In This Story
As the saying goes, if you love something, set it free. If it maps an area, finds a target, delivers a package, and comes back, the trophy is yours forever. Or something like that.
The black quadcopter drone surveyed the field in two quick swoops against a cloud-heavy sky. In just a matter of seconds, it picked up a signal and flew to the sidelines where it landed itself. A small ground rover took off, bumping over the grass to drop a first aid kit a few feet from the goal. Team members on the side of the field exploded with cheers and high fives.
The entire search and rescue took less than two minutes.
Programmer Qeon Edwards jumped with joy at just how well and how quickly their vehicles performed. “What, you doubted us?” he laughed, obviously relieved.
It was late in April when five teams gathered at the Xelevate drone ecosystem field in Leesburg, Virginia, to test their autonomous programming mettle via ground and air vehicles. This was George Mason ĢAV’s second year in the competition, which they won last year. As the reigning champs, they had a lot to prove.
The challenge rules change every year, so the team can’t just copy their predecessors’ codes. “Last year was basically a game of tag,” said Xuesu Xiao, assistant professor of computer science and the team’s faculty lead. They had to fly a drone around to squirt water on their opponents’ ground vehicles while avoiding their own.
This year, instead of trying to shoot others down, they had to perform a rescue mission: Find a randomly assigned “survivor” in a field of battle and send first aid. The faster the find and the closer the kit, the higher the score.
“Our strategy was for a drone to scout the area first, find the target, and then drive another ground vehicle robot to deliver the first-aid kit,” said Xiao.
The team programmed six flying drones and a ground robot that communicated with one another—and nothing else. That means that the programmers had no immediate control over the vehicles; they set their autonomous creations free and hoped that they did a good job coding them.
Overall, their robots performed exceptionally, managing to cut their delivery time and distance to less than a minute and 57 inches, respectively. They were just nudged out, however, by first-place ĢAV of Southern Florida and second-place Virginia Tech.
“It was definitely very tight between the top three!” said Cameron Nowzari, associate professor in electrical and computer engineering. “There was only a two-point difference between first and second, and one point between second and third.”
Still, the team wasn’t disappointed with how they performed. Their robots were a labor of love, and they met the challenge in a big way.