June 29, 2011
Catholic University Wins $4.94 Million for Artificial Intelligence Software
Pamela Clark, physics research professor
Catholic University has won $4.94 million from the Defense Advanced Research Projects Agency (DARPA) to develop artificially intelligent, Web-based software for configuring multiple spacecraft into clusters capable of performing different jobs as needs evolve.
In addition, the software could gauge options and determine the best approaches for future, possible space missions like the exploration of Mars.
The proposed software represents a “game-changing” shift in artificial intelligence (AI), the architecture of computer systems and engineering, says School of Arts and Sciences Dean L. R. Poos.
“This is a truly breakthrough concept. No one has done this before,” notes Poos.
Pamela Clark, Catholic University physics research professor and research faculty member of the University’s Institute for Astrophysics and Computational Sciences, is head of the research team.
At the request of DARPA’s Tactical Technology Office, Catholic University is leading the project through the agency’s System F6 program, short for “Future, Fast, Flexible, Fractionated, Free-Flying Spacecraft United by Information Exchange.”
The agency is eager to find a way to employ groups of spacecraft – like satellites – for a multitude of jobs for changing needs, not just for one function, as with conventional spacecraft, according to Clark.
The proposed AI software tool called, “Frontier: Design Tool for Stably Adaptable Systems” could configure spacecraft into complex systems, enabling their functions and interactions to be modified.
Using a synthetic neural system, Frontier would glean and analyze information from human input and the Web and explore and narrow options, until reaching a solution. The process, performed many times, “trains” Frontier until it “knows” how to make connections unseen by humans beforehand, Clark explains.
For instance, based on past performance of aerospace equipment, investment costs for new technology, and other information, Frontier could see patterns in what’s happening and figure out the best actions to take, she says.
The software would be able to determine whether it’s more cost-effective to build a base on the moon for human exploration or use robotic rovers or, as a project evolves, a combination of the two – and then also, “what makes the most sense for giving us capability down the road for going to Mars,” Clark says.
Moreover, the software tool is expected to have myriad applications in addition to space exploration, for instance, tracking relief efforts in disaster areas or changes in Earth’s climate and expanding communications capabilities in developing countries.
In fact, the project will engage a wide range of stakeholders. Barry Yatt, associate dean for undergraduate studies and professor in Catholic University’s School of Architecture and Planning, will facilitate their collaboration using techniques that civil architects, as opposed to software architects, employ when preparing design concepts for a new public project.
Scientists from the University of Maryland, George Mason University, and several private firms will serve also as team members in the work.
A prototype is scheduled to be finished in six months. The project is expected to be complete in 2 ½ years.