PORTLAND, Ore. Robotic fish could swim in schools of hundreds to perform surveys, environmental monitoring, reconnaissance and other underwater tasks, according to Massachusetts Institute of Technology researchers who recently displayed prototypes.
Measuring just a few inches long, the robotic fish combine flexible polymers with microprocessor controllers to swim, observe and return to report their findings.
"If you use traditional materials like pulleys, cables and gears, you end up with a very complex and expensive mechanism that has a high probability of failing," said MIT researcher Pablo Valdivia Y Alvarado. "We wanted to make robotic fish that were cheap, robust and resilient in the real world, so we enclosed everything in a flexible monolithic body with no parts that can break loose."
The polymer compounds used to make the fish were of variable stiffness in different sections to perform the functions of discrete components. MIT's original design back in 1994 had over 2,000 components, including six motors. Other researchers have continued to design similar robo-fish using traditional materials, but the MIT researchers took a cue from the design of modern prosthetic limbs to make their robo-fish cheaper and more reliable by virtue of reducing the number of moving parts to just 10, including a single motor.
Some prototypes have survived in the lab for four years of constant underwater tests without a leak.
Real fish swim by contracting muscles on either side of their bodies to generate a wave that travels from head to tail. Robotic fish simulate the way real fish swim by using the centrally-located motor to initiate a wave that travels down the body, propelling it forward. The monolithic construction of their flexible bodies makes them very maneuverable.
Since radio communications underwater are limited, even with powerful tranceivers, the MIT engineers proposed to instead release the robo-fish in schools of hundreds, depending on them swimming back home to report any findings. The schools would perform such missions as sensing each other with visual cues and precision pressure sensors that allow the robo-fish to "run" together.
"The way that real fish sense each other, besides vision, is by detecting the pressure in the water around them, which tells them how the body of the school is moving with very high resolution," said Alvarado.
Long-range control over radio channels will not even be attempted, according Alvarado, due to the poor transmission of radio signals in water. Instead, the robo-fish will communication with each other by touching in order to pool data with direct connections. Other observations will be recorded for uploading when they return.
"Underwater radio is very difficult, so what you want is total autonomy," said Alvarado. "The fish will come back to you carrying the data they collected while swimming--and if some don't make it back its OK because they were relatively inexpensive to build."
MIT has developed two basic models: One about five inches long that mimics carangiform locomotion used by bass and trout; the another model is about eight inches long and mimics the peduncle motion of a tuna for faster bursts of speed.
Top swimming speeds have reached one body length per second, about 10 times slower than tuna.
The researchers are testing other types of underwater locomotion, as well as reducing power requirements to enable adding on-board batteries to power the fish (current prototypes consume 2.5 to 5 watts, which is supplied by a tether). The team is also experimenting with different designs that mimic the locomotion methods of other sea creatures.