Team-KAIST won in the DARPA Robotics Challenge 2015 Finals

DRC-HUBO finished all eight assignments in less than 45 minutes, taking first place among 24 international teams and claiming the USD 2 million prize offered by a US defense research agency.   

Article  |  Fall 2015

Team KAIST of the Republic of Korea led by Professor Jun-Ho Oh of the Mechanical Engineering Department at the Korea Advanced Institute of Science and Technology (KAIST), Professor In-So Kweon of the Electrical Engineering Department, and researchers from Rainbow Co., the university’s spin-off company that builds the robots, won the US Defense Advanced Research Projects Agency (DARPA) Finals. The Robotics Challenge Finals 2015 took place on June 5-6, 2015 at the Fairplex in Pomona, California, where the team was awarded a USD 2 million prize.

The DARPA’s Robotics Challenge (DRC) promotes a competition of robot systems and software teams which seek to develop robots capable of assisting humans in responding to natural and human-made disasters such as the Fukushima Daiichi nuclear incident in 2011. The DRC consists of three competitions: a software-based Virtual Robotics Challenge which took place in June 2013; the Robotics Challenge Trials in Homestead, Florida, in December 2013; and the Finals in June 2015.

The winner of the Finals, DRC-HUBO+ is a humanoid robot designed for operation in disasters areas and human environments. The robot was developed in HuboLab (Humanoid Robot Research Center), KAIST, in 2014, and its basic configuration is based on HUBO2, which was developed in 2009. Compared to HUBO2, the maximum power of each joint in DRC-HUBO+ was increased 3 – 4 times by combining an efficient air-cooling system, and the lengths of the arms and legs were increased in order to extend the workspace and the walking mobility. DRC-HUBO+ has 32 DOF (degree of freedom), and it uses 36 BLDC motors. Its height is 173cm, and its weight is 82kg. Its outer aluminum cover case provides excellent protection from the outside environment. Additionally, a LADAR sensor and vision cameras are equipped on the head. These sensors assist the robot in active cognition about the surrounding environment. Two legs, F/T (Force-Torque) sensors, and FOG (Fiber Optic Gyro) sensors allow this robot to walk on the ground. DRC-HUBO+ can transform its posture into wheel mode (mobile robot mode). These outstanding features of DRC-HUBO+ enabled Team-KAIST to win the 2015 DARPA Robotics Challenge.

A total of 24 teams from universities and private and public research institutes from Korea, the US, Hong Kong, Germany, Japan, and Italy participated in the Finals. The participating teams had to finish eight assignments in 60 minutes, during which their robots were untethered and operated wirelessly without communication from their engineers.

Each team was assigned a series of tasks: they included driving a vehicle, getting out of a vehicle, opening a door, turning a valve, drilling a hole in a wall, a surprise task such as pushing a button or turning on a switch, walking over rubble or debris, and climbing stairs. Robots scored a point each time they completed their missions. To win, a team had to complete all the tasks successfully in the shortest amount of time possible.

Team KAIST completed the entire course in 44 minutes and 28 seconds, followed by the Institute of Human and Machine Cognition (IHMC) Robotics in Pensacola, Florida in 50:26, and Team TARTAN Rescue of the National Robotics Engineering Center at Carnegie Mellon University in 55:15.

In summing up his team’s experience at the 2015 DARPA Robotics Challenge, Professor Jun-Ho Oh pointed out that competition was not the main purpose of the event:

DARPA said that the event is not a competition; they call it a “challenge.” A challenge means that we will try something impossible. We cannot reach the final goal, but the important thing is that we shoot the ball toward that goal. That’s the challenge. DARPA is just opening the ways to think and to find out where we are. They are trying to find the limit and the variety of our technologies, and they try to push those limits. So in that sense, the Challenge was quite successful.

From this perspective, it is clear that Team KAIST and DRC-HUBO+ were quite successful as well.