|
In 1994 RoboTuna I was built, at MIT, by David Barrett. Robot Pike
is a "relative" of this project. Before beginning the formal design of
Robot Pike, a species of fish to mimic had to be chosen. The pike was
chosen because of its excellent accelerating and turning abilities.
Once choosing to build a pike, it was possible to begin work on the
actual design of the robot. As an introductory exploration into the
workings of underwater vehicles, a model submarine was built. Despite
being a simple hobby shop model, this submarine gave valuable insight
into the difficulties of powering an underwater vehicle. The next
step of this project was to obtain a fiberglass model of a Chain
Pickerel, the specific type of pike that had been chosen. Using this
model, cross-sections of the fish were able to be drawn.
The next step was to build a kinematic model of the pike, and thus
study the power and motion of a body driven by three servos.
At last, it was time to work on the final design of the robot.
There were many critical design decisions that had to be made. In a
robot which must be strong enough to swim and turn, flexible enough to
mimic a fish's movements, and light enough to nearly float, material
selection is very important. Much thought was put into the choosing of
each individual part of this robot. Acetal plastic was chosen for most
of the structural elements, as it has low water absorption, low
specific gravity, high tensile strength, and high resistancy to
corrosion. The final design was arrived at by an iterative process of
conceptualizing the design, drawing a layout design, and then
detailing areas of the design. During the actual construction of the
robot the design drawing was updated as problems were encountered.
Design Decisions:
Control of the Pike
- The robot is to be controlled by a supervisory controller. The
navigation will be performed by a human, and a computer will interpret
the controls so that the robot can perform as expected. The computer
will, after interpreting the controls of the person, send the
appropriate signals to a radio transmitter. Inside the robot is a
radio reciever, which then sends the control signals to each
servo.
Three Bending Segments
- By using only three segments, the size of the fish could be kept
down, while still using inexpensive actuators. This means that while
the shape of the fish cannot be made to resemble exactly the shape of
a real fish, a close shape can be obtained.
Courtesy of Dr. Jamie Anderson
Flexibility
The desired flexibility is very high, with external skin strain
in the range of 50-100%. Components, which in an inflexible body
would be easy to package, are in danger of colliding with other
components. Components must be mounted carefully to allow maximum
flexibility.
The rigid components inside the robot needed to be able to move
such that they:
- would not collide with each other
- would not push up against the flexible hull of the robot
- would not suffer from the repeated motions of swimming
In order that the components would not fall prey to the above
problems, we decided to implement the following design aspects:
- Large bulky components were put where the fish does not bend very
much; in the forward part of the body.
- The rigid components were kept away from the skin.
- All bearings and hinges were designed for a million or more
cycles.
Flooded Hull
Making a waterproof hull has advantages and
disadvantages. The reasons for waterproofing the hull are:
- The fish can be filled with air to provide a large amount of
buoyancy, allowing the components to become heavier and stronger, and
possibly cheaper.
- Filling the robot with a nonconducting fluid, such as air or
silicon oil, would allow the electrical components to remain unsealed.
This would save space.
The reasons for not waterproofing the hull are:
- Modern seals do not reseal themselves in a non-clean environment.
Thus, any opening of the robot would require conspicuous cleaning
before the resealing of the hull.
- Any leak would destroy the electrical connections which are not
sealed. Sealing the components against the water, however, removes
one of the reasons for making a sealed hull.
- A sealed flexible skin would be much more difficult to fabricate
than a nonsealed flexible skin.
Because a flexible waterproof hull would be more difficult to open
and then reseal, and as it would also be more difficult to
manufacture, it was decided to use a flooded hull. This made it so
that many components would have to become waterproof.
Waterproof Components
-
Each component not capable of operating in chlorinated water is
waterproofed individually. While this takes more space, and a little
more weight than the components do individually, it makes the system
more durable. A single leak will not destroy all the
components. However, because there are more seals, it is more likely
that there will be a single leak. Because the system is meant to
accomodate an array of sensors and actuators, some of which are very
expensive, we opted to seal each individual component with a reliable
seal. This will make waterproof connectors desirable in the design.
 Spiral Spring Exoskeleton- The spiral spring
that forms the fish was made from Cast fiberglass, using West system
epoxy. This allowed us to cast the exact shape of the spiral spring,
and keep the component very light.
The spline that connects to the spiral spring exoskeleton is 1/16 inch
delrin. Delrin, a resilient plastic that was tested with "Charlie"
(Robot Tuna), was chosen due to the necessity that the fish be highly
flexible.
Missing Ventral Fins-
By watching the motions of fish, it was determined that the
elimination of the ventral fins would not signifigantly reduce the
performance of a robot fish whose desired activities are turning and
forward swimming. Fish primarily use these fins for hovering
activities. With the ventral and pectoral fins they can move
backwards, sideways and slowly forward. We will lose the ability to
mimic these motions. But we are designing a robot fish, which can
maneuver and turn, not hover. The elimination of the ventral fins
allows the robot to stay smaller. Another reason for not attempting to
implement the ventral fins is that it is unclear how many degrees of
freedom would be needed to control them the same way a fish does. By
comparing a submarine with a fish, one can see that the pectoral fins
are like the hydrofoils on a submarine, and control the dive rate of
the fish.
| |
