SUBJUGATOR GENERATION 3
Version 3.4
|
The 4th major revision of SubjuGator occurred after the 3rd AUVSI competition. This
generation of SubjuGator has lasted 4 seasons as we continue into the 6th AUVSI UUV competition.
Version 3.4 of SubjuGator has been customized to perform in the 2003 competition.
This year we are working to develop 3 major systems on the SubjuGator:
a computer vision system to determine the active target at the decision point
as well as target recognition, a hydrophone system to track down the active target
via its pinger, and an electromagnet triggered marker dropping mechanism.
Computer Vision System:
To accomplish underwater computer vision, we have developed and constructed
both hardware and software capable of capturing images and processing them
completely on-board the submarine. We are using a Unibrain Fire-i400
progressive-scan camera, capable of 640x480 resolution at 15fps. This camera
has an interchangeable lens and interfaces to our embedded computer through
IEEE1394 (Firewire). The camera is mounted in an exterior enclosure.
The latest Linux kernels have built in support for plug-and-play Firewire
devices. Using the digital camera libraries available for Linux, we have
written custom software for both frame grabbing and acquiring video.
Using our camera and computer vision algorithms, SubjuGator is able to
accomplish the detection, localization, and classification of the underwater
targets and the decision point used for this competition.
At the beginning of each run the submarine takes a sample data of its environment.
This data is used as the basis for its image processing calculations. Any pixels
that significantly deviate from normal brightness are assumed to belong to an LED
on the arrow pointing to the target for the submarine's current run. Once the arrow
is used to determine the direction of the target, the submarine pursues the target
using its hydrophone system.
Hydrophone System:
The hydrophone system consists of four basic stages that aid in obtaining and
processing the signals transmitted by a target's pingers. These stages are:
1. Hydrophones
The hydrophones provide a means with which to detect acoustic vibrations in
the water, such as the signals transmitted by the pingers corresponding to each
target. SubjuGator utilizes three hydrophones mounted in a triangle configuration
to create the geometry needed to identify the direction from which the received
signals originated.
2. Amplifier
Because of the weak signal strength produced by the hydrophones, the hydrophone
outputs are fed directly into an amplifier circuit where the signal is amplified
by a gain of approximately 100. The amplifier output is then passed into a Schmitt
trigger to create the square wave necessary to trigger logic levels in the ESOI.
The resulting three square waves are then passed on to the ESOI for further processing.
3. Embedded Signal Occurrence Identifier (ESOI)
The ESOI system, consisting of a MAX7128S CPLD, takes the three digitized signals and
judges whether or not they are the signals of interest. If they are, the system sends
time of occurrence information to the ATMEGA8 microcontroller, otherwise it waits for
the correct frequency to occur.
4. ATMEGA8 microcontroller
The ATMEGA8 interface controller provides an RS232 interrupt driven state machine,
which controls the ESOI logic as well as the electromagnetic relay circuitry. The
microcontroller is in constant communication with SubjuGator's
main computer system, which tells the microcontroller which pinger frequency it
needs to pursue to find its target as well as when to drop the markers in order to
hit that target.
Marker Dropping Mechanism and Electromagnet:
The dropping mechanism was designed to safely carry and deliver 1.5" steel marker balls
to the active target and release them when the target is detected. The dropping mechanism
is mounted externally on the ventral side near the frontal cone of the submarine.
It is actuated through SubjuGator's aluminum hull wall by an electromagnet that attracts
a rectangular piece of steel on the arm of the mechanism. Throughout the mission,
the mechanism carries the markers within a holding tube made of PVC pipe. When the
target is detected, the electromagnet is activated, attracting the steel on the mechanism
and pulling a pin that allows a trap door to open and the markers to fall onto the target.
Chassis, Power System, Computing, and Basic Sensors:
The remainder of the SubjuGator's chassis, power system, computing, and basic sensors
is much the same as SubjuGator Version 3.3.
History of Generation 3
SubjuGator Version 3.3
(2002 AUVSI Competition)
Following the 2nd AUVSI competition, we again had an opportunity to
reflect on what we learned and how we could improve our vehicle.
It was felt that the current body was pushed to its limits and
we wanted a more robust and reconfigurable body style. We also
wanted enough power (actual and computing) to handle the current and
future goals. We classified the new structure into four categories:
Power, Mechanical, Sensing, Computing.
Power:
Solving the motor driver problem in the last design, we concentrated
on selecting batteries which would provide enough power to run the
sub for a length of time without having to change either the power or
the computer batteries.
Mechanical:
There were two goals that we wanted to address with the new body. The
first one was internalizing the batteries. An external battery wasn't
going to cut it anymore. The second was to make a sub that was reconfigurable.
Dave was tired of having to try to bolt something new on for each competition.
Clearly, a sub that had external mounting points for the current and future
sensors was needed.
Sensing:
As a minimum, for an autonomous sub, you need a depth sensor (in
our case, it was a pressure sensor) and heading sensor (compass with roll and
pitch). We also thought it would be nice to add an altimeter to sense the height off the
bottom. For this reason we added a Datasonics PSA-916 sonar altimeter. Our altimeter model
has been modified to measure distances from 30cm to 100m with a resolution of 1cm over an
RS-232 connection.
For the competition we also had three mission specfic sensors. These
included a hydrophone array to sense the acoustic pinger, a CdS array
to sense the strobe, and a camera to locate and align the sub with
the ring.
Computing
We are using a Pentium-3 700MHz enbedded processor system. The is the workhorse for our vision processing system.
A Motorola 68HC11 processor is also used to provide A-to-D and control signals for our motors.
|
SubjuGator Version 3.2 and Earlier
(2001 - 2000 AUVSI Competitions)
|
Power
|
The batteries were Power Sonic Sealed Lead Acid batteries. They
are 12 volts and provide 12 Amp Hours of power. Four were chained
together in parallel to provide power for the motors and a single
one was used for the processors and sensors. After 3 hours of
pool time, running the motors almost constantly at 30%, the batteries
droped from 13.2 V to 12.7. Apparently we didn't need to worry about
running low on power.
|
Mechanical
|
We pulled off everything that we could use from the old body to put on
the new one. This included the on/off and kill switch. Also, it was
our initial intention to pass all the external connections through
two 20 pin Burton electrical connectors.
However, we wanted an entirenly new design. It was felt that the current
generation was taken to its limits. For the second generation, a box
was made to fit all the new computational hardware. If we wanted to add
anything else, a new box would need to be made since that one was custom
made for that hardware. Also, the power battery for the motors was exposed
to the elements. We wanted to eleminate all that, while making it robust
and reconfiguarble.
The first step was coming up with the design. We wanted something
more streamlined, but also easy to fabricate and that would mate well
with our connectors.
|
For a vehicle this size, safety is something that cannot be ignored. A
kill switch is required. When we are in communication with the sub, we
can stop it via software, but we cannot assume that will always be an
option. Therefore, we required a physical kill switch mounted externally
that, when activated, would interrupt ALL power for the motors. We wished
to find a way to do this without also shutting down the computer, since
shutting down a unix machine without warning is not a good thing. Since
we isolate our computer / motor power systems, this turned out to be not
that difficult. We simply ran the power for the motors through a Single
Pole Double Through (SPDT) relay, and used the waterproof switch to
control the throw on the relay. When the kill switch is pushed, the relay
throws and the motors get no power regardless of the computer commands.
This switch is a medium power (about 6 amp) switch from Giannini. We use
a similar switch to power our electronics.
|
Water leaks was a fact of life for the Gen One sub. Since we were
completely upgrading all of the electronics, a water leak would be worse
than ever. The estimated value of our electronics went from $800 (Gen One)
to about $3800 (Gen Two). That box needed to stay bone dry. In the past,
the connectors were the real problems, so we looked far and wide for
connectors that would solve our problems. Burton Electrical Engineering
was our answer. Their connectors, in our eyes, are the best built objects
on the sub. These things are simply amazing, and the connectors are now
the last thing we worry about.
|
Sensing
|
|
As with the first generation vehicle, a Precision Navigation TCM2 digital
compass was used. It is based upon a proprietary triaxial magnetometer
system and a biaxial electrolytic inclinometer, contains no moving parts,
and can output compass heading, pitch, and roll readings via electronic
interface to a host system.
|
The previous pressure sensor returned a voltage from 0-100
millivolts to represent a depth from zero to 60 feet! The
resulting resolution basically said "According to my voltage, we are
somewhere between the surface, and 1.3 feet deep." We needed better
resolution. We upgraded to a Measurement Specialties Inc MSP-300 depth
sensor rated to 100 psi. It outputs an analog DC voltage between 1 and
5 volts, with each 10 feet of water depth changing, the sensor voltage
by 0.225 volts. For better resolution, we built an amplifier which
produces a 5 volt swing for a depth change of 20 feet. That allows us to
resolve to about 2 inches of depth.
|
Computing
|
|---|
Top-level control is handled by a Win-Systems LBC-586Plus single-board computer with
32MB RAM, running Red Hat Linux. All sensor information, gathered on one system, is
evaluated, and consequent instructions are then is-sued to all subsystems.
|
The Motorola DSP56309 is an 80MHz 24-bit fully pipelined DSP.
Of the many features of this system, the ones we are exploiting are (1)
a serial communications interface, (2) system interrupt timer pins, and
(3) a data acquisition time resolution of 27ns. The system interrupt timer
pins extract phase information from the acoustic localization system to
determine the bearing to the beacon. The SCI system receives instructions
from the main processor, and transmits phase information to the main processor.
|
 Running Linux gave us a multi-user OS that fit perfectly in a network
environment. We just had to find a way to get it on a network. In the
lab, we could plug in a PCMCIA network card, allowing multiple users to
work on the sub simultaneously. In the past, a physical tether was
required to communicate with the sub. Poolside, this became quite
inconvienent, and quite slow. We desired a wireless ethernet style
connection. In steps Harris Semiconductors. We recieved an evaluation kit
for the PRISM wireless ethernet cards. These cards are intended for
design engineers testing the technology for possible inclusion in a
product of some sort. What that means is that we had the flexibility of
adding our own antenna. We placed the card in the sub, and ran the wire
through the hull to an external antenna. During all development work, we
placed the antenna on a styrofoam cylinder that could fload free (the
wire is 8 feet) behind the sub, and could therefore relay sensor data in
real time. For competition runs, the antenna is mounted on the hull, and
communication is lost during submergence.
|
|
