A lot of marine scientists are more than happy to spend time in the ocean, placing instruments, taking samples, and getting a firsthand look around. Mark Patterson, an affable Harvard-educated professor of marine and environmental sciences at Northeastern University, takes it a bit farther. He's lived on the seabed for a grand total of 87 days. Yet Patterson knows that sometimes there's no substitute for a good remote-controlled robot, like the AUVs that he and his lab colleagues have designed and built. The acronym stands for autonomous underwater vehicle – basically an ocean-going space probe.

"All these robots are taxicabs for sensors," says Patterson, the father of an AUV aptly called Fetch that gets programmed to go out and retrieve data. This torpedo-shaped AUV is about 6 feet long and weighs 220 pounds. Actually, with a new robot rolling of the line, there are now seven in the Fetch family.

Rigged with the necessary sensors, Fetch bots have been dispatched to get krill counts in the southern ocean, cruise over a coral reefs in Florida, and assess oxygen levels in various seas. But different types of data collection call for different AUVs. The new Robolobster, a creation of Patterson's Northeastern University colleague Joseph Ayers, has eight plastic legs, fiber-optic antennae, and an industrial-strength plastic shell. It not only resembles a lobster but scuttles around like one.

Now Ayers is working to imbue Robolobster with real lobster behaviors, a field known as biomimetics, essentially by using electronic neurons to construct a simulated nervous system. The thinking: The more Robolobster can adapt and maneuver like a real lobster, the better chance it has of carrying out its missions in the unpredictable and ever-changing ocean. This kind of neurotech could help an AUV like Fetch, too.

"If you think of the nervous system simulation as a robot controller," Ayers says, "that's a way to think about what I do."

Or just imagine Dr. Frankenstein as a marine biologist.