Everybody's been blogging the robot basilisk lizard ("Jesus lizard") which was recently built by the nanorobotics lab at Carnegie Mellon. Make Magazine also took the helpful step of linking to how real Jesus lizards run on water. Turns out it's not super sneaky: just slap the water hard enough to boost yourself up a while, and as you push mostly backwards for locomotion, also push to the side a bit to stay balanced; then do it again with the other foot. It's mostly a matter of being light, fast, and having wide slappy feet. (Though to be fair, pushing sideways adds an element of control not required for many robot legs, and balancing on a squishy medium like water instead of a hard surface has got to be a nightmare to control.) It's just a proof of concept, but ultimately they envision bots like this having features like "biochemical sensors that monitor water quality; deployed with cameras for spying, search-and-rescue or exploration; or outfitted with bacteria to break down pollutants." I think the bio-sensors and bacteria are an excellent idea, particularly for something that automatically stirs water up as it moves. I might also suggest mine-clearing, although anything light enough to run on water might not be heavy/strong enough to set off a mine. But spying? "Hey, Ahmed, what's that wet slapping noise?" I don't think so.
However, this is not the only fish in the sea, as it were. There are a few other robots in labs that can locomote both on land and on water: a snake, a water-strider, and a crab. Each one is exciting in its own ways...
If quiet and subtlety is what you want, the nanorobotics lab is doing that, too, by mimicking a water strider. This tiny bot has just a half-inch body and two-inch legs, and according to Wired was built onlya year after the first researchers discovered how real water striders move on water. It turns out they're not swimming or paddling, they push the surface of the water to bend it, then surface tension "bounces back like a trampoline to push the insect forward." Since the water strider robot cost just $10 in materials (though models with sensors & brains would cost more), this little beast would be a perfect candidate for a swarming robot. Cheap, distributed sensing across bodies of water could map effluent plumes from factories, temperature gradients, oxygen or other nutrient levels, toxic spills, any number of things. I doubt they would work in the ocean or other highly-splashy bodies of water, because of the surface tension they require (they don't float, they just stand on water), but there are myriad applications for them in lakes, rivers, and pools.
If you want something quiet, big, and gloriously creepy, you need to leave Carnegie Mellon and go to Japan, where NEDO (the New Energy and Industrial Technology Development Organization) has funded the creation of an amphibious robot snake. TechEblog has a great video of it going from land to water and then swimming--very eerie--but I haven't been able to find a translation of NEDO's page about it, so I don't know the details. It has its own camera and ring of LED headlights, perhaps other instruments too, and is intended for helping to locate disaster victims trapped in rubble (though I suppose it won't be long before someone uses it for warfare / espionage). Can anyone write a quick translated summary of the NEDO page in Comments?
Of course, no discussion of biomimetic robots would be complete without mentioning the folks at Berkeley in Robert Full's PolyPedal lab and their colleagues at Stanford. Their contribution to amphibious robotics is Ariel the robot crab, which walks only on solid ground, but is equally happy doing so underwater in churning surf. It imitates various aspects of crab physiology and behavior to handle turbulent waves that would tumble away any normal robot, wheeled or legged. It also does something that real crabs can't: if it's flipped over onto its back, it simply bends its legs the other way so that its bottom becomes its top. This shoreline crawler has a useful niche that none of the other robots above could likely fill. For instance, it might be particularly good for clearing mines in otherwise impossible-to-search locations.
No doubt this is just the beginning of amphibious robot design; perhaps ten years from now, no robot will qualify for search-and-rescue or surveying operations without being water-capable. These designs will also no doubt inform devices in other applications as well. Perhaps micro-robots walking through your bloodstream to unclog arteries will use crab tricks to avoid being swept away by your blood flow. Perhaps special power-assist boots could be made to allow people to run on water. Who knows?
I'm not sure if it's actually biomemetic, but there's also a robot carp in japan:
" A carp robot, jointly developed by Ryomei Engineering Co Ltd and other companies, can make water quality inspections and will be set out for examination of seabed resources and bottom feeders in the future.
The robot is modelled after a Nishiki koi carp. Though the tail movement is very smooth and lifelike, the remote-controlled koi is capable of moves that a genuine koi is unable to perform, such as swimming in reverse and rotating in place. "
note the youtube.com video seems to be different than the carp in the story.