Exyn Technologies, a spinoff of the University of Pennsylvania's GRASP laboratory, has revealed groundbreaking artificial intelligence software that essentially lends autonomous drones smarter navigational skills.
Exyn AI-Powered Drones
With the help of Exyn's AI software, drones may be able to fly and wade through the dark environments, dodge obstacles hindering its path, or travel to locations beyond the reach of GPS. For this to work, Exyn employs a technology called sensor fusion, which lends situational awareness to drones, helping them travel in the way humans naturally do.
No Human Intervention
In a demo video showcasing the technology the company shared to TechCrunch, a drone integrated with the company's AI software activates, then immediately scans its surroundings. From a starting point in a densely populated workplace, it launches, making its way toward the nearest exit, doing so with nary a human intervening.
The route, unlike most autonomous drones, isn't preset, and the drone's pilots didn't mangle the controls in order to affect the way the drone behaved. In other words, they simply told it to find the nearest exit, and off it went.
Drones Have Harder Jobs Than Driverless Cars
Vijay Kumar, the founder of Exyn, and a consummate roboticist, whom all told is the dean of University of Pennsylvania's School of Engineering, says that the navigational hurdles of drones are far more complex than that of driverless cars, or any autonomous machine that's ground-based.
"Artificial intelligence that lets drones understand their environment is an order of magnitude more complex than for self-driving cars or ground-based robots."
Drones have far more complex environments to chew on because they're essentially traversing 3D space. On top of following traffic regulation, and carefully avoiding civilians, autonomous drones must recognize obstructions and steer away from them. This all happens aloft in unmapped skies, where there's no stable internet connectivity. Additionally, drones are fighting against gravity.
"Cars roll along on wheels and can carry large batteries. But drones must preserve all the power they can for flight," he said.
Exyn's AI software will function with any type of unmanned aerial vehicles, ranging from mainstream drones to research or even industrial models. The whole thing works by fusing a slew of sensors from different parts of the spectrum for the drones to create a 3D space, the process of which happens in real time. Exyn only provides a drone with a goal and a starting point.
"But it takes off, updates its map and then goes through a process of planning and replanning until it achieves that goal," said Jason Derenick, Exyn's chief engineer.
More importantly, because the technology is essentially kept inside the drone, the vehicles won't rely or even require external infrastructure, such as pilots manning the drones to accomplish a given route. Moving forward, Exyn will also cull and integrate data with the help of sources in the cloud.
Nader Elm, the company's CEO, hopes Exyn's AI software will help propel new usage for drones, especially in environments where it's dangerous for human workers to operate. The CEO cited dark mine shafts, warehouses, or buildings-in-construction that may require constant safety checks, as examples of this notion.
Thoughts on Exyn's AI software that makes autonomous drones smarter? Watch the company's demo video below to get an idea. Also, feel free to sound off in the comments section!