Inside Unmanned Systems provides actionable business intelligence to decision-makers and influencers operating within the global UAS community. Features include analysis of key technologies, policy/regulatory developments and new product design.
Issue link: http://insideunmanned.epubxp.com/i/696732
42 unmanned systems inside June/July 2016 AIR NEW APPLICATIONS "What it fundamentally is doing is ensuring that the inventory is indeed in the place and at the quantities that you expected," said Matt Yearling, PINC's CEO. For large operations with hundreds of dis- tribution centers across the country the cost of lost inventory can be substantial. "Your inventory management system says you've got five pallets of this but you can only find one. Each pallet is $50,000—now you've got a problem," Yearling said. "That's really fundamentally it. If you are operating at scale, that's where this is important." Learning As They Fly PINC builds its own platforms from off-the- shelf components, with one key exception—a very directional RFID antenna. That antenna narrows the sensor field so that it gives the RFID reader a high degree of confidence that it's looking at a specific item. "It's just like a spotlight on the as- set," Yearling said. In the process of scanning the shelves, the drones are also learning the layout of the warehouse, building knowledge of where they are in the build- ing as they go. A PINC drone uses both RFID and OCR to deter- mine it s loc ation within the warehouse. The drone can read the tags on the shelves as well as on the items in stock, keeping track of its direction and speed as it moves around. The processing power to do this is contained on the platform so the drone is not dependent on an external link or network. "We do simultaneous location and mapping (SLAM) inside," Yearling said. "…Once you've identified where you are, then as you're look- ing at a piece of inventory, you know how fast you're going and where you're going. You use that as well as information collected from what you're looking at. It's all those multiple reads that we feed into our algorithm to determine where the location of that asset is." The drones must learn their environment, Yearling said, because it is not realistic to add a layer of infrastructure across a firm with hun- dreds of large distribution centers or ask the company to change the way it labels its inventory. "(A drone) has to understand where it is just like a human," Yearling said. "I walk down the rack I know where I am based on the labeling on the rack. …We have some understanding of the environment. We navigate based on the environment and we read the labels on the in- ventory in whichever way is appropriate based on how they do it." For a million-square-foot warehouse, Yearling estimated it would take "less than a handful" of drones—typically working during a relatively quiet period like the early morning hours—to handle the inventory reconciliation work. Photo courtesy of Exponent Technology Services by Ayhan Kamil