Marc Gyongyosi, CEO and founder of Intelligent Flying Machines, discusses in the second of this two-part story how he leveraged his research in robotic safety into developing safer, more accurate forklifts.
As time progressed and technology advanced, designers focused more so on humans and robots working cohesively, rather than two separate units. “Now, we’re finally ready for collaborative robotics that you can put a human next to,” Gyongyosi said.
One project Gyongyosi worked on to help assist workers was a robot for BMW that attached the insulation in a car door. Seemingly simple, this task required a human consistently to apply insulation, and if it wasn’t consistent, the car door would have gaps where rain could get in. Additionally, workers’ hands were cramping from this task. So, Gyongyosi said this simple task was taken over by a robot and the human was then moved to a different location in the process.
“It wasn’t about creating AI technology, it was about creating the framework,” said Gyongyosi. “Technical capabilities are expanding and require a different approach. It’s not about pushing what’s possible, but how you do it safely.”
Looking back, he says everyone should push the boundaries of what’s possible, but it needs to consist of a closer feedback loop on what individuals need and do in the industry. From here, Gyongyosi saw a new category of technology—making existing systems smaller and safer, and integrating it into technology.
“There was a huge opportunity here,” he said. “But for me, I have never quite understood the ‘powered by AI’ concept. If you want to build a product that’s reliable, it comes down to solving very specific problems extremely well. It doesn’t matter if it has AI integrated.”
It was with this philosophy that Gyongyosi saw an opening. While working at BMW, the robot was working cohesively with employees and the car, but oftentimes the parts didn’t get there in time.
“I saw an opportunity,” he says. “Sometimes, the warehouse screwed up or humans misplaced pallets. I wanted to build flying robots.”
Before, workers were using binoculars to locate labels on boxes and count/check inventory. He wanted to use drones to autonomously fly past boxes, locate the information and compile all the information into a system. At first, his drone-flying method wasn’t working when applied to different warehouses with lighting changes and altering elements. But, after seeing how machine learning worked, using cameras and convolutional neural nets, his method was finally successful in all different types of warehouses, despite lighting differences and changes in the environment.
From here, Gyongyosi started his own company called Intelligent Flying Machines (IFM), but instead of pursuing his flying drones, he took a step back and recapped the real problem—warehouses were misplacing stuff.
He asked a series of questions: Do you need drones? Is this the best solution? How safe is this? Am I adding complexity to a process?
“Our mantra at IFM is to think simple,” he said. “The real problem was humans were making mistakes in the processes they had. So, I needed to build technology to make the process smarter.”
Humans were misplacing things that were leading to inventory errors, and were hurting one another with forklift accidents.
“Ultimately, everything we were looking at came back to one thing…forklifts,” he said. “Forklifts were the reason why things would get screwed up.”
So, he put the technology he had integrated into the drones, into forklifts.
“We created OneTrackCoPilot,” he said. “Think of it as Google Assistant for the forklift. You can attach it to any system and use it to prevent misplaced inventory, maximize throughput, and increase safety for drivers.”
“We can tell you were you are, where you’re placing your stuff. There’s no IT infrastructure, it’s just powered by cameras. You can attach it to anything, but right now we’re starting with forklifts.”
In turn, this provided rich data connected to the cloud that provided an opportunity for warehouses.
“This is a good understanding of how to bring valuable tech into warehouses today,” said Gyongyosi. “Everyone wants to compete with Amazon, and that’s what we’re helping them do. This is an example of how powerful the right technology can be when applied to the right problem.”
Filed Under: Rapid prototyping, Robotics • robotic grippers • end effectors