German clothing chain Adler Modemärkte is among a handful of retailers using an RFID-enabled robot called Tory to count inventory and identify the locations of merchandise on store shelves each day. The robot and the software that manages the data it collects are provided by German technology firm MetraLabs.
Adler is carrying out a pilot project involving two Tory robots that it purchased, one for use within its store in the city of Erfurt and the other at the store located at its corporate headquarters in Haibach. The company plans to expand the deployment to other stores later this year. Adler, which already attaches passive EPC Gen 2 ultrahigh-frequency (UHF) RFID tags to most merchandise it sells, has been using handheld readers at all of its 177 stores for some time, says Roland Leitz, the company’s head of IT. Compared with manual or bar-code-based inventory checks, he adds, checking inventory via RFID handhelds “speeds up stocktaking significantly.” However, Leitz notes, because the process requires that an employee walk through store aisles and past shelves, waving the reader at nearby items, “it is manual work that ties up capacities of staff.”
Tory (the name is derived from the word “inventory”) offers an alternative. The robot can be set loose on a store floor, and will use sensors to navigate its way around the sales area, reading tags as it goes. “Our aim is to reduce administrative tasks even further so that resources can be allocated to sales activities,” Leitz explains. “With the help of a robot, stocktaking can be conducted more often so that data on the availability of goods is always highly up to date.”
Adler launched the robot deployment in October 2015, Leitz says, and plans to expand the pilot to include as many as 10 stores during the course of this year. “The number of stores that will be permanently equipped with the system has not been determined yet,” he states, “and depends on the outcome of the pilot phase.”
Currently, there are more than 200 MetraLabs robots deployed at stores, industrial sites, museums and restaurants around the world, says Johannes Trabert, MetraLabs’ co-founder and executive partner. A handful of them, Trabert reports, are the newly released Tory model being tested for RFID inventory tracking.
MetraLabs, founded in 2001, released a robot in 2007 to help customers locate goods within a store. That early version employed a 124 kHz low-frequency (LF) reader mounted to its underside and passive LF RFID tags installed on floors to navigate its way to a particular product. MetraLabs designed and built the reader itself, says Christian Reuther, the company’s senior software architect. The tags, as well as laser- and camera-based sensors, provided navigation for the robot since each unique ID number encoded to a tag was linked in the software to a specific location, enabling the robot to use those IDs to understand where it was located at any given time. Each product name was linked to a particular location, and the robot knew the IDs of the floor-installed tags it expected to read while moving toward that spot.
By 2009, MetraLabs had enabled a robot to perform inventory tracking by building a UHF reader into it, and tested the system in a University of Tübingen laboratory, which assisted with the testing. However, Trabert says, there were not enough companies or organizations at that time that tagged items within their facilities to provide value for RFID-based inventory tracking. That has changed during the past five years, as large retailers are now tagging their goods or receiving tagged products from suppliers.
The Tory robot can accomplish two functions: using its built-in Impinj UHF RFID reader to count inventory, and utilizing its laser and camera sensors to locate specific products for shoppers. The UHF reader comes with a custom antenna array that MetraLabs developed “to yield high accuracy and read rates in typical retail scenarios—very tall and low shelves, stuffed boxes of merchandise, multi-path-propagation problems due to metal shelving,” Trabert says. To determine the location of a specific product, a shopper would use Tory’s touchscreen to input that item’s name. The robot would look up that product’s assigned location on the floor map stored in its memory, and then use its internal sensors and odometer to guide itself to that product’s assigned shelf or display fixture. (Unlike earlier models of MetraLabs’ robots, the Tory does not have a 124 kHz reader for navigation purposes.) In this way, the Tory could be used to capture inventory data at night, and to provide customer assistance during the day.
To launch the inventory-tracking functionality, a user would first need to set up the robot’s route or coverage area. This is accomplished using the device’s touchscreen to communicate with its onboard software. The software displays instructions on the screen to guide users through the process of setting up a new “scan area.” The retailer first places the robot on a selected start point to begin its inventory count, then creates a name for the scan area or route and presses “start.”
The user can then guide Tory along the desired scan area or navigation route (the screen will show the route recorded so far), and press “stop” when finished. The robot, which measures 1.5 meters (4.9 feet) tall and 0.5 meter (1.6 feet) wide, can be guided remotely via a game pad, keyboard, laptop or smartphone, Trabert says, but the most precise navigation is offered by pushing the machine manually. “The robot’s wheels make it very easy and effortless to guide it through the store in exactly the way that the inventory run should ideally be performed later,” Trabert states. Remote-controlled methods like a game pad or smartphone, he explains, tend to be too imprecise to accurately reflect the navigation route for inventory. “As this is a one-time procedure, we usually advise our customers to ‘take a stroll through the store’ with Tory. After that, setting up the robot is finished and automated inventory is ready for operation.”
Tory saves the route in its memory and can later be dispatched from the same start point anytime that inventory counts are required. If there are any unforeseen obstacles or if any shelves have changed, Trabert says, the “navigation software will recalculate the route to best fulfill the ideal scan route, and if it’s not possible, it will determine how much of the blocked area to skip for minimal detrimental effects to the inventory count.”
There can be a virtually unlimited number of routes or areas per store, Reuther says, but a full inventory run will typically be the most commonly used route. Other routes could include a certain merchandise family or an area of the store for specific inventory counts.
The robot can travel at speeds of up to 1 meter (3.3 feet) per second, but that speed can be adjusted during inventory counts according to the density of RFID tags and the activity within the area (for instance, the presence of shoppers and store personnel). Typically, Tory can read tags located as far away as 8 meters (26.2 feet), at a rate of up to 250 tags per second. Its built-in memory can store more than one million tag reads. The robot forwards data to a server via a Wi-Fi or wired connection.
MetraLabs software, which resides on the retailer’s database, stores information regarding the inventory that should be located at each store shelf or rack. If the robot fails to capture the tag IDs of items expected to be found at a certain location, the machine can return to that section and perform another reading. It can also circle back for a second read in areas where tagged items are very densely packed together, making the likelihood of a missed tag read high. The robot can then forward data indicating which items are missing, via a Wi-Fi connection. These options are dependent on the retailer’s actual requirements. “Our software serves to interface Tory’s data output—tag EPCs, timestamps and spatial locations of tags—with the ERP [enterprise resource planning] system of the retailer.”
The robot accomplishes its inventory-counting tasks about 10 times faster than would be possible via a manual count with a handheld reader, Trabert says, based on user trials performed with retailers.
In addition to using the robot to track inventory, Leitz reports, Adler also plans to test RFID technology with intelligent changing rooms. Such a system would identify clothes being carried into the fitting room, and would display data about those items by reading each garment’s RFID tag. “We are planning to install intelligent changing rooms in selected markets [stores] in the course of 2016,” he says.
MetraLabs will be attending this year’s RFID Journal LIVE! conference and exhibition, to be held on May 3-5 in Orlando, Fla. At the event, the company will display the Tory robot at its booth (#1044) on the show floor.