RoboSimian finishes in fifth place — the robot is a collaborative partnering effort between the Jet Propulsion Laboratory / Caltech and UCSB’s Robotics Lab led by ECE Assistant Professor Katie Byl and her graduate students
Showing off its robustness and versatility, the ape-like RoboSimian robot, developed at NASA’s Jet Propulsion Laboratory, Pasadena, California, took fifth place in the DARPA Robotics Challenge (DRC) Finals, held June 5 through 6 in Pomona, California. RoboSimian squared off against 22 other robots in the international robotics competition, which promoted the development of robots that could respond to disaster scenarios too dangerous for humans.
RoboSimian and its competitors faced a variety of complex tasks during the tournament. Each robot had one hour to:
• Drive a vehicle through a slalom course and then exit the vehicle
• Open a door
• Turn a wheel to open a valve
• Cut a hole in a half-inch-thick panel of drywall using a cordless power drill
• Cross a field of debris or walk over uneven terrain
• Walk up a set of stairs
In addition to the tasks known in advance, DARPA officials gave competitors an additional surprise task each day. These included throwing a switch on an electrical panel and pulling a plug from an electrical socket and reinserting it.
Making the challenge even more difficult, the JPL group and the other teams faced degraded communications as they tried to control their robots. DARPA officials had planned this element of the competition to mimic the disorientation of a disaster scenario.
The DRC Finals are the culmination of a nearly three-year program to develop robots capable of assisting humans in responding to natural and man-made disasters. The challenge was launched in the wake of the 2011 earthquake and tsunami that devastated the Tohoku region of Japan, with the goal of better preparing humans to confront the threats posed by future disasters. Through two preliminary rounds of competition, DARPA and the DRC teams redefined what is possible in supervised autonomy, physical adaptability and human-machine control interfaces.