August 7, 2017
Formed just this year in 2017, the interdisciplinary Texas Aerial Robotics (TAR) team was established with the purpose of competing in the International Aerial Robotics Competition (IARC). IARC is a highly sophisticated competition requiring teams to learn and apply skills such as computer vision, computational simulation, parallel computing, controls, design, fabrication and corporate interaction.
In this Q&A, we asked two TAR members about their first experience at IARC this summer, which took place in July 25-27 at Georgia Tech. Umer Salman, the TAR controls lead, is a sophomore in electrical engineering and Eric Johnson, TAR president, is a junior in aerospace engineering.
This was the Texas Aerial Robotics team’s first major international competition. How long did your team prepare for this? What kind of “aerial robot” did your team design and how did you go about this?
Umer Salman: We’re a pretty new team, only really getting started early 2017. We spent all 7 months preparing for the competition. Typically, each of the four sub-teams works about 10 hours a week. However, that number goes up when we have specific goals we want to meet and does not include the work most members put in on their own outside of meeting times.
As for the robot itself, we sat down and figured what would work best — a straightforward, efficient design that needed to have interfaces for interacting with the target/ground robots and room for multiple sensors to make up for not having a GPS. The competition is intentionally held in a GPS denied environment, which necessitates the sensors we have onboard. We laser-cut wood boards for our base to allow for rapid iteration as we worked through problems with sensor types and locations. That ideology carried through to other parts of the quadcopter as well, which came in handy when we changed to more powerful motors after adding our onboard processing computer, an Nvidia Jetson TX1.
One big design decision we made is to not have landing legs. Our bottommost wood layer is almost completely flat, which allows us to have a much larger surface area for ground robot interaction compared to a quadcopter with legs. At competition, it seemed the judges really liked our rapidly iterable, efficient design. After this competition, we have a list of improvements to make to the design, and thanks to our design strategy, we should be able to implement those improvements quickly so we can spend more time focused on the challenge itself.
What were the mission requirements for this year’s competition?
Umer Salman: Completing Mission 7a of the International Aerial Robotics Competition requires a fully autonomous drone to fly around a 20m x 20m grid and “herd” Roombas (ground robots) out one side of the square field. The Roombas change direction randomly, so the aerial vehicle must tap the top of the Roombas to rotate them. Additionally, some Roombas have poles on them to try to knock your vehicle out of the sky. The most challenging part — and why teams have not been able to complete the mission after four years — is that the quadcopter must decide on its own which ground robots to target and autonomously fly and interact with, all without having a Global Positioning System. Without GPS, the vehicle has no defined references to figure out where it is. Mission 7b is the same competition; however, teams will compete head-to-head to score more Roombas through their side.
Your team placed 4th overall. How many teams actually competed and how many different U.S. teams and countries around the world competed?
Eric Johnson: 30 teams representing 9 countries registered; 13 teams showed up and competed in the American venue. Another 30 teams are scheduled to attend the Asian-Pacific venue.
What was the team’s biggest challenge for this project?
Eric Johnson: The team’s biggest challenge was integrating all of our sub-systems which were developed by separate sub-teams.
What is your team most proud of concerning this competition?
Umer Salman: That we were able to execute our goals for the competition exactly as we wanted, which showed the judges and other teams exactly where we stand after just a few months. Additionally, I was very proud that we exceeded the goal we set early in the year. That goal was to be competitive with the other teams, including those who have been working on this specific challenge for 3 or 4 years. By actually placing towards the top rather than the middle, like we expected, we surpassed that goal.
Will the team be competing again next year? If so, what will you do to improve upon this year’s design?
Eric Johnson: Each mission for the International Robotics Competition is played until the mission is solved. Mission 7 was not solved this year and Texas Aerial Robotics will be returning to Atlanta next year to solve the mission. We will be improving across the board. We will be improving our vision system, drone body, decision making and will be adding more sensors.
How do you think working on a team like this has affected your educational experience at UT Austin?
Eric Johnson: This competition is a great way for me to take what I learn in the classroom and apply it to a real-world engineering problem. It is also an opportunity to learn skills that are not taught as part of my aerospace degree.
What was the most valuable take away you and your teammates experienced from participating on TAR and in the competition?
Eric Johnson: For me, the biggest take away were the friendships I developed this year. It is truly a unique and rewarding to be part of a team that is as passionate about a project as you are.
