MIT researchers finally cracked the computer code for X-ray vision—at least for finding hidden objects like missing t-shirts or lost keys. Instead of providing users with stereotypical X-ray specs, however, the team’s newly dubbed “X-AR” setup offers something more akin to a geolocation program, a la Apple AirTags for wearable augmented reality headsets. Once fine-tuned, the team’s new system could become a useful tool within retail companies’ massive warehouses and distribution hubs, alongside casual consumers’ everyday home activities.
To pull off the impressive digital sleuthing, developers first designed a flexible, transparent RFID tag-communicating antenna sheet that adheres atop a Microsoft Hololens headset. RFID locators, which adhere to products like stickers and reflect signals back to antennas, generally rely on multiple antennas spaced apart from one another. In this case, the team needed to optimize a single antenna to handle a high enough bandwidth for the tags. Not only that, but the antenna couldn’t obscure any portion of the Hololens’ other visual tools and sensors.
Once fine-tuned, the team utilized synthetic aperture radar (SAR) to locate desired RFID-tagged items, like a boxed phone or article of clothing. To do this, the X-AR measures an area utilizing the antenna at various vantages while a user transverses the room. The headset then combines the data points like a standard antenna array to localize on its target.
After extensive testing X-AR’s creators were able to guide users with nearly 99 percent accuracy to items scattered throughout a warehouse testing environment. When those products were hidden within boxes, the X-AR still even boasted an almost 92 percent accuracy rate.
“While it presented a challenge when we were designing the system, we found in our experiments that it actually works well with natural human motion,” project collaborator Laura Dodds said in a press release. According to Dodds, a research assistant at MIT’s Signal Kinetics lab, humans’ frequent movements allows X-AR to take measurements from many different locales, and thus more accurately localize items. While wearing the AR headset, the X-AR system’s simple visual menu cues and icons help users move towards a desired location and let them know when the correct RFID has been scanned.
In future iterations, the team hopes to expand its sensing capabilities to incorporate WiFi, mmWaves, or terahertz waves. Right now, researchers say X-AR also only works within three meters of an object, so expanding its radius will be key for any future public deployment, alongside coordination abilities between multiple headsets.