|
The RoboTuna, designed and built by David Barrett, is a
revolutionary undersea vehicle, with a flexible hull, propelled by an
oscillationg foil. It is, in effect, a carefully instrumented,
precisely engineered robotic fish and was developed to explore the
hydrodynamic mechanisms employed by a biological fish for
propulsion. It is unique not only in its biologically inspired design,
but in its ability to learn to swim with high performance.
Can a biological system made from muscle, bone, and skin
legitimately guide the design of a man made system? There are a number
of examples of this in the robotics community where a careful
examination of the construction, dynamic behavior, and control of
various biological systems led to the succesfull implementation of a
five fingered anthropomorphic hand, a walking 6 legged insect, and a
dynamic hopping kangaroo. Each of these implementations did not seek
to exactly copy nature, but to distill what was fundamental to the
behavior of the system and then implement that as best as was possible
with available materials and technology. In effect the fundamental
physics were preserved between biological model and man-made system.
In this light we have begun a detailed study of the design,
kinematics, fluid dynamics and construction of large, fast, pelagic,
predatory fish. The information gathered will be used to guide the
design of an efficient, highly maneuverable, high speed, undersea
vehicle in the future.
Very early on it became apparent that there was an extrodinary variety
of creatures to look at. There are more living species of fish than
any ofther vertebrate (30,000 is a conservative number) and at least
75 species of living cetacea (dolphins, whales,etc.). The variation in
body types and propulsion techniques among this vast collection is
staggering. In order to proceed at a reasonble rate we decided to
narrow the scope of the study to a single family of fish. Namely one
that was capable of high speed, efficient propulsion, and one that had
physical attributes that would be conductive to replication in a
man-made system. In short, a fast submarine shaped fish with a
relatively rigid torso that swims with fairly small body and tail
motions.
There is a family that meets these specifications quite well: the
Scombridae. This family includes a wide range of body sizes sharing a
reasonably common hull form ranging from the Skipjack Tuna, 36 inches
long and weighing 77 lbs., up to the grand Bluefin Tuna, 14 feet long
and up to 1500 lbs. The fact taht even across a wide range of body
sizes the basic tuna-form shape remains very similar leads to the
observation that this is somehow an optimal hydrodynamic design and we
could perhaps benifit greatly by using it as the fundamental model for
our vehicle.
Design Details
Incorporating all of the sophisticated engineering
suggested by a live biological tuna into a man-made robotic vehicle
capable of autonomous free swimming and maneuvering in an ocean
environment is a dounting task. Rather than attempt to attain this
goal in one giant leap, it has been persued in a number of shorter
pragmatically achievable steps.
The first such step is to design, build and experimentally test a
full sized, carefully instrumented robotic tuna in the M.I.T. Testing
Tank. This is suspended from a single surface piercing streamlined
mast from the Testing Tank's carriage in a fashion commonly used to
test the hydrodynamics of undersea vehicle models. As such its gross
body motion will be constrained by the Testing Tank's carriage to be
in a straight line over the length of the tank at the midway depth
between the tank's bottom and the free surface. The tuna's body
however is able to rotate freely about its center of mass in a plane
parallel to the water's free surface.
The experimental apparatus that makes up the testing tank consists of
2 major mechanical subasseblies; the robot fish body itself and the
support structure which attaches it to the testing tank's carriage.
The robot's body design closely parallels that of the biological
tuna. As such it must fit into the carefully prescribed volume of the
biological Tuna-form shape. (below)
The major structural component of the robot fish is a segmented
backbone made up from 8 discrete rigid vertebra connected with low
friction ball bearing joints. As in the biological fish this is the
framework on which all the other subassemblies are mounted and the
orientation of these vertebra to each other defines the overall
curvature of the fish's body.
These 8 vertebra are driven through an elaborate system of pulleys
and cable tendons by 6 brushless DC servo motors mounted outside the
fish above the waterline inside the carriage support structure. These
tendon drives are the mechanical analog of the biological fish's
muscles.
In between each discrete vertebra is strung a flexible spring steel
spline element on which are mounted lightweight body ribs that define
the overall tuna form shape of the outer hull. As the discrete
vertebra move with respect to one another these spline-rib assemblies
bend to provide smooth continuous structural support for the flexible
outer hull much as the real fish's ribs do.
The robot fish's outer hull consists of a thin layer of flexible
retuculated foam covered by a conformal Lycra sock. The foam acts as
the analog of the biological fish's fless, the Lycra sock as it's
skin, to provide a smooth conticuous hpefully wrinkle free flexible
outer hull.
The body itself is suspended from the support structure by a single
surface piercing streamlined mast which both supports the body and
acts as a conduit through which all the tendons and sensor wires
pass. This mast in turn is mounted to the support structure in a
fashion which will allow it to rotate up out of the water for robot
maintenance and storage.
The support structure consists of a rigid welded aluminum space
frame which attaches to the to the existing testing tank carriage. In
addition to supporting the mast and fish it also houses the servo
motors and various bits of sensor electronics.
Connecting all these subsystems and subassemblies todether are two
experimental apparatuss controllers (one for the fish and one for the
testing tank's carriage), and two data collection systems (one for the
coarse real time data viewing and one for the precise high speed data
collection). These for systems work together in parallel to both
produce fish like motion by the tuna and to record the output from its
sensorsystems while it swims down the tank. The end result is a
vehicle that both looks and moves like a real biological tuna, but
with which we can do careful experimentation.
| 
