DARPA Contest Seeking Humanoid Rescue Robot

The Defense Advanced Research Projects Agency’s latest robotic challenge calls for a human-shaped robot to carry out a series of search-and-rescue tasks.

The Defense Department agency in the first round of competition in December 2013 had teams from all over the world put bi-pedal robots through an obstacle course influenced by the Fukushima Daiichi nuclear disaster in 2011.

A robot going into a radioactive disaster zone has to remove and manipulate objects that are designed for humans, literature describing the program stated.

Yet, Gil Pratt, DARPA robotics challenge program manager, said search-and-rescue robots that can function in toxic environments in the place of first responders are not necessarily the competition’s ultimate goal.

“To be clear, we do not expect to exit the [challenge] with fully capable disaster response robots, but we will have made sufficient leaps forward in many vital technology areas such that others in the U.S. government, commercial industry, and elsewhere have the confidence they need to make the additional investments required to refine the technology and turn it into a finished product,” he said in an email.

One of DARPA’s first contests open to the public was the Grand Challenge in the mid-2000s, which had autonomous vehicles driving on a closed city course. That contest sparked interest in self-driving cars. Pratt said the state of humanoid robot development is similar to where autonomous vehicles were a decade ago. There were many institutions working on bits and pieces of the technology, but not in any coordinated way.

“DARPA determined that the …  piecemeal model of robotics research and development would not deliver the technologies required for disaster-response robots anytime in the near-term — certainly not in time to mitigate the effects of the next natural or man-made disaster — so we set about to tackle several of the technical challenges at once.” 

Teams competing can either supply their own robots capable of carrying out the series of tasks, or use a humanoid machine provided by DARPA and built by Boston Dynamics. That’s because some teams are better at software development than the mechanical side of the problem.

“Before we started the [challenge], robots were limited in their application to defense missions and civilian tasks in similar ways,” Pratt said. “Existing robot systems tend to be complicated to operate, have highly specialized and limited functionality, possess extremely low levels of autonomy in perception and decision-making, and suffer from limited platform mobility, dexterity, strength and endurance. Arriving at a robot with the capabilities required for complex missions such as disaster response means overcoming all of these limitations.”

Pratt said the first round of competition that took place in Miami in 2013 is already paying dividends. It catalyzed the field of humanoid robotics, inspired a wave of new investment domestically and internationally, and expanded the horizon of what’s possible, he said.

There is one indication that a search-and-rescue robot may be closer to reality. Team SCHAFT, a start up from Japan and the first-place winner in the opening round, was bought by Google a few weeks prior to the competition. It won the contest handily, according to media reports and its robot performed the eight tasks nearly flawlessly. The team has withdrawn from the upcoming competition in order to commercialize its technology, DARPA said.

While there may be many applications for humanoid robots, search and rescue is a mission of universal relevance and urgency, Pratt said.

“We expect to see even greater progress on display at the [DARPA Robotics Challenge] finals in June 2015 when we test new areas such as cloud robotics,” he said.

The winner of the finals, to take place in Pomona, California, June 5-6, will take home a $2 million prize.

Topics: Robotics, Science and Engineering Technology, DARPA