When workers manually take warehouse inventory, it's a laborious process. Warehouses have to shut down for days at a time for workers to manually check each box, sometimes peering through binoculars to check product on the top shelves. This process is also prone to errors and lost inventory.
Marc Gyongyosi's solution to the problem? Autonomous flying robots.
Gyongyosi is the founder and CEO of IFM Technologies (which stands for Intelligent Flying Machines). The startup has developed devices that fly up and down warehouse aisles scanning barcodes to track inventory, at a rate 400 times faster than the manual process, he says (about 20 minutes to complete one aisle).
But don't call it a drone. "First of all, 'drones' is a buzzword right?" Gyongyosi said. "Everyone is using that to describe whatever piece of hardware they’ve managed to get up in the air to take a picture of themselves. Here at IFM, we’re building autonomous systems robots that happen to be flying, but whose main job is to collect data repetitively and with extremely high accuracy."
Warehouses are just the first application of their products. Gyongyosi said he sees IFM Technologies as a data analytics company using computer vision and robotics to automate data capture. Eventually, he envisions using their software and hardware to capture spatial data that could be integrated with augmented reality, virtual reality and Internet of Things technology as indoor spaces become more connected.
"Instead of people manually mapping out the information, we can automate that process with our flying robot," he said.
But first step: warehouses. Gyongyosi and two Northwestern computer science PhD students, Siddarth Jain (who specializes in robotics and perception) and Nathan Matsuda (who focuses on computational imaging and sensors) have been working for the past two years to develop a hardware and software solution that can handle the unique challenges of flying indoors. Unlike typical outdoor drones, it doesn't run on GPS or have sensors. Instead, it relies on computer vision and their own algorithms to locate itself in space. The processing is also all done on board the hovering robot.
"There’s no external systems, no markers, no beacons," he said. "This is a system that knows down to the centimeter where it is in space and it doesn’t need any external guidance for that."
Warehouses pay a one time set up fee, and then pay a subscription to their software. The price varies depending on the size of the warehouse, but Gyongyosi said their prices are similar to industrial robotics products.
Over the past year they've been validating their business model and testing their tech through pilots with automotive manufacturers, food and beverage distribution and industrial supply. Gyongyosi declined to name the companies they are working with and whether IFM is making revenue. But he said, "We’re in a very good position."
IFM Technologies won $15,000 at Northwestern's Venture Challenge in May, and have been working out of The Garage this summer. In April they presented at the GPU Technology Conference organized by NVIDIA in Silicon Valley, in September they presented at TechCrunch Disrupt, and they're competing at Hello Tomorrow in Paris this month. They're also looking to raise funds, but Gyongyosi declined to comment in any detail.
That support and exposure will be key, as they aren't the only company that's working on the warehouse inventory problem using robots: Walmart, France-based Hardis Group's Eyesee, and South African startup DroneScan, are all creating tech that goes after the same issue.
Previously Gyongyosi worked in robotics at BMW in Munich and MDAR Technologies, a Northwestern robotics startup. But he said the genesis for IFM Technologies goes back to when he was 15 and built a 737 flight simulator in his basement.
"What I realized is that I can combine my passion for flight and my passion for robotics," he said. "And that’s what I’ve been working on ever since."
