Jul 16, 2019 | Atlanta, GA
According to Greek mythology, Icarus’ demise came from flying too close to the sun. While Team Icarus, Georgia Tech’s elite drone racing team, won’t be flying near the sun they do hope to fly quickly and efficiently using only artificial intelligence.
The team is one of nine finalists that will be competing in the AlphaPilot – Lockheed Martin AI Drone Racing Innovation Challenge this fall. The winner will take home a $1 million prize.
To reach the finals, the team – comprised of ten Georgia Tech students ranging from undergraduate to Ph.D. students – competed in a series of qualifying rounds. Each round tested a different component of the drone including drone control, vision, and a team game plan. Icarus successfully completed each round knocking out over 415 teams from 81 countries.
The finals will bring together all of those components to see which drone can complete the course as quickly as possible. The team is excited to debut an algorithm that was developed by a team member called the model predictive path integral (MPPI). The algorithm combines the control and planning aspects together – something the team hopes will give them an edge over the competition.
“I think the model makes us pretty unique amongst the nine teams. I’m excited to show off the power of this algorithm and to see it perform in competition. If it fails it will be a great learning opportunity for our lab, so it’s a win-win,” said Manan Gandhi, a fifth-year Ph.D. student in the Daniel Guggenheim School of Aerospace Engineering and team captain.
Grady Williams, Brian Goldfain and Paul Drews have been the key enablers for the development of the technology that the team is using in the Alpha Pilot Challenge.
While drone racing has become an immensely popular sport, including airtime on ESPN, Gandhi says it’s the other real-world applications of drones that excite the team and motivate them to create the best one possible. The underlying technology created in this competition could be used to improve drones used in scenarios like disaster relief, where time is of the essence and there is not a lot of sensory data to go off of.
When asked about the million-dollar prize, Gandhi says they try not to think about it too much.
“If we win, we want the money to go back to the team so that they won’t have to worry about funding for the next few years. Graduate school is stressful and difficult, and this prize money would help secure their futures while they pursue their research,” said Gandhi.
Beginning in September, the finals consist of four races, with all nine teams competing in the first three. The top three teams will compete in the grand finale in November.
Team Icarus is advised by Evangelos Theodorou, an associate professor in the School of Aerospace Engineering and affiliated faculty member of the Machine Learning Center at Georgia Tech.
Explore an interactive data visualization of all nine finalists here.