Just over a year ago, the head of the Pentagon's esteemed research agency called an urgent staff meeting. Snipers in Iraq were killing U.S. soldiers. Often, troops in noisy Humvees didn't even know they were being shot at until someone was hit, said Tony Tether, director of the Defense Advanced Research Projects Agency. So could DARPA's famous brainiacs concoct an inexpensive system to spot incoming bullets instantly and pinpoint where they came from? If the system saved just one life, it would be worth it, Tether told the group.
So began a crash project that not only has saved lives but also offers larger lessons about technology development. Research teams in France, Canada, and elsewhere have developed bullet-tracking systems. But the Pentagon wanted one capable of functioning on the move and cheap enough and small enough to place on many vehicles, enabling troops to return fire quickly.
It wasn't an easy task. "It's hard to innovate on command -- it takes a certain breed of craziness," explains Robert G. "Tad" Elmer, president of BBN Technologies in Cambridge, Mass., which succeeded in building the system. In this case, it required a tough deadline, a team approach, extensive feedback from the troops, and new ideas. As an improved version of the system now undergoes its final tests, it also illustrates how the demands of war are a powerful spur to progress.
A longtime DARPA research and development contractor and builder of the precursor of the Internet, BBN had won a DARPA contest for a cruder stationary bullet-detecting system in 1997. So when Tether issued his challenge, DARPA called BBN. Could the company create a system that worked on the move? In 60 days? And deliver 50 working units? The answers, recalls BBN Chief Technologist Steve Milligan, were "yes, yes, and can we call you back Monday morning?"
By Monday, Milligan thought it was possible, and on Nov. 17, 2003, BBN signed the contract with DARPA. The company's researchers knew that the physics was straightforward. Like supersonic airplanes, bullets create shock waves -- mini-sonic booms -- as they speed through the air. So if engineers arrange seven microphones like the spines of a sea urchin, a shock wave from a bullet will hit each microphone at a slightly different time, like a wave lapping against different pebbles on a beach. By measuring those time differences, it's possible to calculate the trajectory.
As it turned out, "it was much harder than we thought it would be," says Milligan. Not only is the math difficult, but the system also had to work in the cacophony of urban warfare, including echoes from shots -- and do it on the move.
The first step was hastily assembling the seven-microphone array and a van full of electronic equipment over the Thanksgiving weekend and driving down to Quantico Marine base in Virginia for a week of collecting data from actual shooting. The conditions were horrible -- "snowy, freezing, and rainy," recalls DARPA Program Manager Karen Wood. Once it had hundreds of shots' worth of data, BBN's team tackled the software. It was Milligan who came up with the breakthrough, recalls Wood, employing programs known as genetic algorithms. These start out with equations that offer potential solutions, then mutate them repeatedly, like evolution on fast-forward. Ultimately, a solution emerges that fits the observed data almost perfectly. "We can run what amounts to thousands of generations in a fraction of a second," says Milligan.
Then his team romped around Cambridge in a pair of Humvees to tune the system so that it wasn't fooled by normal urban bangs and jolts. They cajoled suppliers into making special production runs of key components. And everyone pitched in to screw and solder 50 units together.
BBN missed the deadline by just six days. By early March, 40 of the systems, dubbed Boomerang, were installed on vehicles in Iraq. They did save lives, but there were glitches. The biggest problem was that radio signals confused the units. Marines also thought the array was too big a target.
So the company went back to the drawing board to create an improved Boomerang. Engineers shrank the array to one-third of its former size, shielded the unit from radio interference, improved the display, and added networking capabilities and the ability to tell soldiers the elevation and range of a sniper, in addition to just the direction. In December, the new version sailed through radio interference testing. In the first week in January, it passed a durability test. The week after, it aced the performance tests. "I had no idea we were going to end up being as successful as we were," Wood says.
The next step will be building a set of Boomerangs to help the troops. And as the needs of war push the development of technology, the longer-term plans are more ambitious. Linking Boomerangs together in a wireless network will substantially extend their range. Similar systems might help defend helicopters. And there are plans to combine sniper systems with the ability to fire back, as in a Navy/Marine project called Gunslinger. One vision is a completely robotic vehicle capable of patrolling city streets, spotting and responding to any snipers in its path. It all demonstrates how rapid innovation amid the urgency of war can protect soldiers in the combat zone.
By John Carey in Arlington, Va.