In late December, teams from around the world descended on Florida’s Homestead Miami Speedway with millions of dollars of high-performance machinery in tow. But instead of the racecars that usually compete on the track, this was a battle between 16 of the most advanced robots on the planet.

The event that brought these machines together was the DARPA Robotics Challenge Trials, a competition designed to push the envelope in terms of robots that could autonomously serve in emergency and disaster scenarios that would be dangerous for humans. 

Penn faculty and students played a role on two of the teams. Team THOR is a collaboration with Virginia Tech and Korean robotics company Robotis, and is co-led by Daniel Lee, a professor in the School of Engineering and Applied Science and director of the GRASP Lab. Team TROOPER, which included engineers from Lockheed Martin and Rensselaer, features fellow Penn Engineering professors Kostas Daniilidis and Vijay Kumar.  

The robots faced off in eight different trials designed to mimic potential real-life applications for humanoid robots, such as clearing debris, cutting through a wall, or carrying and connecting a fire hose. The connecting theme between these trials was the ability to interact with environments and objects built for humans, like doors and ladders, or the steering wheel and pedals of a car.   

THOR is custom-built, while TROOPER is an Atlas robot from Boston Dynamics, one of the robotics companies recently acquired by Google. In both cases, the main contributions of the Penn teams were to the software and sensing that allow the robots to complete these complex tasks with minimal human input.

In the chaotic situations where they would be most useful, the ability to directly monitor or control such robots might be limited, so the teams all used their own hardware and software approaches to deal with unpredictable environments.

This diversity of approaches was a key motivator behind the Challenge.  

“The idea of the Challenge is for the government to spur creativity and see what the state-of-the-art is,” Lee says. “Robotics is a good example of this, since you can go head-to-head and compare them on particular tasks. It’s not as if there’s one homogenous approach that’s going to be the best at everything.”  

Team THOR, for example, excelled at the “Valve” and “Vehicle” trials, where its more nimble hands had an easier time gripping and turning the wheels involved.

After promising performances in these trials, both team THOR and team TROOPER will go on to compete at the Robotics Challenge Finals for a $2 million prize at the end of 2014.