DARPA Wants Robotics to Rise to the Challenge of Disasters The DARPA Robotics Challenge kicks into high gear today as the organization announces the top teams that will be competing to create robots that can prevent the compounding of human peril in manmade and natural disasters. Spurred by the Fukushima Daiichi nuclear disaster in which the "Fukushima 50" ventured into a nightmare scenario to prevent a nuclear meltdown, DARPA is seeking robotic substitutes so that one individual's life is not weighed against dozens or more of others'. Dr. Gill Pratt, the program manager for the challenge, noted that the program's focus on humanitarian assistance in disaster response is aligned with one of the 10 primary missions of the U.S. Department of Defense that was laid out by the White House and the Secretary of Defense in January 2012. But Pratt called attention to another reason why DARPA chose the subject of this challenge: "[W]e believe that this is very inspirational for participants because it's a universally understood and appreciated mission." The participating teams are divided into two tracks: Track A teams will create the robots themselves as well as the software while Track B teams will be provided with the Boston Dynamics-designed Atlas robot, the descendant of Pet-Proto, and create software for it. Both teams will receive DARPA funding for their projects and they have until approximately December 2013 to complete them. DARPA has chosen seven Track A teams from Carnegie Mellon University, Drexel University, Raytheon, SCHAFT, Virginia Tech, the NASA Johnson Space Center, and the NASA Jet Propulsion Laboratory. The 11 Track B teams are from Lockheed Martin's Advanced Technology Laboratories, RE2, the University of Kansas, Carnegie Mellon University, the Massachusetts Institute of Technology, TRAC Labs, University of Washington, the Florida Institute for Human and Machine Cognition, Ben-Gurion University, the NASA Jet Propulsion Laboratory, and TORC Robotics. To open up the pool of talent to the challenge, which is already an international competition, DARPA has added two tracks. "The reason for doing this is to open the aperture as wide as possible, so that we can gather teams that are not in traditional robotics fields, that may not have been used to sending in proposals to DARPA, maybe from other parts of the world that aren't used to doing business with us," Pratt said. "We really want to see what talent is out there and find the best possible talent." Participants in Track C won't receive DARPA funding up front but should they perform better than the Track A and Track B teams, they might be eligible to receive one of six Atlas robots to work with and funding to progress further in the challenge. Track D is for those who do not want to take any DARPA funding at all but would like to compete anyway. Robots will be tasked with driving a vehicle (under the supervision of an operator at a distance), opening a door, climbing up a stairway, and connecting a cable or a fire hose. "We understand that where disasters matter the most, the environment generally has been engineered for use by human beings and even if the environment is degraded, it is not a random, unstructured outdoor environment," Pratt said. "In fact, it's one that has a lot of things made for people to use; for instance doors and stairways and ladders, things like that. And so the robots that we're developing, the technology that we're developing here, are strongly geared toward operating in environments that have been originally engineered for human beings." The robots will also be required to tackle one of the things that separates humans from many other species: figuring out tools. "Disasters often occur in environments that have plentiful numbers of tools that are meant for human beings," Pratt said. "And by tools, I mean things from screwdrivers to vehicles and everything in between. And often those tools are around both for construction and repair and maintenance. And, again, during the first few days after a disaster, there is no time to bring in specialized tools and, so, can we build robots that can reuse tools that were originally meant for human beings?" Of particular importance in the challenge is human-robot interaction. "Often in a disaster, the experts who know how to handle the disaster are not robotics experts," Pratt said. "And so we want to shorten the amount of time and the amount of work that needs to be done to allow those people to be able to directly use the robots in the disaster." Pratt again cited Fukushima, where robots were rapidly brought in from Japan and the United States, but the plant operators needed to spend several days in training to learn how to use them. The ultimate prize for the team that best meets DARPA's requirements is $2 million, and it will be handed out sometime near December 2014. Additionally, DARPA projects often end up extending into the consumer realm. One of the most nascent examples is the Google self-driving car, which came out of a 2005 DARPA Grand Challenge win for a team led by then head of Stanford Artificial Intelligence Laboratory and Google Vice President Sebastian Thrun. Beyond the creation of a task-driven robot, the robotics challenge has higher-level goals that involve the robotics field as a whole. Pratt said DARPA is investing in an open-source simulation package to advance the state of the art in robotic simulation. "In particular, we want to have these tools outlast the program and be the foundation for catalyzing the field of robotics, particularly helping to make the design of robots move from an art to a science," he said. DARPA is also funding the Open Source Robotics Foundation to further develop a preexisting simulator that will use cloud computing for quick and easy scalability.