Robots mimic Mother Nature 5-18-98

May 18, 1998 Design News

ROBOTICS AND AUTOMATION

Robots mimic Mother Nature

Biological modeling and technology integration take robotics 'where no man has gone before'

Rick DeMeis, Associate Editor

Mention robots and most people will think of machines from robust one-dimensional industrial units on the factory floor all the way to the personable piece of machinery in the classic TV series "Lost in Space." But today, many robotic technologies are being highly integrated--many on what is termed the micro level, leading to autonomous devices measured in a few inches or feet. And many of these devices are intelligent machines patterned to look and swim like fish or fly from the palm of your hand. Such machines could survey and monitor water and air quality and wildlife, or serve as the unobtrusive eyes and ears of military forces.

Once you scale systems way down, integration becomes critical as weight, volume, and power become limited. Often, traditional large-system characteristics, particularly in fluid dynamics, are modified "in the small." Skin friction goes up for inch-sized fish or bird-sized aircraft, and laminar flows predominate.

But what engineering tricks can we learn from trying to mimic nature's systems? Quite a lot, according to Jamie Anderson, principal investigator for the Charles Stark Draper Laboratory's (Cambridge, MA) robotic fish prototype. "Fish can corner, zigzag, approach something very closely, and then rapidly maneuver away." And to investigate such versatility and potential applications, Draper is running a proof-of-concept test bed, the Vorticity Control Unmanned Undersea Vehicle (VCUUV), which, for obvious reasons, is called the robot fish or robot tuna.

Go fish! Once the preliminaries of piscine propulsion have been plumbed by the prototype robotic fish, it offers many practical possibilities. Obvious is its use as a scientific exploration platform. Well suited as a drone doing repetitive tasks, such a vehicle can survey fish stocks or do environmental monitoring. The latter could be accomplished in waves, turbulence, and turning rivers of various depths, even to the point of entering outfall pipes.

Military applications include a reconnaissance platform, mine detection and countermeasures, and locating quiet submarines. Right now military commanders are reluctant to commit to autonomous underwater systems because of limited onboard battery propulsion capacity, notes Anderson. "A change in the propulsion paradigm, along with new levels of energy capacity, may see systems that could go miles up a river and sit in a pipe to monitor a potential chemical weapons plant."

Smart skin materials may allow a more fish-like lateral line of sensors along the sides of a robot vehicle as well as propulsion movements. For the latter, piezo elements or developments of polymer/metal artificial muscle tissue from recent medical advancements, may see use. And finally, highlighting the critical importance of high propulsion energy density in underwater applications, Anderson looks for exotic battery materials with good form factors for packaging, and perhaps small fuel cells, as always welcome.

Fish tale. "Vorticity control is the magic that fish do compared to conventional propulsion," notes Anderson. Their handling of this rotational energy for propulsion using the tail, which is highly integrated with a low drag body, gives an energetically beneficial result--so much so that some researchers estimate fish require only about half the power to achieve the same speeds as a propeller-driven vehicle. Scientists hope to use the tuna to put hard numbers to such judgements.

According to Anderson's previous research at MIT, a fish wake has natural instabilities. "In the past, an unsteady wake was considered a 'bad thing,' " she notes. "It has a natural instability at the frequency that the fish drives it--efficiently transferring momentum to the wake with minimum input energy." Fish can turn within their body length while conventional vehicles require several lengths. Anderson has first-hand experience in studying fish motion gained by observing tuna during tank dives as a volunteer at the New England Aquarium. "In the fish we have propulsion and steering in one device. It takes up more of the hull than a propeller, but it adds a lot more [turning performance and energy efficiency]."

The robot tuna will directly fathom propulsion parameters by measuring fluid forces on an unconstrained swimming vehicle. The prototype program is based on tow-tank work done with smaller robots at MIT. "We're standing on their shoulders," adds Anderson. The proof-of-concept free swimmer is eight feet long and weighs 300 lb. "There's a lot to float in the boat," she notes, "and to advance the technology, we needed autonomy, which drives up the size in order to accommodate the necessary batteries, hydraulics, and computer power." The vehicle will run for at least three hours at four knots.

Tune a fish. Anderson says they saved time and money by reverse engineering one of Mother Nature's working systems: a yellow-fin tuna. "The swimming movements of the tuna are very localized to the tail," she adds. Their amplitude is 10-15% of the body length and only in the very end portion of the fish. In the robot, "The whole forward package is like a big bus, you can put all your stuff in there. Designers don't like to have a vehicle moving from side to side like eels or other fishes."

To gauge a tuna's shape, an ungutted 40-inch, 34-lb tuna was shipped on ice to Draper. "We studied the fins and how they were attached, particularly the caudal fin area, near the tail, where the actual pivot point was," she says. A plaster cast of the fish served as the mold for a rigid urethane replica. This plastic fish was digitized and the data set given to Goetz Marine Technology (Bristol, RI), a builder of composite components. It constructed the robot's pressure hull from carbon/epoxy. The fish's nose is a soft urethane bumper, but in the future this could contain sonar.

In configuring the robot, Anderson cites deciding the ratio of "hull to wiggly tail. Right now one half is forward pressure hull." The tail deflects "a lot farther than a real tuna tail. We really wanted to push the envelope for maneuvering," she says. The tunabot is designed to bat-turn 180 degrees in one body length. Engineers will study "exactly how it moves as we wiggle it--observing the input/output relationships between how you wiggle and where you end up--for transfer functions and system performance," she adds.

To approximate the fish's tail, with its many tens of vertebrae, the robot's tail, which fills with water, has four articulated links, each hydraulically driven. Investigators will study and optimize swimming with the arrangement, right down to driving just one of the highly flexible segments. The composite structure, with aluminum fittings and members, has flexible beams and ribs, as in the fish, including the tail links. About 80 lbs of lead facilitate trimming and submerging. Laminated fiberglass scales between the gaps in the tail support a flexible neoprene skin with a nylon outer layer. The robot's pectoral (side) fins are somewhat larger planes for low-speed diving and climbing. Dorsal and ventral fins act as keels for straight-line stability.

The robot tuna was fabricated in 16 months, after a 4 month preliminary design period. Because time and money were limited, commercial, off-the-shelf (COTS) component use was mandatory (see For More Information box). Exceptions were the Draper designed and made hydraulic manifold block, and the eight hydraulic tail actuators, which were fitted with MicroStrain (Burlington, VT) external position sensors. The Draper team conducted an adroit eight-week design and fabrication effort on these actuators--because the only qualified commercial vendor failed to deliver ordered actuators, with internal sensors, after 20 weeks.

The brain of the animated tuna is "a 486 processor with a gigabyte hard disk to store data," Anderson adds, along with other distributed computing and sensing processors, such as for actuator control. "Networked sensors talking to each other are another key feature." Safety systems allow manual, remote, or automatic recovery. The computer shuts down operations if sensors detect a leak, or overheating, or there is no systems response.

Components that would allow advanced uses mentioned previously might include a "quick-look" GPS satellite navigation system for pop-up position fixes. Anderson cites a wish list, including a high-bandwidth acoustic modem for reliable communication in many environments. "Some modems are available today, but only in limited applications and at significant cost." Multipath signals in shallows, caused by many surface reflections, are the gremlins here. And although optical based communication and sensing is being looked at, Anderson notes, "Whales are capable of talking across oceans, rather than looking at each other nearby."

"There are so many possibilities in robotics and autonomy," Anderson concludes. The VCUUV is a propulsion and maneuvering platform and "guinea pig for demonstrations of new sensors and advances in autonomy. The goal is a user-friendly, intelligent system so someone can plug into the fish's computer and test using real hardware."

Mad about MAVs. Just as mechanical fish gotta swim, electronic birds gotta fly. This May in Florida, the International Society of Structural and Multidisciplinary Optimization (ISSMO) and the University of Florida (Gainesville) Department of Aerospace Engineering are having their second annual Micro Aerial Vehicle (MAV) Competition (For details, http://www.aero.ufl.edu/~issmo).

MAVs are substantially smaller than current remotely piloted aircraft, with efforts aimed at producing vehicles on the order of six inches in wingsp