The new challenge, which will launch in October, asks teams to build robots that can navigate disaster-response scenarios. Robots will be required to complete a number of discrete tasks, including traveling across rubble, removing debris from a blocked entryway, climbing a ladder, and entering and driving a car. Responses from prospective entrants are due by the end of May.
Humanoid robots will likely be among the challenge entrants. Though DARPA indicates that a humanoid form isn't required, the agency will be providing "a robotic hardware platform with arms, legs, torso, and head" to some entrants, and it included in its solicitation materials a graphic of humanoid robots cleaning up an industrial disaster.
DARPA is no stranger to scientific challenges or to robotics. DARPA's past challenges have involved driverless cars, the hunt for red balloons hidden around the continental United States, the reconstruction of shredded documents, and the hunt for quick response codes hidden throughout the Internet. As for DARPA's robotics chops, a DARPA robot recently set a speed record, and DARPA recently built a relatively autonomous robot arm as well as an inflatable robot.
[ Read about DOD's call to developers to strengthen K-12 science education. See DOD Challenges Developers To Improve STEM Education. ]
"For robots to be useful to DoD, they need to offer gains in either physical protection or productivity," said DARPA acting director Dr. Kaigham Gabriel, whose own research has included forays into robotics, in a statement. "The most successful and useful robots would do both via natural interaction with humans in shared environments."
DOD plans to use the research and technology developed in this challenge to help improve the ability of robots to navigate rough terrain at disaster sites, operate vehicles, and use common tools, as well as to make robot hardware and software development "more accessible." This in turn, according to DOD, could help mitigate damage in future natural disasters and industrial accidents.
However, applicants need not apply if their robots represent only "evolutionary improvements," according to the agency's solicitation; DARPA is looking for cutting-edge developments only. "Proposed research should investigate innovative approaches that enable revolutionary advances in science, devices, or systems," the solicitation says. "Specifically excluded is research that primarily results in evolutionary improvements to the existing state of practice."
The tasks robots will be required to perform are complex. For example, the robot must be able to get into an unmodified utility vehicle, drive it across a course that includes curves and obstacles, stop the vehicle, and get out. It must also be able to find a leaking pipe and nearby valve and close the valve. According to DARPA, the agency will intentionally alter details of the disaster scenario so that teams won't be able to optimize their robots "for a small range of conditions."
The challenge will break up entrants into four separate tracks, some of which will be supported by DARPA (which has available $32 million in funding) and some of which will use government-provided equipment. The challenge will take place in two phases and won't be complete until the end of 2014.
InformationWeek's 2012 Government IT Innovators program will feature the most innovative government IT organizations in the 2012 InformationWeek 500 issue and on InformationWeek.com. Does your organization have what it takes? The nomination period for 2012 Government IT Innovators closes April 27.