An AUV Based Experimental System

For The Underwater Technology Education

Keisuke Watanabe
Associate Professor
Department of naval architecture and ocean engineering
School of marine science and technology, TOKAI UNIVERSITY
3-20-1 Orido Shimizu-ku
Shizuoka-shi, Shizuoka, 424-8610, JAPAN
keisukegscc.u-tokai.ac.jp
Abstract - In this paper, I introduce a series of AUV related
basin experiment platforms now we are developing as a set of
underwater technology educational tool. Existing AUV platforms
are very expensive for general students or schools to purchase as __-_
an educational tool so the popularity of AUV technology among
the general people is not still so high in Japan. On the other hand,
land based robotic contest is very popular in Japan and its Al
educational effect seems good for the wide range of students. I gSafti r
think it is important for us to develop a set of handy and low cost 00*0i
tools oriented to the underwater technology education and AUV
related platforms which can use easily in a small basin will be one
of the best choices. In this paper I will introduce our concept of
the AUV based experimental systems as an underwater
technology educational tool. The system is composed of two types "Me ~ SUUM IM
of AUV test-beds, precise ultrasound ranging system, LBLF Conceptualsketchofdeapseaconstructon
coordinate detection system and 2FSK ultrasound communication
system. The hardware and software design is introduced in this
paper. The AUV test beds are designed considering a convenience
of basin experiment using a 7inches-diameter acrylic sphere or
17inches-diameter glass sphere with low-cost embedded system
with DOS/V PC104 based subsystems. We constructed a set of PIhh'W .

iMHz carrier frequency ultrasound ranging system and l00kHz

LBL coordinate detection system. A 2FSK ultrasound
communication system for basin experiment was tested. In our
case, the basin is 3m(W)c3m(D) 6m(L). So we designed both AUTV
testbeds as are enough for the handling and mobility in our basin
experiment. The production cost must be suppressed while it has
enough equipment for our purpose. The hull is transparent so we
can implement optical communication using sets of infrared LED
and phototransistor without using any penetrate connectors for
the data communication between the inside equipment and Fig.2 Conceptual sketch of distributed monitoring
outside equipment like an ultrasound module or a land PC. The
underwater connector is relatively expensive compared to other Production Vesesl
equipment so we used it for power supply only and reduced the
fabrication cost. On smaller one, the thruster is composed of a
4.5W coreless motor with planetary gear and a four-blades
propeller. We designed and fabricated the magnet coupling for
watertight seal of the thruster to reduce the production cost. The
smaller type?s total weight in air is about 3kgf. As the total size is
small (0.4mILI_0.4mIBI_0.2mIHI) we can handle it easily and Transportation
T
4 Robot Agent
many robots can swim in our basin. The hull is designed to bear
the pressure at least l0m water depth. We selected Lithium ion Manganese
battery for power supply and the total capacity is 450Wh. By Oft Nodules
power consumption calculation we may use it about 4 hours with
A"

one full charge of the batteries. The main PC is PC104 module

(Advantech PCM3350) of which MPU is NS-Geode 300MHz and
64MB RAM. I implemented the RT-Linux based program system
in 1GB compact flash card from which the system boots. It
controls the total system through PIC based subsystems. The RS-
232 serial communication is used for gaining the ranged distance Fig.3 Conceptual sketch of collection of deepwater resources
from the ultrasound controller of which MPU is PIC16F876. This
RS-232 signal is also transmitted from the inside MPU to the

1-4244-01 38-0/06/$20.00 ©2006 IEEE 1

 outside PC by infrared data communication. So we can
monitor the information inside the pressure hull without opening
it. The program is implemented through Ethernet. The frequency
of the sensing and actuation is at most lHz this time, the timing is
controlled by the RT-task mentioned later. An electric compass
TCM2 is used to detect the heading angle and roll, pitch. I was
afraid that the magnet compass is affected by the thruster or
inner electric current, but at present, it doesn't matter. Each
device like electric compass has its own PIC and all serial
communication line is multi-dropped using 12C through the
communication interface PIC. The 6 underwater connector are
attached for ultrasound transducers, thrusters, power supply and
ethernet. The magnet switch is used for power on and off of 2 DC-
DC, which corresponds to 12V for thruster and 5V for computers.
The system generates maximum 4 PWM signals for the propeller
speed control with motor driver ICs. As we mounted a wireless
CCD camera inside, we can see the underwater image as far as
the robot is near water surface. We have just finished the basic
function test in the basin by free swimming. We have not
implemented modern controller, we will try path tracking
experiment in the next step..
I. INTRODUCTION
The underwater multi-agent technology has great potential
for the various activities in the near future ocean development
or investigation c,2,3,4]. Fig.o 1 illustrates an imaginary picture
of deep ocean construction as one example of those activities,
Because we can utilize buoyancy force actively to reduce the
self-weight of the structural members in the water and the
external force is relatively small compared with near free
surface in the deep water, there is a new possibility of
underwater structure to be developed if only we can develop
the method of automated underwater construction using many
autonomous underwater robots[5,6].
Fig.2 shows the schematic view of underwater monitoring
system now we are planning to develop as another possibility
of underwater multi-agent. We are aiming to realize the
continuous monitoring system in time and space where many
compact AUV groups distributed in target coastal sea area
from the surface to the bottom. The target monitoring area is
sliced planar hierarchically toward the depth as monitoring
layers, and a group of AUV arranged on each layer tracks the
given trajectory and gathers various environmental information
continuously communicating each other in the layer or between
other layers. There are also autonomous moving buoys on the
surface that have the role not only of the observation around
surface environment but also of the communication interfaces
between underwater groups and facilities on land, satellites etc.
The target area also marked off one lot from another and each
division has an AUV group communicating other groups to
organize information networks from land base through free
surface to the bottom of the sea. We call this system as
PADOMS abbreviating Portable and Autonomous Distributed
Ocean Monitoring System[7]. It will improve the defects of the
present ocean monitoring method, that is, the monitoring
depends on the spotted water sampling so that continuous in-
situ monitoring is difficult.
Fig.3 shows a concept of development of deepsea mineral
resources like manganese nodules by underwater multi agent
robots. Each distributed robot has manipulators to grab the
nodules and they gather the nodules into atransportation
LAND PC
...~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~............
Ul-ut Sound
A
I B
Fig.4 Conceptual view of basin experiment setup
basket. If the transportation basket is full then the basket
releases its weight to go up to the surface where the cellection
vessel is waiting for baskets. As the underwater robots are
staying for long period around the seabed, the power supply
station is needed. The suggested method before now for the
development of deepsea manganese nodules consists of a very
large vessel with massive riser pipe and an underwater suction
unit to absorb the mineral resources from the bottom to the
vessel[8]. The design of the riser system is complicated
because of its flexiblity so the total cost of the system is
expensive. The multi agent method may be one of the
improved plans because of its general versatility and system
flexibility.
Considering about the AUV control algorithm design
procedures, it is important for the designer to synthesize the
balance between autonomy and cooperation in the AUV
character or intelligence. To determine which characteristic is
more important depends on the mission. For example, the
priority is on the autonomy side for the Fig.2 case because the
interactive situation is limited for communication between
each AUV. On the other hand, in the Fig. 1 case, the priority is
on the cooperation side because the interaction is not only on
the communication but also dynamical coupling exists in this
case. And also, in Fig.3 case, we need to implement a kind of
emergent characteristic to optimize the total behavior of the
team. To extract and solve such problems in the development
of AUV multi-agent technologies, we need to enlarge
educational opportunities for students of dealing with various
basic AUV technologies by their own hands. So we should
build up a series of basin experiment system which we can
experience various problems arising from the real system to
verify the result of simulated algorithms. The conceptual
sketch of the experimental system is shown in Fig.4. The
essential parts of the experimental system are AUV platforms,
ultrasound ranging system and ultrasound data communication
system. The AUV platforms consist of a floating commucation
interface between underwater robots and land base PC, and
underwater tiny AUV test beds. In this paper I will introduce
two types of AUV test-bed platforms and ultrasound systems
now we are developing. Although the underwater multi-agent
system will include not only the AUVs but also other non-
vehicle type robotic agents, in this research we restricted our
experimental task using AUV only. More complicated task that
needs non-vehicle type underwater robot is the next step of this
research.

11. ULTRASOUND RANGING AND LBL SYSTEM _

We are building a 1Mhz ultrasound ranging system that is _
detection. And also we are building a set of 40kHz ultrasound
pinger/transponders to calculate the underwater coordinate byFi5Ulrsudanngm ue
LBL with reference to the basin fixed coordinate system[7].
Although it depends on the acitivity range of the multi agents,
we will need both long distance ranging system and short |1800 = R1aDist.Elz] McsrdEmz] Elr[l]m25
distance ranging system in the application of underwater multi 1600
agent system. For example, when each agent works in open 1400 20
water it uses long rang system but if it ought to move into 1200 * 15
some complicated space like coral reef, cave or underwater 1000 .10
artificial structure around, it swithes the ranging system to 800
more precise mode. The switching algorithm is important and 600 m5
we want to realize the situation in the basin experiment that's 400 0
200_=
why we are now developing a precise ranging system using 0 -5
1MHz carrier frequency. 0 500 1000 1500
We designed and fabricated the ranging circuit as an Real Distance[mm]
independent module because of the maintainance and
extensibility. The ultrasound module has its own pressure tightFi6Cabrtoreutf1Mzanngmdl
the module. The phototransistor and LED is used for the
communication between the module and inner main PC which ll i1
activates the ranging. Fig.6 shows the calibration result. The ll 11 Ei_.,.. f; '
calibration is executed like Fig.5 we measured the distance i X
using the reflected ultrasound by the basin wall. Pink dots 1
indicate real distance and blue dots indicate the measured | z lr4
LBL ~~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~~hc depend reernc the theubasinn fixe coordinat systm7e
reflected carrier is decayed. However, when the distance is less I
dstaneptin, ranin
than sysep
the errors are
etwem within
inethe application
5mm, which is
ofc enough
unerate for our
multi I* __7747711 . M
3m[W] X 6m[L] X 3m[D] basin. A PIG in the main hull is used |*_
some ~~
for the interface ~ ~ the~main~ PC~ in ~the ~
between the_ coplcae space likel
hull ~and ~~exe1et coral reef,w cave oroordinate10
ranging module.
Fig.7 shows an experiment setup of 100kHz 2-dimensional
LBL system and anexperiment resul. Th,lain oo Fig.7 Two dimensional LBL experiment setup and a result
transmits 100kHz carrier and at first the transponder A calls ..
back 1lOmsec after it receives the transmitted carrier signal dere ietvt oaodtemlipt aig i. hw
from the robot. The transponder B calls back after 20msec the test condition ofthis FSK development.
whenit ecevesthecarierand he obo cacultesits The communication words which we are planning now iS like
coordinate using measured distances. The right part of the as follows-..........................
abmtxlOy20z30 \n
Fig.ure shows the detected coordinate. Now we are building The first character amxo2zo \
communication starter's isID the
nondiretuidional100khz ultranspondersnt.. . . . ,the , bt's . ,, .this case the AUVmeans iS "a".theAnd in
cte the next character
coordinate in 3 dimension. destination AUV's ID in this case "b". A space iS used to
separate the meaning block. The next block is an action
III. ULTRASOUND COMMUNICATION SYSTEM command or a message from the starter to the destination robot.
In this case "mt" means "move to directed position". Each
Ultrasound communication system.is essential for our AUV has a database of command words and this received
b asi1n exp eriment. We are bu lding a binary FSK serial data command blci. .sm
csedtomm iatoidon stem.e tllision dsire for muti-orwagen ........command list.The next block means the data or following word
communication that the ultrasound oscillators are non-directive,
But in the narrow basin the multi path fading occurs so in this
wis Uth reous can mormes d
m detais M nasin
stepwofltheeexperimentnwe usedathe oscgiator which handsh e r nt n thei 2examl fr ollow the courdiate
(10,20,30) where the robot "b" ordered to go by the robot "a".
The end of the communication is "\n".
In this binary FSK system, the signal 0 corresponds to the
carrier 35kHz and the signal 1 corresponds to the carrier 38kHz.
We selected each carrier frequency closely so as to use one
transducer and receiver. A character data is translated to 8 bits
binary code in the PIG program and integer data is translated to Transduce
16 bits binary code. Fig.9 shows the received binary signals. for~ eeved charActer
The yellow line is the raw signal received. The received binary communii At on.
signals are reprocessed by the PIC and reconstructed into the
characters or integers. The blue line shows the demodulated Fig.8 Binary FSK check result
signal from the carrier 35kHz into 0 signal. The pink line
shows the demodulated signal from the carrier 38kHz into I
signal.
The reconstructed received data is shown in Fig.8. In this
experiment we transmitted "abxl0" as 40bits binary code. Re666 4d zigra
Although the basin in this experiment is a kind of storage tank bf
so it is very small as shown in Fig.9, we can communicate with
no multi path fading. As it is inevitable for underwater robots sii
that the communication rate is very low so we must develop
the efficient communication protocol and well balanced
autonomy/cooperation algorithm using these real system.
Although there are many problems in the development of
this communication system, the first problem to be solved is _
how to share the position information of each agent that is, ........................

how to make and share the mapping information of each agent,

which changes continuously and unpredictably. It affects the
way of a generation or making algorithm of the communication Fig.9 Received binary signal of FSK
words. Especially, if the agents must work together in the
cooperative mission like underwater construction, we easily
imagine the situation that an agent needs help by its neighbors u
but how can it know who is the nearest or nearer one? So we V1
must develop underwater position information sharing
methodology in the development of underwater multi agent
technology as well as the development of efficient
communication protocol and procedures.

IV. A TINY AUV TEST BED SYSTEM

For the development of underwater robotic multi agent
thecnology, we need many (at least 3) AUV or non-AUV type Fig 10 Inner configuration oftiny test bed
test beds. So the testbed must be as low cost as possible and
also as small as possible. In our case, the basin is

hasFenoug.13shows
eq nth forsou pu..rp The presr Te hull is
showhstenobattery modueshandlnLCd doislaty.i The displayed
whic istaapearnouhfrte mobilityckvie
andcheckusviewi
hnln in our basin. i.11BteisadTM ops ipa
thprienttal
apeaane
configurati conn o fanodusytemn
d
theoevelope tinyAouprebasedwin.Fgl1eatresadTM2cmastipa
UV bed.
test Fig 11
Fig13ou show
eup ntheosyste diaramse ofe it.ssThe hull is
compospaedt sof twe
tranospaent imples imenteoptyica
cainches
two 7a acryica commuicatione using epniecmaedtctehqipen
semmui-sphere which epniecmaedt te qipen ow sdio
aret cofnnected byDand0rn seootal.sigtor sihowsth uing ner oe upyol adrdcdtdfbrcto ot h

insidedatain his aseisand

equipment TM2 cmpas
te outside equipmentouput Fig12
like an how
ultrasound geare and a four-blades propeller. We designed and fabricated
module or aland PC. The underwater connector is relatively the magnet coupling for watertight seal of the thruster to
reduce the production cost as shown in Fig.11. The total weight
in air is about 3kgf. As the total size is small (0.4m[L] X
O.4m[B] X O.2m[H]) we can handle it easily and many robots
can swim in our basin. The hull is designed to bear the pressure
at least 1 Om water depth. We selected Lithium ion battery for
power supply and the total capacity is 45OWh. By power
consumption calculation we may use it about 4 hours with one
full charge of the batteries. The main PC is PC 1 04 module
(Advantech PCM3350) of which MPU is NS-Geode 300MHz
and 64MB RAM as shown in Fig.13. 1 implemented the RT-
Linux based program system in 1 GB compact flash card from
which the system boots. It controls the total system through
PIC based subsystems. The RS-232 serial communication is
used for gaining the ranged distance from the ultrasound
controller of which MPU is PIC16F876. This RS-232 signal is
also transmitted from the inside MPU to the outside PC by
infrared data communication. So we can monitor the
information inside the pressure hull without opening it. The
program is implemented through ethernet. The frequency of
the sensing and actuation is at most 1Hz this time, the timing is
controlled by the RT-task mentioned later. An electric compass
TCM2 is used to detect the heading angle and roll, pitch. I was
afraid that the magnet compass is affected by the thruster or
inner electric current, but at present, it doesn't matter. Each I g
device like electric compass has its own PIC and all serial
communication line is multi-dropped using 12C through the __:_
communication interface PIC. The 6 underwater connector are
attached for ultrasound transducers, thrusters, power supply
and ethernet. The magnet switch is used for power on and off
of 2 DC-DC, which corresponds to 12V for thruster and 5V
for computers. The system generates maximum 4 PWM signals
for the propeller speed control with motor driver ICs. As we h
mounted a wireless CCD camera inside, we can see the
underwater image as far as the robot is near water surface. We
have just finished the basic function test in the basin by free
swimming. We have not implemented modern controller, we
will try path tracking experiment in the next step. Fig.13 System diagram of tiny test bed
V. A LARGER AUV TEST BED SYSTEM from the view point of the brittleness of the glass, wide dead
space due to its shape. And the open/close operation is really
Wentionedmadve.
tw
otyprisdesiofAU tone
tor Te platfor isetroublesome because the crude rubber tape is used for water
tight sealing. So we are now planning to develop the pressure
not only in the basin but also in the calm sea. We decided to hull for our convenience in the next step of this research.
develop the platform as reduced in size and weight as possible FigiS shows the appearance of the large AUV.
because, on the field experiment, we experienced the handling A thruster has a 70W motor with tacho-generator for the
of the AUV is very hard if the AUV is massive. From the local feedback control of the thruster revolution speed. We
aspect of cost reduction in both fabrication and operation, we prepared 2 propellers for basin experiment and calm sea field
aimed that we can handle the platform by two men at shore and experiment whose diameters are 9cm, 15cm respectively. The
we can put it into the water or out from the water using a 15cm diameter propeller's bollard maximum thrust is about
general fishing boat with no extra facilities. So we selected 2kgf. Each motor has a motor driver which generates PWM
BENTHOS 17inch pressure tight sphere glass which is widely
used for underwater measurement instrumentation. Fig.14 power signal with tacho-generator local feedback. The thrust
shows the hardware system of the large AUV. The total weight sinls
45kgf,
in air is aboutin'air s09[]X05[]T
0.9m[]X
.5m[]. This ~
Tisissze ffr
thrusters for planarD/aand thendrive. Anr
ardvertical
to motion control,
AuVbutmaythishve4
time
0.7m[L]
sabout45kgf
7mL 0.9mB] X05mH. sizewe equipped only 2 thrusters for horizontal motion direction.
is the limit that we can handle it by two men in land and put it The total battery power inside is 700Wh in this case and we
into or pull it up from our basin by hand. Regardless this size can
we will need small winch when we make an exDeriment in the w put 4 more batteries so the maximum power capacity is
1100OWh
,
in this system. The calculated maximum speed with
-,-

real. Although the glass pressure hull is not so expensive and 15cm propeller is about 0.8kt.
-

widely used for ocean research, the 6500m depth spec is too W eetdteP14cmue ytmadtemi
high for our purpose and it is not necessarily convenient for the CPU is NSGeode300MHz with 64M RAM and a 20G HDD.
non- professional users like students to handle it, The PCMCIA card for wireless LAN and A/D, D/A cards are
equipped. The wireless LAN is used for the remote login _
access from an external computer for program download or _______
sensor data acquisition. One of the important functions of the
large AUV is the interface function between the land based
computer and the underwater multi-agents so the wireless LAN
ismpotan asthe communication device. The attitude sensorUSGOE30
is TCM2 which senses the heading, pitch, roll angles and the I
sensing data is collected by RS232(COM1). The power.
consumption of PC system is less than 25W, and we estimate
each 4 thrustersm- averaged power consumption is about 30W t wbf Mt m
without disturbance like current, so that we can continue one i
experiment about 4 to 6 hours. It may be enough at this stage, I I
S
~~
~ ~ ~ ~ ~ ~~k
g
Am.l Pofw wtg M&, thr 11

but we need further good ideas to realize longer navigation 11I____W______||________A_t_C_____

'P ift M

time.
The ultrasound ranging/LBL module is separated from the
main pressure hull to avoid PWM noise generated by the motor Fig. 14 Hardware system configuration of larger AUV
drivers inside. Because the glass pressure hull is transparent,
we used a set of an infrared LED and a phototransistor for the
infrared serial data communication between main PC in the
main hull and PICs for ultrasound ranging module outside.
Generally speaking, the underwater electrical connector cost is
very high, so we designed to limit the use of the connector only
for the power transmission. We reduced the use of underwater
connector by using the infrared data communication. Once the
watertight seal is equipped we don't want to open the hull
easily so the main power switch is controlled externally using
photo sensors because the magnet switch could not work
because the glass wall was too thick.

VI. SOFTWARE SYSTEM

I selected RT-Linux V3.lpre2 as the operating system of
the both types because it is free and is possible to get hard real- Fig. 15 Total appearance of larger AUV
time operation. The VxWorks is widely used as the operating
system of many AUV platforms. But in our experiment RT-
Linux is enough. Each operations such as the thruster drive or USER PROGRAM
RS232 for ultrasound ranging control is executed by the START/STOP
periodic thread whose period and priority can be varied xHz
independently arranging each operation's necessity. The lower TASK HANDLER
level time consuming tasks such as ultrasound ranging or
sensor data acquisition are executed separately from the main Kernel Modules
thread and processed by PIC programs so the corresponding
thread is a kind of a watchdog of the PIC to synchronize the riontYb
whole threads within the desired sampling time. Because these
threads are implemented as the kernel module it is not easy for Control force
a user to control the threads externally, so we used a user Thread
process to activate or stop the kernel module through RT-FIFO
as commonly used. The acquired data is also transmitted Ultrasound
through RT-FIFO and the user process handles all time Commun C
consumptive file accesses such as acquired data recording to
the compact flash card. The command and data communication | Ultrsoun
between the main PC and each subsystem PICs mainly depend ..l...Ranging COM2
on RS-232(COM1, COM2) in this system. This situation is not
so effecttive not only in the operation but also its fabrication or LL Thread
programming, so we are now planning to develop a PIO
communication using pararell port and PC104 bus. The RT- I < RTFIFOThread
Linux based software development system is very useful for__
students to learn about hard real time task and threads based
programming. Fig.16 RT-Linux software configuration
 VII. SUMMARY
Underwater multi-agent technology is expected to broaden
the chance of underwater activities such as underwater
construction, resource development or ocean environment
monitoring. We must experience many situations and problems
to be solved for the development of its elemental technologies
by using real systems as well as our computer simulations. And
also we should make effort to broaden the popularity of
underwater technology. Educational platforms that we are able
to have students see and touch easily are very effective for that
purpose. In this paper I introduced our concept of the
experiment in the near future and the hardware/software design
of our tiny AUV testbed and ultrasound ..................

ranging/communication system for that experiment. Two types

of testbeds are designed for the basin experiment with RT-
linux based compact flash boot embedded system with NS-
Geode PClO4 CPU card and PIG based multi CPU subsystems.
We also made an 1MHz ultrasound ranging system which can
measure the distance to an obstacle within the 2cm accuracy.
The 100kHz LBL system was fabricated and checked, and
100kHz non directional LBL system is now in progress. We
made a binary FSK ultrasound data communication system
which we succeeded to transfer 3 characters and 1 integer
successive data. The undergraduate students who engaged in
making parts ofthem.
up and create experimental
this And I believesystem all enjoyed
their academic to buildN
skills were
improved extensively through this development. As the basic
component of the basin experiment was developed, we must
assemble the total system to execute some mission for group
behavior experiment in our next step of this research.
Fig. 17 Fabrication scene by students
ACKNOWLEDGMENT [5] H. Suzuki, T. Qi, K. Watanabe, "Learning Tracking Contr
The author gratefully acknowledge the contribution of ollers under Unknown Disturbances for the Installation of
undergraduate students Hiromasa Miura, Kenichi Ishise in my Rigid and Flexible Structures", SNAJ, Journal of Marin
laboratory, Saori Nakada, Masako Fukazawa, Ayano Ishihara e Science and Technology, Vol .4 No.4, pp.187-199, 1999
and Makiko Fujita in Tokai Junior College. This work was [6] K. Watanabe, H. Suzuki, K. Yoshida, "Active Control for
supported by the Grant-in-Aid for Young Scientists (B) Installation of Underwater Structures", SNAJ, Journal of
16760654. Marine Science and Technology, Vo 1.3 No.3, pp.15 1-160,
1998
REFERENCES [7] K. Watanabe, K. Sugiyama, A. Nakamura, "A Concept of
Distributed Compact AUV Groups for the Continuous M
[1] J. Guo, H.Y. Wei, F.C. Chiu, SW. Chen, "A MaximumE onitoring of Coastal Sea Environment", Proc. Oceans Tec
ntropy Method for Multi-AUV Grouping", Proc. Oceans hno-ocean 04, pp.511-516, 2004
Techno-ocean 04, pp.R325s36, 2004 [8] N. Hatta, "Hydrodynamic Examinations and Problems on
[2] P. Didao, E. Batlle, D. Ribas, M. Carreras, "NEPTUNE: Lifting Characteristics of Manganese Nodules by Air-Lif
A HIL Simulator for Multiple UUVs", Proc. Oceans Tech t Pump System", Energy Resources, Vol.50, No.1, pp.50-
no-ocean 04, pp.524-5J31, 2004 56, 1996
[3] A. Okamoto, J.J. Feeley, D.B. Edwards, R.W Wall, Rob [9] J. T. Mian, T. Ishimatsu, and Y. Nagashima, "Compact U
ust Control of a Platoon of Underwater Autonomous Vehi nderwater Vehicle with Acoustic Link for Communicatio
cles", Proc. Oceans Techno-ocean 04pp.505-510, 2004 n and Positioning,", Proc. ofthe 2000 International Symp
[4] C.A. Reeder, D.L. Odell, A. Okamoto, M.J. Anderson, D. osium on Underwater Technology, pp. 237-241, 2000.
B. Edwards, "Two-Hydrophone Heading and Range Sens [10] S. Yu, "Multi-agent based AUV System", Journal of th
or Applied to Formation-Flying for AUVs", Proc. Oceans
Techno-ocea 04, pp.5 7-523, 200 e Robotics Society ofJapan, pp. 718-722, 2004