More than 20 years ago, neurobiologists showed that the lobster's brain detects scent only while it flicks its antennae. New studies now reveal exactly how these flicks are responsible for the lobster's marvelous sense of smell. The research, based at the University of California at Berkeley, and Stanford University, is part of a joint effort at several institutions funded, surprisingly, by the U.S. Navy. The military is involved for one reason: It wants better robots for detecting underwater mines or monitoring toxic waste. And the crustaceans can show the Navy the way.

"If you want to build unmanned vehicles or robots that go into toxic sites, and you want those robots to locate something by smell, you need to design noses for them," says UC Berkeley researcher Mimi Koehl. It just so happens there is no better nose to imitate than the lobster's schnoz, which has had plenty of time to improve over millions of years of evolution. Engineers and biologists are teaming up to learn and steal from designs that nature worked out long ago, a field called biomimetics. Lobsters are not only master sniffers, but they've also adapted nicely to the rough conditions of surf break zones without getting washed away. Both features are crucial for underwater robots to succeed in hazardous coastal areas.

Koehl, a biomechanical engineer, wants to understand how the lobster's nose masters the challenge of smelling underwater. On a recent afternoon in her lab, she explained her work. "Most biomechanics researchers are the guys who develop running shoes or artificial knee joints. But some of us straddle biology and ask questions about non-human organisms."

Like an aerodynamics engineer studying the flow of air over the wings of an airplane, Koehl looks at the flow of water and odor molecules over the pair of antennae attached to the lobster's head. This flow brings odor molecules into contact with sensory receptors in each "nose." Unlike in mammals, where smelly molecules stream into the nostrils with every breath of air, lobsters must move their antennae to "sniff" smells dispersed in turbulent water.

Each antenna is about two inches long and splits into a Y-shaped structure with two pointy tips–"hairy little legs," is how Koehl tenderly describes them. Peering closely through a magnifying glass at one of these antenna tips, she glimpses a dense zone of hair tufts staggered in a zigzag arrangement. It looks like a miniature toothbrush. Each hair is covered with multiple nerve cells that can detect odors. Along the edges of the toothbrush, larger hairs line up like a long alley of tree trunks. Up to five times thicker and taller than the smallest hairs, these hairs control the flow of odor molecules and water to the shorter, inner sensory hairs. For that reason, researchers call them "guard hairs."

"To understand the physics of smelling," Koehl says, "you need to understand the fluid dynamics of water interacting with hairs." She holds up a scaled-up plastic model of the toothbrush-like array of sensory and guard hairs. "If you look at the array of hairs, it is full of holes," she says, poking her fingers into the spaces between the taller guard hairs. Nonetheless, water can't normally flow through these gaps; instead, it takes the path of least resistance and flows around the hairs. It is only when the entire, hairy array is moving fast enough–when the antenna is flicked–that water can flow through the guard hairs. Then, sensory hairs can encounter odor molecules and transmit the scent information to the lobster's brain.

To observe the flow of odors, Koehl uses another plastic model that's about 300 times bigger than the lobster's microscopic guard hairs. She places the model, mounted on a small motorized cart inside a large glass tank. To approximate the drag that occurs when seawater flows through such teeny hairs, Koehl's tank is filled with gluey Karo corn syrup. She adds tiny little red beads that float in the sticky sweetener, like odor molecules in water. Moving the plastic model at various speeds through the syrup creates a flow of the red beads across the guard hairs. Koehl records it all with a video camera.

At slow speeds, the beads merely drift around the guard hairs. But as the speed picks up, the hairs get leaky and let the beads pass through. "Models are powerful tools. You can systematically dissect and understand what role each part has," Koehl says with satisfaction. "With organisms, nature never does that for you."

Koehl's work has established the role of guard hairs as selective gates. Based on her experiments, she predicted a double role for the lobster's antennae: When flicked downward, they act as sieves that trap odor molecules. But when slowly moved back up, the antennae behave more like paddles, pushing water and odors away.

Much like cigarette smoke in the air, traces of scent released by fresh fish form a constantly changing cloud, or plume, in the water. A lobster trawling for dinner is never aware of this full cloud. It senses only tiny slices of the plume from flicking its hairy antennae repeatedly. These series of flicks through odor plumes fascinate Jeff Koseff, an environmental fluid mechanics engineer at Stanford.

Using laser technology, Koseff has developed a method to dissect the plume's structure. In lab experiments, he mixes odor molecules with a special invisible dye into water in a tank. Then he aims a thin sheet of laser light (rather than just a single beam) into the tank, illuminating one slice of the odor cloud. Dye molecules within this laser-lit slice give off fluorescent light, allowing Koseff to record the cross-section with a videocamera. In the resulting image, the dye molecules look like fine, threadlike filaments swirling about, reminiscent of the pattern of an oil slick on the surface of a pond. The picture gives Koseff an idea of what the lobster smells while flicking its antenna.

To find out more, working with Koehl, Koseff built a simple robot out of a molted lobster shell filled with plastic. They added this mechanical creature to his experiment, positioning one antenna to move within the sheet of laser light. With the camera, they recorded which parts of the thin filaments of dye penetrated the hairy brushes on the tips of the antenna. "It's a bit like placing toothpaste on a toothbrush," Koehl says.

For the first time, the scientists could measure what the lobster encounters with each flick. Their results showed that during the upstroke, when guard hairs push seawater away, the odor filaments retain the original shape with which they first entered the sensory brush. The next flick, downward, breaks up the filaments as the antenna captures scent molecules from the water. It stores the odor sample for about a tenth of a second–just long enough for sensory neurons to detect the smell, even as the antenna is already swinging upward again. On the next downstroke, the stored odor is replaced by a new scent sample. Each flick is like a deep sniff supplying new smells.

Together with a neurobiologist in Florida, Koehl and Koseff are now working on combining electrical recordings of the lobster's brain with real–time imaging of the plume structure. If successful, they'll see what smells the antenna is picking up and what the lobster's brain is sensing, simultaneously.

SCIENTISTS ELSEWHERE ARE WORKING on other parts that lobster-inspired robots will need. Underwater devices programmed to autonomously sniff out explosive underwater mines or toxic waste sites require some intelligence. And they need to navigate sand, stones, and rubble on the bottom of the sea. Two other research efforts are focusing on these goals.

At the Marine Biological Laboratory in Woods Hole, Massachusetts, neurobiologist Frank Grasso and his team has designed a robot to study lobsters' behavior in response to clouds of scent. RoboLobster, as it's called, doesn't actually look much like a crustacean. It's a two–wheeled vehicle about 30 centimeters long, featuring two smell sensors in the front and instruments to gauge its position. But its body size and shape are copied from the real animal, as are its speed, pattern of locomotion, and the way its sensors are arranged.

Grasso can program the way RoboLobster reacts to the fishy odor plumes it senses. In a typical experiment, he exposes a live lobster and the robot to the same conditions. Based on the differences in how they respond, he finetunes RoboLobster's programming to better mimic natural lobster behavior.

Grasso and RoboLobster recently returned from a field trip to the bottom of the Red Sea off Israel, where he exposed his baby to its first real-world test. A team of frogmen swum out and covered a portion of the sea floor with long sheets of linoleum so that RoboLobster wouldn't get stuck on pieces of coral or other obstacles. The frogmen then escorted the robot three meters underwater, generated a colored odor plume–and let it roll. It was the first time the mechanical crustacean was exposed to the turbulence of naturally occurring waves, but it behaved just as it did in the lab: As soon as odor molecules reached RoboLobster's nose, the vehicle started moving towards the source of the scent.

For Grasso, it was a terribly exciting moment, to see his invention working in the environment that originally inspired its design. "I felt like a real Indiana Jones-kind-of-scientist," he says with a grin.

While RoboLobster is good at sniffing out plumes, another of its brethren is proving to be a versatile roamer. In just five years, neuroscientist and engineer Joe Ayers and his team at Northeastern University in Boston have constructed a fully biomimetic lobster robot. Even without seeing the robot in action, a viewer of the eight-legged metallic critter has no doubt of its heritage. With its thin legs, two front claws, and a long tail, this vehicle has all the key anatomical features of a lobster. Ayers built in these features not for their natural appeal, but for their function. The claws and the tail, for instance, stabilize the robot while it crawls along the bumpy sea floor.

Within the mechanical lobster sits an electronic "brain" inspired by Ayers' early graduate work at the University of California at Santa Cruz. As a trained neurophysiologist, he's deciphered all the nerve cells in the center of the lobster brain that produce the crustacean's pattern of locomotion. To build the controller that steers his robot, Ayers created a computer model based on these neurophysiological measurements. This artificial nervous system can command the robot to move in all directions exactly like using feedback from the lobster's own walking patterns. "What is really unique," Ayers says, "is that the robotic lobster can change its walking behavior on a step by step basis."

Ayers and his team recently finished building the second generation of the robot. This version can walk entirely on its own, and without the cable support that its predecessor needed. Compact battery packs provide enough energy to keep it going for several hours. Some sensors help stabilize the metallic critter's balance, while others that detect touching and bumping guide it around obstacles. From a base station, Ayers can send directional commands to the robot via sonar communication. He even designed the vehicle so that additional instruments such as cameras can be mounted on the back of its tail.

Though Ayers' efforts were fully focused on creating a robot capable of running on the sea floor to hunt for underwater mines, his invention currently lacks a nose for TNT or any other explosives. Yet he says with confidence, "If the Navy combines Grasso's RoboLobster with our robotic lobster, they will completely solve their problem."

What the Navy wants is a robot that can track explosives in the 30-meter zone off a shoreline. "We think that a legged walker that can search would be the ideal device on the rocky bottom of a surf zone," says Joel Davis, the research coordinator from the Office of Naval Research. Loaded with a camera and explosives, a robotic lobster would search for mines along an area enclosed by sonar buoys. Upon finding a suspicious object, it would transmit an image to a human operator, who could identify whether it had found a real mine–and trigger the robot to explode to get rid of the threat. At $300 a pop, the Navy's self-destructing robotic lobsters would be a cheap way to make seashore operations safer.

Meanwhile, Ayers is turning his attention to biological challenges in robot design. Just like Grasso, he hopes to do real-world experiments exposing his creature to the lobster's original habitats. One day, these robotic lobsters may even start to invade the dens of their natural compadres. "Ultimately," Ayers muses, "our robotic lobster should be able to do all these things a real lobster does–except have sex."