U.S. looks to improve remote killing, er, identification ability

The U.S. Air Force is spending millions of dollars to develop a new, technology-intensive intelligence discipline called Human Measurement and Signatures Intelligence, or "Human MASINT." MASINT has traditionally been the study of electronics signals — radio, radar, etc. — that provide information about enemy forces. Human MASINT is the study of personal and biological traits to determine whether a person is hostile or not.

The idea is to gather a mountain of information about a targeted person, such as their gait, motions, size, shape, "gestures and behaviors that would give clues as to ethnicity, role in a group, or possibly geo-political origin," and even the presence volatile organic compounds (which could indicate the presence of explosives) or biowarfare agents, according to an Air Force request for proposals. All of this information would be compared against a vast database of people and groups to see if a particular target’s characteristics match those of someone up to no good. The approach will be used to quickly identify threatening people, detect and disrupt terrorist groups, prevent friendly fire and track non-combatants, and help special operations troops in "global man-hunting" efforts, the Air Force says.

The problem is that the Air Force must first figure out just how to do all this. On August 9, the Air Force Research Laboratory gave Infoscitex of Waltham, Mass., a $46.9 million contract to help it develop technology that can "discover and exploit human biosignatures" from "stand-off distances." (The next day, Infoscitex was acquired by Alexandria, Va.,-based DCS Corp, a larger defense contractor that specializes in cyber security and intelligence-related systems.) The Air Force and Infoscitex plan to use video, infrared cameras, and possibly radar to examine what humans look like under a variety of circumstances. It will then take all of that information and form a "Global Biosignatures Information System" — a giant database that will be used by software that receives information from UAV cameras and other "exploiting mechanisms" to automatically detect and recognize "human signatures," according to the RfP. Add technology that can sniff out things like the presence of explosives and a database that tracks humans and links their connections and you’ve got Human MASINT.

The technology may be a long way off — the Infoscitex contract extends until 2030 — but it would represent a huge step toward providing the service with the ability to sift through the thousands of hours per week of video data collected by its spyplanes, drones, and satellites quickly enough to act on it. As any Air Force intelligence official will tell you, one of the biggest difficulties they face is making sense of the avalanche of information generated by the service’s eyes in the sky. If the Air Force can make Human MASINT technology a reality, it will automatically alert an intelligence analyst sitting in one of the service’s Distributed Common Ground System intelligence centers any time it spots a potential trouble-maker.

So, imagine you’re walking through crowded city streets somewhere the United States is conducting military operations. You’re hot, not just because the temperature outside is high but also because you’re lugging heavy bomb parts in a duffel bag; the effort and anxiety associated with hauling your deadly cargo is making you sweat. The weight of the explosives is also impacting your stride; you’re constantly shifting the way you’re carrying your load, sometimes slinging the bag over your back, making you bow slightly.

Normally, you might go unnoticed by the drone circling overhead with its regular and infrared cameras. But, thanks to Human MASINT, software at the intelligence center automatically notices that you’re hotter than nearby pedestrians; it also sees the massive duffel bag and notices that you’re straining to carry it, but can tell by your quick clip — despite the weight — that you’re probably young. It checks your estimated size against a database of wanted men in your city. It also notes that you’re walking through a known hotbed of bomb-makers. All of this happens in seconds. Sure enough, the system realizes you match the description of, well, you — a known militant — and it instantly flags you.

You can see where this is going. Someday, a UAV, or maybe a satellite, or even a ground robot spots a suspicious person and automatically identifies him as hostile. A person in the intelligence center is alerted to the situation. Commanders quickly approve action against the target, and a drone opens fire. It’s called "shortening the kill chain" — seriously.

Still, don’t expect such a seamless process to come together anytime soon, at least the killing part. In fact, the entire notion of being able to automatically identify a person from above seems pretty far-fetched to Teal Group analyst David Rockwell, who specializes in UAV payloads.

"It is still pretty difficult to have a high degree of certainty if a target is dressed as a civilian," said Rockwell. "There is a great deal of similarity between similarly clothed individuals, and UAV sensors do not yet have the resolution to do things like retina scans from altitude. If good HUMINT were added, that will be important, but again we’re talking likelihood of both mistakes and collateral damage."

The Air Force and Infoscitex did not answer requests for comment on the program.