the compute error is within the predetermined range.
6 . The method of claim 5 , wherein the non-linear signal strength estimation function is y=c 0 +c 1 ×ln(c 2 +c 3 sin (x)+c 4 cos (x)), where y is the signal strength data and x is the angle data in the horizonal plane, wherein c 0 and c 1 are representative of the natural loss of RSSI as a function of distance, and c 2 , c 3 , and c 4 are representative of a peak angle where the wireless tag is expected to be located.
7 . The method of claim 4 , further comprising using the at least one rotation sensor to determine a plurality of angle data of the computing device in the vertical plane during motion of the computing device.
8 . The method of claim 7 , wherein the gradient descent training comprises minimizing the volume of the surface generated by the non-linear signal strength estimation function while limiting a square distance of points lying outside a boundary of the surface.
9 . The method of claim 7 , wherein the non-linear signal strength estimation function is R(θ, φ)=(c 1 +c 2 ln(1+c 3 +cos (θ−c 4 ))) cos (φ−c 5 ), where R(θ, φ) is the signal strength data, φ is the angle data in the horizonal plane, wherein c 1 is a minimum size of an RSSI bubble, c 2 is an eccentricity/radial variance of the RSSI bubble; c 3 is a rotational significance of the device, c 4 is a yaw direction of the wireless tag, and c 5 is a pitch direction of the wireless tag.
10 . The method of claim 1 , wherein the non-linear regression machine learning model is a Log-Trigonometric regression machine learning model.
11 . A non-transitory computer readable medium having stored thereon instructions for locating a wireless tag, the instructions when executed by a processor of a computing device, cause the processor to: use at least one rotation sensor to determine a plurality of angle data of the computing device in the horizontal plane during motion of the computing device; process a plurality of signal strength data of the wireless tag received at one or more antennas during the motion of the computing device; create a plurality of data points by correlating ones of the plurality of angle data with ones of the plurality of signal strength data; train a non-linear regression machine learning model with the plurality of data points; and generate, on a display of the computing device, a visual indicator of a direction of the wireless tag based on the training.
12 . The non-transitory computer readable medium of claim 11 , wherein training the non-linear regression machine learning model comprises training a non-linear signal strength estimation function to determine coefficients that best fit the non-linear signal strength estimation function to the plurality of angle data and the plurality of signal strength data to define a representative curve.
13 . The non-transitory computer readable medium of claim 12 , wherein the training comprises gradient descent training.
14 . The non-transitory computer readable medium of claim 13 , further comprising determining a maximum value in the defined curve, the maximum value representing a direction of the wireless tag, wherein generating the visual indicator is based on the maximum value in the defined curve.
15 . The non-transitory computer readable medium of claim 14 , wherein the gradient descent training comprises: determining a compute error in the training of the non-linear signal strength estimation function; determining whether the compute error is within an acceptable range; and defining the representative curve as a curve that satisfies the non-linear signal strength estimation function when the compute error is within the acceptable range.
16 . The non-transitory computer readable medium of claim 15 , wherein the non-linear signal strength estimation function is y=c 0 +c 1 ×ln(c 2 +c 3 sin (x)+c 4 cos (x)), where y is the signal strength data and x is the angle data in the horizonal plane, wherein c 0 and c 1 are representative of the natural loss of RSSI as a function of distance, and c 2 , c 3 , and c 4 are representative of a peak angle where the wireless tag is expected to be located.
17 . The non-transitory computer readable medium of claim 14 , further comprising using the at least one rotation sensor to determine a plurality of angle data of the computing device in the vertical plane during motion of the computing device.
18 . The non-transitory computer readable medium of claim 17 , wherein the gradient descent training comprises minimizing the volume of the surface generated by the non-linear signal strength estimation function while limiting a square distance of points lying outside a boundary of the surface.
19 . The non-transitory computer readable medium of claim 17 , wherein the non-linear signal strength estimation function is R(θ, φ)=(c 1 +c 2 ln(1+c 3 +cos (θ−c 4 ))) cos (φ−c 5 ), where R(θ, φ) is the signal strength data, φ is the angle data in the horizonal plane, wherein c 1 is a minimum size of an RSSI bubble, c 2 is an eccentricity/radial variance of the RSSI bubble; c 3 is a rotational significance of the device, c 4 is a yaw direction of the wireless tag, and c 5 is a pitch direction of the wireless tag.
20 . The non-transitory computer readable medium of claim 11 , wherein the non-linear regression machine learning model is a Log-Trigonometric regression machine learning model.

Description

This application claims priority from U.S. Provisional Patent Application No. 63/717,245, titled “RFID DIRECTION IDENTIFICATION” and filed on Nov. 6, 2024, the contents of which is incorporated herein by reference. BACKGROUND Major retailers have started tagging their inventory with wireless tags such as radiofrequency identification (RFID) tags. The use of RFID tags provides advantages including improved inventory counting and item location. However, locating an item tagged with an RFID tag in a large facility presents technical and practical challenges because RFID does not provide inherent positional data. Rather, it only indicates the presence of a tag within the detection range of a reader. As a result, “pickers” are required to traverse the space to attempt to locate the items. However, due to the size of the facility and the lack of directional information, it can take the pickers a long time to find the items. BRIEF DESCRIPTION OF THE SEVERAL VIEWS OF THE DRAWINGS The accompanying figures, where like reference numerals refer to identical or functionally similar elements throughout the separate views, together with the detailed description below, are incorporated in and form part of the specification, and serve to further illustrate embodiments of concepts that include the claimed invention and explain various principles and advantages of those embodiments. FIG. 1A is a diagram of computing device used to locate an RFID tag. FIG. 1B is a diagram of the components of the computing device of FIG. 1A. FIG. 2 is a flow chart illustrating the operation of the computing device to locate the RFID tag. FIG. 3 is a graph illustrating a curve representing a plurality of collected data points that is used to determine the direction of the RFID tag. FIGS. 4A, 4B, 4C, and 4D are polar plots illustrating example curves for different distributions of data points. FIG. 5 is an example of a display on the screen of the device for directing a user to the location of the RFID tag. FIGS. 6A, 6B, and 6C are graphs illustrating example results of training an RSSI estimation function. Skilled artisans will appreciate that elements in the figures are illustrated for simplicity and clarity and have not necessarily been drawn to scale. For example, the dimensions of some of the elements in the figures may be exaggerated relative to other elements to help to improve understanding of embodiments of the present invention. The apparatus and method components have been represented where appropriate by conventional symbols in the drawings, showing only those specific details that are pertinent to understanding the embodiments of the present invention so as not to obscure the disclosure with details that will be readily apparent to those of ordinary skill in the art having the benefit of the description herein. DETAILED DESCRIPTION Examples disclosed herein are directed to a method of locating a wireless tag via a computing device. The method comprises: using at least one rotation sensor to determine a plurality of angle data of the computing device in the horizontal plane during motion of the computing device; processing a plurality of signal strength data of the wireless tag during the motion of the computing device; creating a plurality of data points by correlating ones of the plurality of angle data with ones of the plurality of signal strength data; training a non-linear regression machine learning model with the plurality of data points; and generating, at the computing device, a visual indicator of a direction of the wireless tag based on the training. The visual indicator of the direction of the wireless tag is displayed at the computing device. Additional examples disclosed herein are directed to a non-transitory computer readable medium having stored thereon instructions for locating a wireless tag. The instructions, when executed by a processor of a computing device, cause the processor to: use at least one rotation sensor to determine a plurality of angle data of the computing device in the horizontal plane during motion of the computing device; process a plurality of signal strength data of the wireless tag received at one or more antennas during the motion of the computing device; create a plurality of data points by correlating ones of the plurality of angle data with ones of the plurality of signal strength data; train a non-linear regression machine learning model with the plurality of data points; and generate, on a display of the computing device, a visual indicator of a direction of the wireless tag based on the training. FIG. 1A illustrates a computing device 100 (also referred to herein as the device 100). The device 100 may be a mobile device such as handheld computer, vehicle mounted computer, or wearable computer. The device 100 may be a lightweight handheld computer such as a smartphone or a more rugged device for use in a warehouse, for example. The device 100 can, in other examples, include any of a wide