PAL Robotics Rolls Out Tag-Reading Robot

By Claire Swedberg

The Barcelona firm expects to provide the newest version of its motorized StockBot to retailers for long-term pilots later this year, to count inventory by reading RFID tags.

PAL Robotics plans to launch its first large-scale pilot of its motorized RFID-reading robot in Europe during the second half of this year. The newest version of the Spanish company's StockBot will be tested for its ability to read the RFID tags attached to products, while software will identify where those tagged items are located within stores.

Now that RFID tags are ubiquitous on products in some larger apparel retail stores, a number of companies are seeking ways to make it easier to conduct inventory counts without having to draw from employees who would otherwise be assisting customers, says Sergio Ramos, PAL Robotics' product manager. RFID makes inventory checks much easier compared with bar-code label scanning and visual counting, he explains. However, the technology typically requires an individual with a handheld reader to walk along the sales floor, waving the device past each group of products.

The StockBot has Impinj R420 readers and eight antennas onboard to capture the RFID tag IDs and link them with its location.

Fixed RFID readers are still somewhat expensive. Therefore, PAL Robotics has developed an alternative that is more automatic than handheld readers and less expensive than a fixed reader infrastructure.

PAL Robotics is currently in discussions with several European retailers in preparation for a long-term pilot of what is now the third-generation version of the StockBot. Citing non-disclosure agreements, the company declines to name the stores that may be piloting the robot.

Other technology companies are also selling RFID reading robots for the retail market. Ramos believes that StockBot differs from many of its competitors' robots because it was designed specifically for the use case of RFID in retail. StockBot is designed to adjust to a changing environment, such as the presence of customers, or the movement of displays or shelves. First, retailers use the robot and a joystick to map out the store and establish a route that the machine will travel while conducting an inventory count. The joystick (which is similar to the kind used in PlayStation devices) sends data to the robot via a 2.4 GHz signal. PAL Robotics also offers an application enabling users to control their robot via a Wi-Fi connection from an Android phone. The robot stores its route in its software. In the future, it will be able to automatically update that route, based on changes in conditions as detected by its built-in laser and 3-D camera sensors.

Since its launch in 2004, PAL Robotics has created a variety of humanoid robots designed to be interactive at conferences and exhibits, as well as autonomous mobile bases onto which items can be transported, for instance. The robots use laser sensors to navigate their way.

In 2014, Ramos reports, the company designed a model specifically for reading RFID tags. The StockBot has been through two prior versions, and the third is now being released, designed to have a smaller form factor than its predecessors, while still able to read tags on shelves at a height of 2 meters (6.6 feet) or more. The company has been testing the second version of StockBot at Barcelona Outlet, a clothing retailer located near the robotics firm. At that location, PAL Robotics has tested a StockBot's ability not only to follow a navigation route, but also to adjust that route as the environment changes—such as a shelf being moved to another location.

With most other robots on the market, Ramos says, a retailer would first use a joystick to guide the machine through the store in order to map out the site, and then configure the desired route that the robot would navigate. "It works fine, but a few days later, your shop is not the same as the map of the robot," he explains. "This produces two effects. The first one is that the robot gets lost easily. And the second one: The robot can't find a path to go to the places that were configured."

Typically, Ramos says, stores would then need to make a new map and reconfigure the route. "We don't like this solution," he notes. "It implies that a person has to make the configuration every few days." Instead, he says, when the StockBot encounters changes to its environment, based on data from its sensors, it can automatically modify the navigation route stored in its software.

"All this navigation work has been tested in Barcelona Outlet," Ramos states, as well as at several larger stores that have asked to remain unnamed.

PAL Robotics' Sergio Ramos

In addition to lasers and 3-D cameras to identify its location and move around obstacles, the StockBot has Impinj R420 readers and eight antennas onboard to capture RFID tag ID numbers and link them with its location. That location is determined via data from the robot's laser and camera sensors and odometer (which measures distance by tracking the number of times that the unit's wheels revolve). PAL Robotics software then receives all of that data via a Wi-Fi connection and displays it on a map.

When first deploying StockBot in a store, a retailer employs the wireless joystick to move the robot forward, back, left and right, in order to map out the area. The company then accesses that map via the StockBot's graphical user interface to indicate where on that map it wants inventory counts to take place. The robot stores that data and can then travel through that area autonomously, reading tags at a distance of approximately 1 meter (3.3 feet). The speed at which the machine moves is set according to the environment—for instance, a store containing a lot of tight turns, or the presence of shoppers during the business day, would require a much slower pace to accomplish inventory counts. Testing carried out by PAL Robotics has shown that if the environment is fairly simple, a 1,500-square-meter (16,160-square-foot) space can be fully scanned within about an hour while a store is closed. With that in mind, Ramos says, the robot can typically take an inventory count of an entire store within a single night.

"A store is a very dynamic environment," Ramos says. "We designed the StockBot to adapt to that." He says the third version's slim form factor—about 6 feet in height, and 1.6 feet in width and depth—as well as its ability to be easily programmed by a user for a specific route make it one of the best options for a retail-based RFID reading robot.

The robot detects its own location on the mapped-out route, based on its own sensor data, and that location is stored with each RFID tag read. The software then links that location with the tag's ID number and can display that product's location on a map.

The software can be used not only for inventory collection, Ramos says, but also for analytics. Stores could collect data regarding the locations of tagged goods, for instance, and compare that information against sales data to determine when goods from specific parts of the store are selling more often than items located elsewhere. PAL Robotics' software can be integrated with the user's own store-management software.

Ramos says his company is presently in discussions with several customers located throughout Europe, including larger retailers that already have RFID tags on their goods and have stores large enough that it can be too time-consuming to perform manual reading via a handheld device. Some of PAL Robotics' potential customers have not RFID-tagged their goods, he adds, but are in the process of launching such a deployment and hope to use the robot.

PAL Robotics will be exhibiting at the RFID Journal LIVE! conference, in Orlando, Fla., on May 3-5, at Booth 549, where it will be demonstrating its StockBot robot.