Paul J Heckman.

Free-swimming submersible testbed (EAVE WEST) online

. (page 3 of 4)
Online LibraryPaul J HeckmanFree-swimming submersible testbed (EAVE WEST) → online text (page 3 of 4)
Font size
QR-code for this ebook

Table 4. Summary of NOSC/USGS Free-Swimming

Vehicle's Performance Characteristics.


A brief functional breakdown describing some details of existing and
planned capabilities will now be given. All subsystems are part of the tech-
nologies being investigated for potential use with unmanned, untethered sub-
mersibles (figure 2). Other aspects of these technologies, such as acoustic
communication and bottom transponder navigation, are being investigated by the
University of New Hampshire and will be tested or demonstrated on their EAVE
EAST submersible (see reference 7).

Communication Subsystem

For entirely autonomous operation before launch, the communication link
to the EAVE WEST vehicle is simply an umbilical cable which is disconnected
after preprogramming the submersible and transferring the resulting data base
to the vehicle. The umbilical cable is disconnected while the vehicle is
executing a preprogrammed time delay.

The use of fiber optics as a communication link to this submersible is
being investigated. This link is an attempt to fulfill the real-time console
communication requirements through an extremely small, lightweight, low-drag,
single fiber-optic strand. This fishing-line-size cable will be able to
handle all data and control requirements of the entire vehicle at the full
bandwidth of existing visual data sensors. If this is achieved, it will be


possible to take advantage of a real-time high bandwidth data link; use exist-
ing, relatively inexpensive data sensors, such as television cameras and sonar
systems; and not suffer the disadvantages of high cable drag experienced in
tethered submersibles. For the application of a pipeline survey and inspec-
tion vehicle, where the submersible will be deployed from a compact canister
along relatively straight-line runs for long distances, this approach is quite
feasible and attractive. Up to 5 km of cable can be stored on a coil which is
approximately 3 in long with an outside diameter of 8 in and an inside dia-
meter of 5 in. For the application of structures inspection, this approach is
promising but not without fault. Cable entanglement within the structure can
break a fiber-optic link with a 6-1 b tensile strength; however, with some
autonomous capability the free-swimming submersible could conceivably navigate
out of the structure. Use of a fiberglass-sheathed, fiber-optic cable with a
100-1 b tensile strength is also being investigated for potentially reducing
the risk of breakage due to entanglement.

Components of the fiber-optic link are shown in figure 14. Subtechnolo-
gies to this investigation are listed below:

1. A method involving a high signal-to-noise ratio (SNR) for modulating
large bandwidth signals for transmission of the uplink. NOSC's approach has
been to use pulse frequency modulation (PFM) (see reference 9).

2. A duplex operation for sending relatively low bandwidth command sig-
nals to the vehicle and for receiving relatively high bandwidth visual sensor
information at the operator console.

3. Penetration of the fiber-optic link into the underwater containers;
the penetration method must be easily disconnected and operate at a depth of
2200 ft or greater.

4. Deployment of the fiber-optic cable from the submersible without ex-
ceeding the maximum tension of the cable and without causing vehicle drag.

5. Precision winding of the cable in the laboratory for installation in
the deployment canister. The resulting coil must incorporate pretwisted wind-
ing to allow deployment without creating kinks.

6. A way to avoid cable entanglement in the vehicle propellers.

The results of the investigation of item 1 are discussed in reference 1. The
investigation of the remaining items and inspection of pipelines and struc-
tures are discussed in reference 10.

Approximately 80 percent of the work required for complete demonstration
of a deployed, fiber-optic link on the EAVE WEST vehicle has been completed.
This includes the PFM techniques, duplex operation, penetration to the vehicle
housing, and a precision fiber-optic winding machine. Preliminary tests of
possible deployment methods have been made, and the vehicle has been operated



in water in a real-time control mode with a good video picture from an under-
water television camera mounted to the vehicle with a small, twisted-pair
tether. Thus, since the fiber-optic subsystems have been fabricated and
individually tested, all that remains is to install and test the various
components on the submersible.

Navigation Subsystem

Three basic navigation methods have been designed for the EAVE WEST sub-
mersible: preprogrammed trajectory and magnetic pipe following when operating
under the autonomous mode and visual orientation when operating under the pro-
jection mode.

The vehicle's preprogrammed trajectories are presently executed through a
dead-reckoning navigation sensor, i.e., a magnetic compass, a depth sensor,
and an internal timer are used to navigate from one point to another. Such an
approach, of course, leads to drift errors because of ocean currents. If,
however, an off-the-shelf bottom transponder navigation system were added to
the vehicle, software programming of the control equations could be accom-
plished in the same general manner used for the existing dead-reckoning ap-
proach. Error signals would be generated to compensate for the difference
between a desired position and the actual position measured by the navigation
system. The cost of such a system could, however, approach the present devel-
opment cost of the vehicle. Because of this and also to avoid redeveloping
technologies already existing, a simple dead-reckoning approach was installed
for demonstration purposes.

A magnetic-pipe-following system will soon be incorporated in the vehicle
to allow autonomous tracking of a pipeline located either on the ocean bottom
or buried beneath it. The system uses magnetic induction to autonomously
follow a 48-in pipeline at a vertical distance of up to 18 ft. The magnetic
sensor and transmitter are separated from each other and the vehicle to limit
interference (see figure 3). A block diagram of this approach is shown in
figure 15. The system uses two receivers and one transmitter operating at
frequencies selectable from 40 Hz to 4 kHz. The basic problem involved is to
discriminate the received signals from the transmitting signal with as high a
signal-to-noise ratio as is possible. Details of this sensor system will be
discussed in a separate report.

A television camera has been installed on the vehicle to provide a visual
means of navigation along a pipeline or within a structure when the vehicle is
operated in projection mode. Transmission of the signal to the console opera-
tor will be via the fiber-optic link when it is installed and operational. In
addition, however, NOSC is pursuing a means of transmitting the video acous-
tically to the surface. Display of 128-by-128 and 256-by-256 picture elements
has been tested and determined to operate to depths of 4000 ft (reference
11). This is another technology subelement being investigated as a part of
the free-swimmer technology development program.

Data Acquisition Subsystem

A Subsea Systems, Inc., silicone diode array camera with a 600-line
resolution capability and a 75-W quartz iodide light is presently installed as
one of the major sensors for visual inspection of pipelines and structures. A



Z ± -I

e> q o

£ i z
o o
oc °



< .


_l z

< to
^ u






. o

J w

_ o

co O

cc o -J

5 O


lil (J










OC w
uj to

-1 _l

0. UJ


< z

<o <






C3 1- O


< z or



O c

o 2! x?

CO - ^
Q Q. 8

2 5 o


UJ (-


"■ CO (J

w uj _,





< < *<

Z 2 O

* < X





fc " s


CO ,5










1 J





si z £

!2 < K
o oc - 1



< oc a-
oc lu £
ui a _
Z O i







o o: -1

Z UJ u.

5 u. co





Z uj £

3 u- co

O jtj
oo 5 Q



ui t j
d z 5

° j S?

02 Q

2 < z

< < <

7 < z

2 m <




2 m <




Remote Ocean Systems Super-8 film camera is used for gathering both still and
motion picture photographs on keyboard or preprogrammed command. The future
importance of the data acquisition subsystem is not what or how many sensors
are incorporated, but whether these can be used to provide better autonomous
control of the submersible and whether information concerning structural
faults or leaks can be automatically derived to alert an operator as to their
location. Such technologies will be investigated in the future as a part of
this program.

Physical Tasks Subsystem

With the intent of using EAVE WEST as a testbed and with financial sup-
port from both the Office of Naval Research and NOSC's Independent Exploratory
Development funding, a manipulator is being developed. The electrically driv-
en manipulator (figure 16) has 5 deg of freedom in addition to jaw closure.
Degrees of freedom from top to bottom include shoulder pivot and rotate, elbow
joint, wrist rotate, and wrist pivot. Its six pressure-compensated, oil-
filled motors incorporate position feedback potentiometers and harmonic drive
gearing. Total lift capability is 25 lb vertically and approximately 8 lb at
full extension. The total weight of the manipulator is 34 lb in air. Motor
housings are fabricated of hard black-anodized aluminum, whereas the other
support components are composed of fiberglass with the arm itself filled with
syntactic foam to conserve weight in water, which is about 30 lb. The arm it-
self has already been fabricated and tested.

Plans are to drive the arm in a supervisory-controlled fashion using LSI
11/23, 16-bit minicomputers packaged both in a vehicle housing and at the
surface console area. This will reduce the vast amount of bandwidth and
operator training otherwise required when using a manipulator on a remotely
controlled vehicle. Computer software is being supplied by MIT as an adapta-
tion of MIT's effort in supervisory-controlled manipulators and man/machine
interface for ONR (reference 12). A diagram showing the supervisory-
controlled manipulator as designed for integration with the LSI 11/23 printed
circuit cards and control components is in figure 17. The device will be
installed on the vehicle, as indicated in figure 3C. Although the manipulator
is shown mounted at the front of the vehicle to allow reaching into crowded
areas, system modularity will also allow it to be mounted at the center of the
vehicle. This would especially be useful for picking up instrument packages
or seafloor samples when operating along a pipeline or in an open area.

Technological problems to be solved include the following:

1. The use of a measurement arm to facilitate operation when relative
motion is present between the vehicle and the work site.

2. The use of electrically induced compliance to facilitate final ap-
proach and grasp operations.

3. The use of low bandwidth measurements of the position of the manipu-
lator's endpoint to update slow scan television pictures as to the
continual status of the manipulator position.


Figure 16. Electrically driven manipulator.


Figure 17. Supervisory-controlled manipulator.


Potential tasks involved for the manipulator relative to the USGS pipe-
line and structure inspection are as follows:

1. Positioning a cavitation, erosion-cleaning nozzle developed under
contract to the USGS R&D program.

2. Placing a self-contained sensor package at selected positions on a
structure; these packages may be attached either mechanically or

3. Positioning imaging subsystems to allow visual inspection of welded
joints and structural members; candidates include still and motion
picture photography and conventional and solid-state television.

4. Attaching a line to instrument packages or navigation transponders
through a hook arrangement to recover such packages from underwater
structures or in the vicinity of pipelines on the ocean bottom.

Details of this work are in reference 13.



Several in-water tests of the EAVE WEST testbed submersible were con-
ducted during its development. Most of these tests, designed to validate the
proper operation of various subsystem components and software control rou-
tines, were performed at NOSC's Transducer Calibration Facility (TRANSDEC)
test pool, where the water was clear enough to allow easy visual and photo-
graphic observation of the vehicle's performance. Pool tests were also a pro-
tection against loss of the vehicle due to unexpected behavior, system mal-
functions, or container flooding during its early stages of development.
Tests were later performed in San Diego Bay off the NOSC pier. Although the
water clarity and resulting visibility were poor, an idea of the vehicle's
performance under more stringent conditions and in saltwater was obtained.
Tests will eventually be performed at the NOSC San Clemente Island Facility,
where performance of simulated inspection missions can be observed in rela-
tively clear water (the optical volume attenuation coefficient is
approximately 0.2/m).


Initial tests of the vehicle were performed at TRANSDEC in October 1978,
while the vehicle was programmed to operate directly from a "dumb" ADM-3 ter-
minal and was launched entirely in a preprogrammed trajectory mode. All con-
sole man/machine interaction programs and the vehicle's routines were stored
in firmware within the vehicle's existing 8-bit 12 kilobytes of PROM memory.
Only 2 kilobytes of RAM were required within the vehicle's memory to perform
the scratchpad calculations to execute these programs. Most routines were
originally written in Intel's 8080 microprocessor assembly language. There
were no visual or data sensors on the vehicle at the time, and there was no
real-time data or control communication link to the console subsequent to the
vehicle's launch. The vehicle was simply preprogrammed to transit to the
center of the pool area, execute a series of maneuvers, and return to the side
of the pool. One of the significant features, however, was that it success-
fully performed these exercises using only a single small microprocessor (a
total hardware cost of only $3000). An underwater photograph taken of the ve-
hicle during this first operation is shown in figure 18. The results of these
tests and a description of the status of the EAVE WEST submersible at the time
are described in reference 14.


During the first single-computer configuration test at TRANSDEC, the fol-
lowing results were achieved:

1. The vehicle was trim-ballasted according to theoretical calculations
based upon estimated system component weight. The submersible was
found to be 40 lb positively buoyant rather than the 86 lb previously
calculated, which represented an error of less than 10 percent of the
total vehicle weight.



2. The vehicle successfully executed various preprogrammed trajectories
as modified by the operator. Each trajectory was found to be repeat-
able. The vehicle could even be made to return to the side of the
tank and to stop when finished with a given run.

3. Data were obtained for the motor thrust and vehicle speed with re-
spect to various programmed values.

4. Still and motion picture photographic coverage was taken of both the
surface and subsurface runs.

5. The emergency and abort routines operated well.

6. The system was found to be relatively easy to transport, and it with-
stood moving with little difficulty.

Data Measurements

Forward thrust was measured to be an average of 38.5 lb when the vehicle
was programmed for full speed.

Forward speed was measured by traversing a 60-ft timed run after the ve-
hicle was up to speed. At full speed, the vehicle ran this distance in 19.5
s, indicating a maximum vehicle speed of 1.85 knots or 2.1 mi/hr. The speed
was found to be independent of whether the vehicle was on the surface or sub-
merged. These values are in excess of the maximum 1.5-knot speed predicted
during the design phase.

Problems Encountered

Problems encountered with the system were minor. They are mentioned pri-
marily to aid future designers of robot submersibles.

STATIC BALANCE. Because it was necessary to place all the batteries in one
container to separate them from potential spark-producing electronics, the
vehicle tended to list to one side. The problem was easily corrected during
the trim and ballasting procedure by merely adding more weight to the other
side of the vehicle to counteract the battery weight. Additional foam (figure
18) was also added to compensate for the resulting loss of buoyancy. These
additional blocks of foam, however, probably increased the vehicle's drag in
the water, resulting in a lower vehicle speed. In the future, as new elec-
tronic payloads are added to the submersible and more foam is added along the
top corners of the frame, this problem is expected to disappear.

OPERATOR CONSOLE INTERACTION. Fast interaction with the terminal was a
problem during the tests; too much time was wasted when it was necessary to
reprogram the vehicle. Since that time, however, the trajectory design
software has been upgraded not only to allow a selection of preprogrammed
tracks, but also to allow development of new patterns based on keyboard input
by the operator. In addition, this software has since been modified to accept
a much faster operator interaction type of editing rather than the pure
prompting-of-questions approach used during the TRANSDEC tests. Reprogramming
now requires only 1 min rather than 5 min. Unfortunately, the disadvantage is
that the operator must be a little more familiar with the program.


REAL-TIME CONTROL. The lack of real-time control during the TRANSDEC tests
was not as serious a problem as expected, particularly since the surfaced
vehicle experienced a definite water current caused by a relatively strong
westerly wind. The current was rather constant, and it was observed that the
vehicle could actually be programmed to return repeatedly to the side of the
tank. However, in most circumstances, a lack of real-time control and some
type of postlaunch reprogramming capability could either prevent recovery of
the vehicle or cause the vehicle to be misdirected or lost.


During September 1979, another series of tests was performed with the
vehicle to determine if the supervisory-control hardware and software archi-
tecture were functioning as designed and to study the dynamics of the fiber-
optic link deployment. Again, the tests were performed in the TRANSDEC pool
to aid visual and photographic observation and to protect against loss of the
vehicle. Topside console programs used approximately 12 kilobytes of memory,
and the major portions of the programs were updated to PLM. A television
camera and light were mounted on the vehicle, but were not electrically con-
nected or used. The fl uxgate-updated gyro compass was used for this series of
tests, replacing a magnetometer compass used during the initial tests. Most
of the remaining electronic hardware stayed the same. A coffee can, nylon
funnel, and 1/2-in PVC pipe were used as the fiber-optic deployment canister,
which was temporarily strapped to the bottom of the vehicle's frame (see
figure 19). Although deployment tests were performed, no actual fiber-optic
communication link existed and there was no additional software to support
real-time control of the vehicle.

Accompl ishments

The two-processor, supervisory-controlled configuration operated without
difficulty. There were no more problems encountered using this more complex
structure than were observed during the initial tests with the single vehicle
processor architecture. The more versatile trajectory design program, added
to the console since the initial configuration, greatly enhanced the operator-
to-console interaction speed.

The basic approach for deploying the optical-fiber link was validated,
amd photographic coverage of the vehicle as it deployed the fiber-optic cable
was obtained. Several runs were successfully made deploying unclad fiber,
fiberglass-clad fiber, and dummy fiber (nylon line).

Problems Encountered

BATTERY LIFE. Although the use of lead-acid batteries is not a technological
problem, much time was wasted because of the poor condition of some of the
batteries. In general, the use of the parallel-series chain of the 2-V,
5-A-hr cells caused much of the problem. This eventually led to the adoption
of a single-series chain of the more powerful 2-V, 25-A-hr cells.

MAN/MACHINE INTERFACE. Although much of the new software added to the console
programs greatly enhanced the console-to-operator interaction speed, the man/
machine interface problem remained an important consideration. It is very
difficult in a short amount of time to communicate to a vehicle a-priori the


0& r f m h


exact trajectory designed. The use of analogic controls and displays, such as
graphics of the planned trajectory, analogic displays of panel meters, and
joystick controls, greatly simplifies the procedure of operator interaction to
the system. Symbolic displays, such as digital readings and digital depth in-
dicators, require much time in the field to study and to understand their im-
plications. The use of a computer keyboard is also somewhat limited for field
test operations. The experience gained from this particular test, therefore,
caused further investigation into the use of analogic controls and displays.

ENTANGLEMENT OF THE FIBER-OPTIC LINK. No problem was observed in deploying
the fiber-optic link along a straight-line or forward-directed run. Difficul-
ty was experienced on occasion, however, in deploying the more flimsy unclad
fiber or the substitute nylon fishing line when performing simulated docking
maneuvers. It was possible to entangle the link in the vehicle propellers
when backing into the deployed cable. Although the link was held away from
the thrusters by the deployment tube (or "stinger") hanging from the rear of
the vehicle, this was proven to not be sufficient. Either a mechanical guard
on the propellers or the more expensive, stiffer, fiberglass-clad fiber must
be used. Further tests will have to be run.


Tests on the ability of the vehicle to respond to real-time control com-
mands were made in March 1980 at TRANSDEC and in San Diego Bay. The experi-
ments were a preliminary step to the implementation of the projection mode of
operation when the final fiber-optic link is installed. The cursor keys on
the control console keyboard were used to control the direction and magnitude
of the vehicle operated under proportional control. Hitting a given key once
produced a 1/4 full thrust change in the control signals applied to the ve-
hicle motors. Progress of the vehicle was visually monitored through an un-
derwater television camera mounted on the vehicle when the vehicle was sub-
merged. The vehicle was operated in the supervisory configuration with a min-
imum of top console display operator interaction. The console computer merely
formatted and relayed the commands input from the keyboard to the vehicle.
Vehicle response to the operator was the primary concern. A photograph of the
entire operational set up as used during the TRANSDEC tests is shown in figure 6.

Accompl ishments

During the TRANSDEC portion of the tests, experiments were devoted to
operating the motor controls in a real-time mode using the television camera
to guide and position the vehicle. These preliminary tests illustrated the

1 3

Online LibraryPaul J HeckmanFree-swimming submersible testbed (EAVE WEST) → online text (page 3 of 4)