Tuesday, August 4, 2015

2015-08-04: Autonomy Incubator Intern Nick Woodward Designs UAV Geo-Containment Software, Plastic Bananas

Nick, surrounded by recent projects, edits a design for the 3D printer.

The most noticeable thing about Nick Woodward's workspace is not the complex code covering his two desktop screens, nor is it the vivid illustrations adorning his PowerPoint slides as he prepares for his exit presentation on Thursday. It's the bananas. When he's not elbow-deep in unmanned flight safety research (which is most of the time), Nick relaxes by designing and 3D printing tiny plastic props for the Autonomy Incubator's demonstrations. Once they're implemented, the intention is to offer them as souvenirs to high-profile visitors as well. Some offices give out pens with their logo on them; we send people home with bananas and Pac-Man figurines.

NASA's tiniest airplane, banana for scale.

Blinky, with "AUTONOMY INCUBATOR" printed on his head.

Pac-Man, also representing the AI in his own way.

We bet you're REALLY wondering what we're doing with a plastic human heart.

"I sort of inserted myself as the CAD guy," he said, citing a long history of 3D printing experience. "It's a lot of fun."

His adventures in 3D printing are only a casual hobby during his breaks from his real project, however-- his area of research this summer has been autonomous flight safety, and his end result is a piece of software that sets and enforces geographic boundaries for autonomous UAVs during outdoor flight. Think of it like an invisible fence for your dog.

"We have the actual safety net up in the flight area to protect the humans," Nick said. "This project is aimed toward maintaining not only [outdoor] vehicle safety, but the safety of those around the vehicle."

Nick's program divides the geo-containment area into a flight zone, where the UAV is approved to fly, a "soft border" zone, and a "hard border" boundary. If a vehicle strays across the soft border, "The software basically tells the vehicle, 'Hey, you're close to leaving the flight zone, but as long as you don't make a dash for the hard border, we're good,'" Nick explained.  The program also steps in and the UAV's flight trajectory to predict if it's preparing to leave the geo-containment area while it's in the soft border zone.

On the off chance that a vehicle crosses the hard border, two things can happen: The first thing the software will do is command the vehicle to land immediately, since that's one possible safe maneuver. If the land command doesn't work for some reason, then the program can shut off the motors.

Nick is proud of his finished product, but says that working on emergency safety protocols has been an unusual experience for him.

"The weird thing about this project is, this is a piece of software I hope never gets used," he said.

This summer has been Nick's second in the AI; he first joined the team in 2014 when he came on board to help build a "dual hardware-in-the-loop simulator." Since the AI didn't have an in-house flight range last year, building sophisticated simulators was the only way for the team to run tests.

"We would strap the [UAV] components onto a table, feed it false information, and pipe its output back," he said.

Of the incredibly diverse roles he's played in the Incubator, he merely shrugs the challenge off as part of the job.

"I'm a robotics engineer, so I'm supposed to be able to tackle things as they pop up," he said. "Being able to evolve and adapt to the problem at hand is an important part of being an engineer."

Nick and high school intern Nick in one of their friendly debates.

Soon, Nick will depart the AI for Worcester Polytechnic Institute, which is apparently pronounced "WOO-ster," not "Wor-kes-ter." We're just as baffled as you are.

No comments:

Post a Comment