Artificial intelligence and machine vision for unmanned vehicle smart cameras is aim of DARPA Mind's Eye program
ARLINGTON, Va., 28 March 2010. Artificial intelligence and machine vision experts at the U.S. Defense Advanced Research Projects Agency (DARPA) in Arlington, Va., are asking industry to develop machine visual intelligence, which would enable machines to recognize not only objects, but also actions and how actions influence objects.
Posted by John Keller
ARLINGTON, Va., 28 March 2010.Artificial intelligence and machine vision experts at the U.S. Defense Advanced Research Projects Agency (DARPA) in Arlington, Va., are asking industry to develop machine visual intelligence, which would enable machines to recognize not only objects, but also actions and how actions influence objects.
DARPA released a broad agency announcement (DARPA-BAA-10-53) Thursday for the Mind's Eye program to develop in machines a capability that currently exists only in animals: visual intelligence.
Of particular interest to DARPA is to develop technology for a smart camera on man-portable unmanned ground vehicles (UGVs) with sufficient visual intelligence to report on activity in an area of observation.
The visual intelligence developed in this program would be applicable to a wide range of platforms such as vehicle-sized UGVs, perch-and-stare micro air vehicles, or fixed surveillance cameras, DARPA officials say.
DARPA scientists would like to see proposals from developers of visual intelligence software and from systems developers who would integrate these new algorithms to create an end-to-end smart camera.
Humans and animals perform a wide range of visual tasks with ease, which no current artificial intelligence can do in a robust way, DARPA officials explain. Humans have inherently strong spatial judgment, and learn directly from visual experience.
At the same time, humans can visualize scenes with objects and actions, and have a powerful ability to manipulate those imagined scenes mentally to solve problems, DARPA officials say. They want the same capability for smart camera-equipped machines.
Now apply this notion to ground surveillance, which human scouts in the Army, Force Recon in the Marine Corps, and other similar units normally perform. These units are specially trained, scarce, and work at great risk because they operate far from the main body. The current capability of ground forces to conduct persistent stare missions is limited by the number of ground forces available.
Military leaders expect a significant increase in the role of unmanned systems in future operations, including functions such as persistent stare. A significant capability area for unmanned ground systems is battlespace awareness that requires the ability to translate sensor data into a shared understanding of the environment.
An envisioned capability in that roadmap is a small robot with a sensor package that can navigate autonomously to a specified point and not only perform persistent stare in support of surveillance missions, but also detect and report on significant activity. In this way, camera-equipped UGVs would take human scouts out of harm's way.
Initial closing of this solicitation is 10 May 2010, and final closing is 21 Sept. 2010. An industry day for program briefings will be 20 April 2010 in Washington. More information on the industry day briefings is online at http://tinyurl.com/MindsEyeIndustryDay.
For questions or concerns contact James Donlon, the DARPA program manager, by e-mail at DARPA-BAAfirstname.lastname@example.org, by fax at 703-812-5052, or by post care of DARPA/TCTO, ATTN: DARPA-BAA-10-53, 3701 North Fairfax Dr., Arlington, VA 22203-1714.
More information is online at https://www.fbo.gov/spg/ODA/DARPA/CMO/DARPA-BAA-10-53/listing.html.
Join the PennWell Aerospace and Defense Media Group on Linkedin at http://bit.ly/9MXl9
Become a fan of Military & Aerospace Electronics on Facebook at http://bit.ly/1VGM0Q
Join your industry colleagues in the Command Post community online at http://community.milaero.com