Drones

Drones with RFID-relays promise more accurate inventory tracking

Drones with RFID-relays promise more accurate inventory tracking
MIT wants to use drones as RFID-relays in warehouses
MIT wants to use drones as RFID-relays in warehouses 
View 1 Image
MIT wants to use drones as RFID-relays in warehouses
1/1
MIT wants to use drones as RFID-relays in warehouses 

Tracking stock in modern distribution warehouses is a huge challenge for even the best-run businesses. Lost items reportedly cost American retailers over $45 billion annually, so tech that makes inventory tracking easier has the potential to have a huge impact. A new MIT system could do just that, using small drones and RFID tags to monitor inventory.

According to the MIT research team, the biggest challenge in developing its RFID-reading drone fleet was finding a way to make it safe. The only drones safe enough to fly in close proximity to humans are small, lightweight ones with plastic rotors. Unfortunately, those drones aren't able to carry RFID readers with a meaningful range.

Rather than relying on the drones to carry readers, the researchers have developed a way to use them as relays, allowing the use of drones without developing new RFID tags, tag readers or software. But the approach doesn't come without its challenges.

The team found that trying to determine an item's precise location by relaying RFID signals causes problems when it comes to signal processing. RFID tags are wirelessly powered by the reader, meaning the transmitter and tag work on the same frequency. Adding a drone-mounted relay adds two more transmissions to the mix, which means there are four transmissions happening simultaneously on the same frequency.

The way around interference would usually be to decode the initial transmission from the RFID tag, re-encode it and transmit to the reader. But the drone is constantly moving, so the time that process takes makes it impossible to accurately record an item's location. To sidestep the issue, the researchers developed an analog filter capable of isolating the tag's signal and re-packaging it.

MIT says this problem was made worse by the fact its system needs to accurately locate the RFID tags to be of use. Location-detection systems traditionally use antenna arrays to work out where the RFID signals are coming from, but the drone is too small to carry one. The fix for this is simple: because it's constantly moving, readings taken at different times will correlate with different locations, essentially simulating the role of a traditional antenna array.

On top of this, because the drone is constantly moving around, the signal reaching the reader is affected not only by the position of the RFID tag, but the position of the drone as well – and the reader isn't able to work out how much the drone's movement has impacted the location of the signal.

MIT researchers have fitted each drone with its own RFID tag as a workaround. The drone alternates between relaying the signal of the tagged item it's locating, and its own signal to the reader, allowing the system to work out how much the drone's movement has affected the result and correct it.

In a set of experiments, complete with RFID-tagged objects hidden away to mirror merchandise haphazardly stacked on warehouse shelves, the system was accurate to 19 cm (7.5 in), and increased the range of RFID readers by 10 times in every direction. It will be put through a second phase of trials with a retailer in Massachusetts.

Source: MIT

2 comments
2 comments
chase
Way too many easier ways to accomplish the same task. I see it over and over concerning UAS's where they try to force fit it into yet another use. Trying to force a square peg into a round hole. Which can be done if you hit it hard enough. It's just silliness if given any amount of serious thought.
Koolski
Have you ever worked in a warehouse full of parts? I have. Every part, no matter how small, has to be tagged if this is to work. It's not sufficient to tag a box full of couplings that only cost a couple of cents a piece. Boxes break because they got wet or the fork lift driver hit the shelf or the slip sheet driver damaged the pallet while unloading or etc... You get the idea. Murphy's law is in full swing. So then the box is placed on the shelf and because it's broken, parts spill under, around, into every place possible. What if some spilled when they were enroute from point A to point B? When pallet C is brought down from the top shelf, what if the pallet of boxes next to it gets damaged at that point and starts to spill items? What happens if the RFID tag falls off or is bumped off? Or someone stole an item and dumped the tag somewhere? There's a lot of potential problems other than just reading a tag.
My point is that 10 x distances won't be enough; this definitely isn't a one size fits all solution. And the only hope of this working is if every single item is tagged. I think the potential is better for a warehouse of appliances than for smaller items.