Snipers have always been the soldier's bane. Today, the problem is especially acute for U.S. troops engaged in urban combat in Iraqi cities. When gunmen hidden inside buildings shoot at troops, they're abetted by echoes coming off surrounding buildings, which obscure the source of the shots.

But help is on the way.

In Boston last week, a crowd gathered in the noise-filled, two-story atrium of the Boston University Photonics Center to watch a man simulate gunshots by banging on a well-dented metal panel. Each time he repositioned himself -- on a set of stairs, for example -- and struck the metal, a small, suitcase-sized robot would instantly swivel its cigar box-shaped head and aim two clusters of bright-white LEDs at the metal panel.

The robot's all-important head, named the Robot Enhanced Detection Outpost with Lasers, or Redowl, is the creation of a team at Biomimetic Systems in Rosindale, MA, led by Socrates Deligeorges, and professors at BU. It's the most recent of several new technologies to help troops quickly identify the source of enemy attacks.

BBN Technologies of Cambridge, MA, for instance, has already sent more than 100 Humvee-mounted devices to Iraq for detecting gunshots. Their system includes a two-meter mast with a half-meter microphone cluster, and weighs about 25 kilograms. Radiance Technologies of Huntsville, AL, sells another gunshot detection device, a box about the size of the BU robot, but weighing nine kilograms.

But Redowl stands out because of its small size and much lower weight. The acoustic detector, mounted as the "head" of the robot, fits into a box the size of a hard-bound book. And its developers say a new version using digital electronics should fit into a space the size of two cigarette packs. The entire Redowl system, excluding the robot it's mounted on, weighs about two kilograms, around one-quarter the weight of the Radiance device.

The Redowl system can be so small and light because it's composed mainly of electronics and software. Deligeorges says he developed the system to mimic the human auditory system, which can detect the direction of a sound using two natural sensors, the ears, located just a few centimeters apart.

"Our external ears, our ear canal, and our middle-ear bones, like the ear drum and the associated pieces, all perform processing that enhances certain features of a sound," says Deligeorges. He and his advisors at BU, biomedical engineer David Mountain and electronics engineer Allyn Hubbard, studied each part of this mechanism step by step. Then they built "a very intricate mechanical model" of how the ear translates pressure waves in the air into neural signals.