August 4, 2011

UAV team

"Of course we all have our limits, but how can you possibly find your boundaries unless you explore as far and as wide as you possibly can? I would rather fail in an attempt at something new and uncharted than safely succeed in a repeat of something I have done."

So said author A.E. Hotchner, and he could have been talking about the ambitious, can-do attitudes of The University of Texas Unmanned Aerial Vehicle team members this year. The team faced more challenges in preparing for the Association for Unmanned Vehicle Systems International (AUVSI) Student Unmanned Aircraft Systems (UAS) Competition than perhaps any previous team in UT's history: The flight computer for their in-house auto-pilot failed six weeks before competition, their plane crashed during a test flight, hundreds of hours of work ended up for naught. Subsequently, they may have learned more than any previous team—and they still brought home $500 in prize money from the June 15 competition at Patuxent Naval Base in Lexington Park, Maryland.

"To be frank, it wasn't the best result—a lot of the software we spent 50% of the semester developing never made it onto the airplane—but the team just pulled together so much," said Miki Szmuk, the UAV team captain and a recent aerospace graduate who began graduate school here last spring. "Out of the three years I've been with this team, this is the year the team has been the strongest, and I was quite proud of what I saw."

The UAS competition requires students to fly a plane on an unmanned surveillance mission in which the airplane must autonomously navigate and detect specific targets in a designated search area. Unlike the Design-Build-Fly competition, UAS competitors must have strong programming skills and a greater variety of interdisciplinary skills. In fact, this year's team of twenty students included one mechanical engineering student and three electrical engineering students.

"The reality is that today, more and more aerospace engineers are required to know more about other disciplines," said Vishnu Jyothindran, a December graduate who headed the team's work on the ground station. "DBF is all about designing and building and flying the airplane and has no autonomous component, so I think the challenge is very different."

Although the competition rules do not require teams to build their own autopilot systems - many teams purchase off-the-shelf kits and install them on their planes—UT's team has focused on building and programming everything in-house from scratch, from the autopilot system to the ground station to the target acquisition software.

"Our philosophy has always been that we want to learn everything we can," Szmuk said. Whereas other teams might purchase a kit and treat the whole autopilot system as a "black box" that simply does what they need, the UT team wants to understand the system inside and out. "Our belief is that when we go out into industry, understanding of what's happening in the black box is crucial. You can't just say it's an airplane with a propeller and there's a box that does this and another that does that."

As it turned out, however, the team did need to use an autopilot kit when their system's circuit board shorted out. Despite their best efforts—even contacting a former UAV team member in Spain (Dr. Sergio Rodriguez) who mailed them a board that might replace it—the team was forced to set aside six months worth of work on their homegrown autopilot system when they couldn't restore the board or find a working replacement. That meant losing all the time Jyothindran spent building and programming a ground station to work with their system and the time spent working on the communication link between the two computers that would connect to the autopilot system and the onboard video.

"It's basically taking the work we did in a semester and having to do it on a different platform in two weeks, and a lot of issues arise when you make last-minute changes," Szmuk said.

Their difficulties plagued them all the way to the competition. Just under a month before the competition, a power connector came lose during a test flight and all the controls went static: the airplane nose-dived from 300 feet at full throttle. Though the team was fortunate that no electronics were damaged, their morale did take a hit, and some proposed skipping the competition altogether. But they pulled together once more, and ten students attended the competition—where radio interference sabotaged their control system and the plane's range dropped from a half mile to a couple hundred feet. The airplane was never able to leave the vicinity of the runway.

"When we designed everything ourselves, we had the opportunity to put in our own features, but now we were constrained by the features of the off-the-shelf autopilot kit," said Jon Tamir, the electrical engineering December graduate who handled the team's target acquisition software. "That made integrating the system more difficult."

Yet the team achieved their most important goal: learning.

"They learned a huge amount this year, and they did extremely well," said Mark Maughmer, the team's staff advisor. "One thing they learned is that testing all the components together is extremely important in having a successful product, but this time, time didn't quite allow it. We'll be able to use a lot of what they learned in many other projects coming down the line in future years."

Szmuk, Jyothindran and Tamir all emphasized how beneficial the experience was.

"I think people are scared to use the F-word, 'failure,' but it's an elemental part of any project," Szmuk said. "You never hear about failure in class, you're not taught how to deal with it, and you don't hear about it when you hear the great names in science, but I think it makes us stronger in character and makes us realize the things we've done wrong so in the coming years, we won't make the same mistakes."

View the entire set of UAV competition photos on our Flickr site.