Autonomous Robots 8, 7–24 (2000)

°
c 2000 Kluwer Academic Publishers. Printed in The Netherlands.

Design and Control of Autonomous Underwater Robots: A Survey

J. YUH
Autonomous Systems Laboratory∗, 2540 Dole St. Holmes 302, University of Hawaii, Honolulu, Hawaii 96822
yuh@[Link]

Abstract. During the 1990s, numerous worldwide research and development activities have occurred in underwa-
ter robotics, especially in the area of autonomous underwater vehicles (AUVs). As the ocean attracts great attention
on environmental issues and resources as well as scientific and military tasks, the need for and use of underwater
robotic systems has become more apparent. Great efforts have been made in developing AUVs to overcome chal-
lenging scientific and engineering problems caused by the unstructured and hazardous ocean environment. In the
1990s, about 30 new AUVs have been built worldwide. With the development of new materials, advanced computing
and sensory technology, as well as theoretical advancements, R&D activities in the AUV community have increased.
However, this is just the beginning for more advanced, yet practical and reliable AUVs. This paper surveys some key
areas in current state-of-the-art underwater robotic technologies. It is by no means a complete survey but provides
key references for future development. The new millennium will bring advancements in technology that will enable
the development of more practical, reliable AUVs.

Keywords: underwater robots, autonomous underwater vehicles, underwater navigation and control

1. Introduction structured, hazardous undersea environment make it

difficult to travel in the ocean even though today’s tech-
The ocean covers about two-thirds of the earth and has nologies have allowed humans to land on the moon and
a great effect on the future existence of all human be- robots to travel to Mars.
ings. About 37% of the world’s population lives within Most commercial unmanned underwater robots are
100 km of the ocean (Cohen et al., 1997). The ocean tethered and remotely operated, referred to as remotely
is generally overlooked as we focus our attention on operated vehicles (ROVs). This paper focuses on au-
land and atmospheric issues; we have not been able tonomous underwater robots, often called autonomous
to explore the full depths of the ocean and its abun- underwater vehicles (AUVs). Extensive use of manned
dant living and non-living resources. For example, it submersibles and ROVs are currently limited to a few
is estimated that there are about 2,000 billion tons of applications because of very high operational costs,
manganese nodules on the floor of the Pacific Ocean operator fatigue, and safety issues. The demand for
near the Hawaiian Islands. Only recently we have dis- advanced underwater robot technologies is growing
covered, by using manned submersibles, that a large and will eventually lead to fully autonomous, special-
amount of carbon dioxide comes from the seafloor and ized, reliable underwater robotic vehicles. In recent
extraordinary groups of organisms live in hydrother- years, various research efforts have increased auto-
mal vent areas. Underwater robots can help us bet- nomy of the vehicle and minimized the need for the
ter understand marine and other environmental issues, presence of human operators. A self-contained, intel-
protect the ocean resources of the earth from pollu- ligent, decision-making AUV is the goal of current re-
tion, and efficiently utilize them for human welfare. search in underwater robotics.
However, a number of complex issues due to the un- There are more than 46 AUV models. Most of the
current AUVs are survey research vehicles without ma-
∗ [Link] nipulators. Only a few of them have performed in deep
8 Yuh

Table 1. Development of autonomous underwater vehicles (AUVs) in the 1990s.

Year Vehicle Purpose Depth (m) Developer

1990 UROV-2000 Bottom survey 2000 JAMSTEC, Yokosuka, Japan

1990 No Name Testbed precise control vehicle 10 JAMSTEC, Yokosuka, Japan
1990 Musaku Testbed precise control vehicle 10 JAMSTEC, Yokosuka, Japan
1990 UUV (II) Testbed NA Draper Laboratory, Cambridge, MA
1991 AROV Search and mapping NA SUTEC, Linkoping, Sweden
1992 AE1000 Cable inspection 1000 KDD, Japan
1992 Twin Burger Testbed 50 IIS, University of Tokyo, Tokyo, Japan
1992 ALBAC Water column 300 IIS, University of Tokyo, Tokyo, Japan
1992 MAV Mine countermeasures NA DARPA, Washington, DC
1992 Doggie Bottom/sub-bottom survey 6000 Yard Ltd., Glasgow, Scotland
1992 Dolphin Water characteristics monitoring 6000 Yard Ltd., Glasgow, Scotland
1992 ABE Bottom survey 6000 WHOI, Woods Hole, MA
1992 Phoenix Testbed 10 Naval Postgraduate School, Monterey, CA
1992 ODIN Testbed 30 ASL, University of Hawaii, Honolulu, HI
1993 Ocean Voyager II Science mission 6000 Florida Atlantic University, Boca Raton, FL
1993 Odyssey II Science mission 6000 MIT Sea Grant, Cambridge, MA
1993 ARUS Bottom survey NA EUREKA (European Consortium)
1993 ODAS Survey 900 Marconi Underwater Systems, UK
1993 Hugin Survey 600 Norwegian Defense Establishment, Norway
1993 Marius Survey 600 IST, Lisbon, Portugal (w/France and Denmark)
1994 Large-D UUV Military/testbed 300 Naval Undersea Warfare Center, Newport, RI
1994 OTTER Testbed 1000 MBARI, CA
1994 Explorer Pipeline inspection 1000 Shenyang Institute of Automation, China
1995 ODIN II Shallow water 30 ASL, University of Hawaii, Honolulu, HI
1995 R1 Bottom survey 400 Mitsui Engineering, IIS, U. of Tokyo, Japan
1995 Autosub-1 Environmental monitoring 750 Southampton Oceanography Centre, UK
1996 Theseus Survey under Arctic sea-ice 1000 ISE, Canada
1997 REMUS Survey 150 Woods Hole Oceanographic Institution, MA
1997 VORAM Testbed 200 Korea Research Inst. of Ships & Ocean Engr., Korea
1998 Solar AUV Testbed N/A Autonomous Undersea Systems Institute, NH
1998 AUV-HM1 Testbed N/A National Taiwan University, Taiwan
1998 AMPS Military 200 Pacific Missile Range Facility, Kekaha, HI
1998 SIRENE Undersea shuttle 4000 DESIBEL, European project led by IFREMER, France
1999 SAUVIM Military/scientific intervention 6000 ASL, University of Hawaii, Honolulu, HI

water and under ice so the performance capabilities are AUVs have various potential applications and great
still embryonic. Table 1 shows a list of AUVs developed advantages over ROVs in terms of operational cost and
in the 1990s. Configurations of some existing AUVs are safety. However, there are still many crucial research
summarized in Table 2 and potential applications of un- issues to make the vehicle fully autonomous and reli-
derwater robots are summarized in Table 3 (Smith et al., able, such as robust underwater communication tech-
1995; Adakawa, 1995; Yoerger et al., 1991; Blidberg, nology, on-board sensors for x-y navigation, and high
1991; Bellingham and Chryssostomidis, 1993; Dane, density power source. The objectives of this paper are
1993; Adam, 1985; Robinson, 1986; Tucker, 1986; to identify key subsystems of the autonomous underwa-
Adam, 1991; Ashley, 1993; Judge, 1992; Kok et al., ter robots; to survey recent developments in each sub-
1984; Bellingham and Willcox, 1996). system; and to summarize the current state-of-the art in
Table 2. Configurations of some existing autonomous underwater vehicles.

Operating Other
AUV Year system Main CPU processors Power Thrusters Sensory system Remarks

AE 1000 1992 V × Works VME 3 DSP + image Lead-acid 3 AC magnetometers; camera; VCR Max 2 knots 1,000 m
KDD, Japan MC68040/4M processor recorder; laser; obstacle depth
avoidance sonar; Altimeter;
depthometer; accelerometers;
rate gyroscope; acoustic
transponder; radio beacon, etc.
Phoenix 1992 OS-9 GESPAC Lead-acid gel 6 with 8 control Datasonic PSA900 altitude sonar Max 1 knot 10 m
NPS, USA MC68030/2M fins ST1000, ST725; collision depth
avoidance sonar; Gyros
ABE 1992 OS-9 68CH11 T800; SAIL Lead-acid gel 6 Fluxgate compass; magnetic 2 knots 6,000 m depth
WHOI, USA network alkaline heading; angular rate sensor
lithium
Ocean 1993 V × Works VME Neuron chips; Lead-acid silver- 1 with servo Watson 3 axis angle/rate; whisker Max 5 knots 600 m
Voyager II MC68030/8M LONTalk zinc controlled sonar; sonic speedometer; depth
FAU, USA network rudder and pressure sensor; mosotech
stern plane altitude; sonar; RF modem, etc.
Odyssey II MIT, 1993 OS-9 MC68030/8M MC68HC11; Silver-zinc 1 with servo Altimeter; temp. sensor; acoustic 6,000 m depth
USA SAIL network controlled modem; obstacle avoidance
rudder and sonar; Pinger, etc.
elevator
OTTER MBARI, 1994 V × Works MVME167 MVME167; Nickel-cadmium 8 Stereo CCD; fluxgate compass Max. 4 knots
USA (68040) NDDS 2-axis inclinometer; motionpak 1,000 m depth 1
protocol 3-axis angle/rate; pressure mechanical arm
sensor; sharp sonic ranging and
positioning system
ODIN II UH, 1995 V × Works VME MC68040 Lead-acid 8 Pressure sensor; Watson 3-axis Max. 2 knots
USA angle/rate sensor; Kaiyo sonic 30 m depth 1
ranging and positioning system mechanical arm
Design and Control of Autonomous Underwater Robots
9
10 Yuh

Table 3. Potential applications of underwater robots. loop simulation) with actual parts of the vehicle and
Science • Seafloor mapping the environment is more desirable than completely nu-
• Rapid response to oceanographic and merical stand-alone simulation. Integrated simulation
geothermal events packages, including 3D graphics and virtual reality ca-
• Geological sampling pabilities, are useful for developing advanced under-
Environment • Long term monitoring (e.g., hydrocarbon water robotic vehicles since actual field-testing is very
spills, radiation leakage, pollution)
expensive (Choi and Yuh, 1993; Brutzman et al., 1992;
• Environmental remediation
• Inspection of underwater structures, Kuroda et al., 1995).
including pipelines, dams, etc. The six degrees-of-freedom nonlinear equations of
Military • Shallow water mine search and disposal motion of the vehicle are defined with respect to two
• Submarine off-board sensors coordinate systems as shown in Fig. 1. The vehicle co-
Ocean mining • Ocean survey and resource assessment ordinate system has six velocity components of motion
and oil industry • Construction and maintenance of (surge, sway, heave, roll, pitch, and yaw). The velocity
undersea structures vector in the vehicle coordinate system is expressed
Other applications • Ship hull inspection and ship tank as q̇ = [u v w p q r ]T . The global coordinate system
internal inspection OXYZ is a fixed coordinate system. Translational and
• Nuclear power plant inspection
• Underwater communication & power
rotational movement in the global reference frame are
cables installation and inspection represented by x = [x y z φ θ ψ]T that includes earth-
• Entertainment-underwater tours fixed positions and Euler angles. The equations of mo-
• Fisheries-underwater ranger tion for underwater robots without manipulators can be
written as follows:

ẋ = J(x)q̇
underwater robotic technology for future advancement.
Various subsystems are listed in Table 4. Sections of Mq̈ + C(q̇)q̇ + D(q̇)q̇ + G(x) = τ + w
this paper are devoted to dynamics; control systems; τ = Bu
navigation and sensors; communications; power sys-
tems; pressure hulls and fairing; mechanical manipu- where J(x) is a 6 × 6 velocity transformation matrix
lators; and summary and evaluation. that transforms velocities of the vehicle-fixed to the
earth-fixed reference frame; M is a 6 × 6 inertia matrix
as a sum of the rigid body inertia matrix MR and the
2. Dynamics hydrodynamic virtual inertia (added mass) MA ; C(q̇) is
a 6 × 6 Coriolis and centripetal matrix including rigid
Dynamics of underwater robotic vehicles, including body terms CR (q̇) and terms CA (q̇) due to added mass;
hydrodynamic parameter uncertainties, are highly non- D(q̇) is a 6 × 6 damping matrix including terms due
linear, coupled, and time varying. Several modeling to drag forces; G(x) is a 6 × 1 vector containing the
and system identification techniques for underwater restoring terms formed by the vehicle’s buoyancy and
robotic vehicles have been proposed by researchers gravitational terms; τ is a 6 × 1 vector including the
(Fossen, 1995; Goheen, 1995). When one or more control forces and moments; w is a 6 × 1 disturbance
manipulators are attached to the vehicle, it becomes a vector representing the environmental forces and mo-
multi-body system and modeling becomes more com- ments (e.g. wave and current) acting on the vehicle;
plicated. The effect of the hydrodynamics of each link B is a control matrix of appropriate dimensions; and
of the manipulator on vehicle motion has to be consid- u is a vector whose components are thruster forces.
ered in modeling the vehicle and manipulator (Mahesh As the robot moves underwater, additional force and
et al., 1991; McMillan et al., 1995). The effect of thru- moment coefficients are added to account for the effec-
ster dynamics on the vehicle also becomes significant, tive mass of the fluid that surrounds the robot and must
especially when the vehicle has slow and fine motion be accelerated with the robot. These coefficients are
(Yoerger et al., 1990). Therefore, accurate modeling referred to as added (virtual) mass and include added
and verification by simulation are required steps in moments of inertia and cross coupling terms such as
the design process (Lewis et al., 1984; Pappas et al., force coefficients due to linear and angular accelera-
1991). Integrated simulation (or HILS: hardware in the tions. The hydrodynamic added mass may be written
 Design and Control of Autonomous Underwater Robots 11

Table 4. Subsystems of autonomous underwater robots.

Systems Subsystems Needs/requirements Methods/models

Mission Sensors Long range information for detecting and inspecting Sonar
a target of interest
Planner Plans for the mission goals, unexpected events or Traditional planner
system failures
World modeling Set of models for the AUV system and its mission Objective & subjective
environment models
Data fusion Meaningful & correct information from massive Analytic methods, AI
data of multi-sensors
Computer Software Tools for developing computer codes for vehicle, System software, application
support and simulation systems, fault-tolerance software
operation
Hardware Integration of electronic modules in a powerful, System architecture,
robust & flexible manner communication network,
mass storage
Fault-tolerance Accommodation of hardware & software failures Redundancy design
Platform Hull Platform for mission package; depth & power Steel, aluminum, titanium,
requirements; stability; modularity for different composite, ceramic
mission parameters; materials; drag reduction
Propulsion Navigation/stationkeeping
Power Power for propulsion, mission systems, & payload
Workpackage Tools for cutting, sampling, cleaning, marking, Manipulators
stabilization, docking, retrieval & launch
Emergency Initiating appropriate action in response to the Emergency buoy, drop
abnormal vehicle condition and providing weight, flame smoke,
means for locating a disabled AUV beacon, water dye
Vehicle sensor Navigation AUV position relative to a fixed coordinate system Acoustic, Doppler,
fiber-optic gyro, GPS,
inertia system
Obstacle avoidance Detecting & avoiding obstacles: order of 50 m & Acoustic, laser
system (OAS) order of 10 degrees
Self-diagnostic Monitoring and evaluating the vehicle operational Sensors for voltage, thruster
parameters for subsystem status rpm, speed sensor, leak,
& temperature
Communication Transferring commands and data between a Fiber-optics, acoustic, radio,
surface station and vehicles laser
Development & support Logistic support Organization, equipment, spares, repair &
maintenance, documentation, etc.
Simulation Tools for testing the vehicle design and interface Stand-alone simulation,
mechanism for the analysis of the vehicle integrated simulation,
operations hybrid simulation in the
virtual environment
User interface Tools for displaying data, inputting command data Virtual reality device,
joystick, 3D graphics

with the SNAME (The Society of Naval Architects and YA = −Yu̇ · u̇ where Yu̇ = ∂Y/∂ u̇. Triantafyllou and
Marine Engineers) convention such that for the hydro- Amzallag (1984) discussed how to calculate the vari-
dynamic added mass force YA along the y-axis due ous elements in MA for different geometrical bodies. In
to a linear acceleration u̇ in the y-direction is shown an ideal fluid, MA is strictly positive and symmetrical.
12 Yuh

Figure 1. The coordinate systems for underwater robots.

Based on the kinetic energy of the fluid, E = q̇T MA q̇/2, locations. Therefore, B may not be a square matrix.
the added mass forces and moments can be derived The thruster force, u, will be the output of each thruster
by Kirchhoff’s equations (Kirchhoff, 1869; Sagatun, whose dynamics are nonlinear and quite complex. For
1992). Then, the added mass forces and moments can bladed-propeller type underwater thrusters driven by
be seen as a sum of hydrodynamic inertia forces and brushless DC motors, an experimental study to com-
moments MA q̈ and hydrodynamic Coriolis and cen- pare four thruster models including those by Yoerger
tripetal forces and moments CA (q̇)q̇. In an ideal fluid, et al. (1990) and Healey et al. (1994) was conducted by
the hydrodynamic damping matrix, D(q̇), is real, non- Whitcomb and Yoerger (1995). Under the assumption
symmetrical and strictly positive. With rough assump- that the thruster dynamics have much smaller time con-
tions such as a symmetric robot and non-coupled mo- stants than the vehicle dynamics, a simple static thruster
tion, it can be simplified to a diagonal matrix D(q̇) = model is often used as each thruster force u is propor-
diag(d1 + d2 |q̇|)i i = 1, . . . , 6 where d1 is a lin- tional to |Ä|Ä, where Ä is the thruster propeller angular
ear damping coefficient and d2 is a quadratic (drag) velocity and proportional to the motor drive voltage.
damping coefficient. In the hydrodynamic terminol-
ogy, the gravitational and buoyant forces are called
restoring forces, G(x). The gravitational forces will act 3. Control Systems
through the center of gravity while the buoyant forces
act through the center of buoyancy. Environmental dis- Major facts that make it difficult to control underwa-
turbances, w, due to waves, wind, and ocean currents ter robots include: the highly nonlinear, time-varying
and their mathematical expressions are discussed in dynamic behavior of the robot; uncertainties in hy-
detail in Fossen (1994). Components of the control drodynamic coefficients; the higher order and redun-
matrix, B, are dependent on each robot’s configura- dant structure when the manipulator is attached; distur-
tion, control surfaces, number of thrusters, and thruster bances by ocean currents; and changes in the centers
 Design and Control of Autonomous Underwater Robots 13

of the gravity and buoyancy due to the manipulator domain while the control and identification sections
motion which also disturbs the robot’s main body. It were discrete.
is difficult to fine-tune the control gains in air or dur- Yuh (1990b, 1994) has proposed a neural network
ing operation in water. Therefore, it is highly desirable control system using a recursive adaptation algorithm
to have a robot control system that has a self-tuning with a critic function (reinforced learning approach).
ability when the control performance degrades during The special feature of this controller is that the sys-
operation due to changes in the dynamics of the robot tem adjusts itself directly and on-line without an ex-
and its environment. plicit model of vehicle dynamics. Lorenz and Yuh
Various advanced underwater robot control sys- (1996) present experimental results on ODIN using
tems have been proposed in the literature, such as the method proposed by Yuh (1994). Ishii et al. (1998)
sliding control (Yoerger and Slotine, 1985; Healey have proposed a neural network based control system
and Lienard, 1993), nonlinear control (Nakamura and called “Self-Organizing Neural-Net-Controller Sys-
Savant, 1992), adaptive control (Goheen et al., 1990; tem” (SONCS) for AUVs and examined its effective-
Yuh, 1990a, 1996; Cristi et al., 1991; Tabaii et al., 1994, ness through application to the heading keeping control
Choi and Yuh, 1996; Nie et al., 1998), neural network of an AUV called “Twin-Burger.” In their study, a quick
control (Yuh, 1990b, 1994; Lorenz and Yuh, 1996; Ishii adaptation method of the controller called “Imaginary
et al., 1998), and fuzzy control (DeBitetto, 1994; Kato, Training” is used to improve the time-consuming adap-
1995). Various control architectures for autonomous tation process of SONCS. DeBitetto (1994) has inves-
underwater vehicles were also discussed in the litera- tigated a 14-rule fuzzy logic controller for the depth
ture (Valavanis et al., 1997; Girard et al., 1998). This control of a UUV. Tsukamoto et al. (1999) experimen-
section focuses on low-level control systems rather than tally implemented four model-free control systems for
control architectures. the position and velocity control of a single thruster
Yoerger and Slotine (1985) have proposed a slid- system: on-line neural net controller, off-line neural
ing mode controller for an underwater vehicle to con- net controller, fuzzy control, and non-regressor based
trol trajectory. They have investigated the effects of adaptive control. The off-line neural controller used
uncertainty of the hydrodynamic coefficients and neg- the Intel i80170 Electrically Trainable Artificial Neu-
ligence of cross-coupling terms. Healey and Lienard ral Network (ETANN) chips (Intel, 1992).
(1993) have used the sliding mode methods for the While many underwater remotely operated vehicles
control of underwater vehicles and separated the sys- (ROVs) have mechanical manipulators, most AUVs do
tem into non-interacting (or lightly interacting) subsys- not have manipulators. For a large robot, effects of the
tems, grouping certain key motion equations together arm motion on the main body may be negligible, and
for the separate functions of steering, diving, and speed the main body and arm can be considered as two sepa-
control. Nakamura and Savant (1992) have proposed a rate systems in different bandwidths. For a small robot,
nonlinear tracking control of a 4 dof (surge, roll, pitch coupled effects of the main body and arm are significant
and yaw) AUV considering kinematic motion. They and must be considered in the overall control system
have made use of the nonholonomic nature of the sys- design. With the arm attached to the vehicle, the overall
tem without considering the dynamics of the system. system becomes a multi-rigid body system. The vehi-
Goheen et al. (1990) have proposed multivariable self- cle main body continuously moves in water and high
tuning controllers as an autopilot for underwater vehi- performance of arm control, in terms of speed and ac-
cles to overcome model uncertainties while performing curacy, requires highly accurate information about the
autopositioning and station-keeping. Yuh (1996) and vehicle position and velocity. Most commercial sen-
Choi and Yuh (1996) have developed and implemented sors for vehicle position and velocity do not meet the
a new Multiple Input Multiple Output (MIMO) adap- accuracy requirements of the arm control. Therefore,
tive controller using bound estimation for underwater there are many challenging engineering problems for
robotic systems and experimented with the control sys- vehicles with manipulators.
tem on an AUV, Omni Directional Intelligent Navigator There are very few papers about the coordinated mo-
(ODIN). A hybrid adaptive control (suggesting that the tion of the vehicle and manipulator (Mahesh et al.,
procedure is a mixture of continuous and discrete op- 1991; McLain et al., 1996; Shoults, 1996; Antonelli and
erations) of an AUV was investigated by Tabaii et al. Chiaverini, 1998; and Canudas-de-Wit et al., 1998).
(1994). The system was simulated in the continuous Mahesh et al. (1991) have developed a coordinated
14 Yuh

control scheme, using a discrete-time approximation task. Several methods for fault-tolerant control of au-
of the dynamic model of underwater robotic systems, tonomous underwater robots have been discussed in the
which controls the vehicle and manipulator simultane- literature (Babcock, 1990; Healy, 1992; Dunn, 1992;
ously and compensates for end-effector errors resulting Orrick, 1994; Takai, 1994; Yang et al., 1999).
from motion of the vehicle. McLain et al. (1996) have
conducted experiments at the Monterey Bay Aquarium 4. Navigation and Sensors
Research Institute (MBARI) using the OTTER vehicle
and have shown that dynamic interaction between robot The sensory system is one of the major limitations in
arm and vehicle can be very significant. They pointed developing vehicle autonomy. The vehicle’s sensors
out that coordinated motion control strategy along with can be divided into three groups: (1) navigation sensors,
an accurate model of the arm/vehicle hydrodynamic for sensing the motion of the vehicle (Cox and Wei,
interaction forces enhance the station-keeping capa- 1994); (2) mission sensors, for sensing the operating
bility and end-effector accuracy. Shoults (1996) has environment; and (3) system sensors, for vehicle diag-
investigated a nonlinear model based control scheme nostics. Different tasks require different sensors: opti-
that simultaneously controls the position and orienta- cal, x-ray, acoustic imaging, and laser scanners for in-
tion of the vehicle and manipulator. Canudas-de-Wit spection; Doppler, sonar inertial system, and gyroscope
et al. (1998) have designed a robust nonlinear con- for navigation; sonar, magnetometer, laser scanner,
trol for a vehicle/arm system to compensate for the magnetic scanner, and chemical scanner for recovery;
coupling effects due to an onboard robot arm. They and force, tactile, and proximity sensors for construc-
have used different bandwidth characteristics of the tion. Blidberg and Jalbert (1995) described mission and
composite vehicle-manipulator dynamics as a basis for system sensors, and reviewed current navigation sen-
the controller design via singular perturbation theory. sors and sonar imaging sensors.
They have pointed out that both the robust controller Multiple sensors are often needed for the same task.
and partial linearized controller achieve similar per- For instance, information concerning the objects and
formance even in the presence of saturation. Antonelli local terrain surrounding the vehicle can be gathered via
and Chiaverini (1998) proposed a task-priority based a combination of sonar imaging, laser triangulation and
redundancy resolution scheme for kinematic control of optical imaging. Sonar can provide most of the obstacle
an underwater vehicle-manipulator system by suitably avoidance information. Video images plus specialized
using the null space vector. machine vision algorithms can provide high-resolution
The unstructured and hazardous ocean environment information concerning the shape and range of near ob-
also presents many challenging problems in the event jects and terrain. Laser triangulation can provide the
of system failures for autonomous underwater robots. same type of data at a slower rate but with the addi-
For any major failure of the robot’s subsystems, the tional capability of operating in turbid water. Geomet-
robot should rise to the surface and signal for retrieval; ric information concerning the vehicle’s surroundings
however, for any tolerable failures, the robot should from multiple sensing systems may be redundant and
be able to adjust for the failure and complete the as- conflicting. This resulting sensor fusion problem must
signed task. Therefore, an efficient and effective fault be handled by the intelligent system. An absorbing,
tolerant system becomes imperative for AUVs. A fault back-scattering, and color-distorting medium such as
tolerant system consists of three areas: fault detection, the ocean environment causes difficult problems in us-
fault isolation, and fault accommodation. The fault ing video images since the illumination is highly non-
detection process is a high-level function that moni- uniform and multidirectional. Additional complexities
tors the robot’s overall systems—both hardware and arise because the artificial light sources mounted on the
software—for any signals that exceed any preset tol- vehicle move with the vehicle. The movement of both
erance or measured sensor values. Once a fault is de- plants and fish also creates confusion in perceived bot-
tected, the fault isolation process determines the ex- tom topography.
act cause and location of the fault and its severity (i.e. Another difficulty is in x-y position sensing be-
whether it is tolerable or not). If the fault is evalu- cause there are no internal system sensors for the x-y
ated to be tolerable, the fault accommodation process vehicle position. The most common approaches that
either accommodates or reconfigures the robot’s con- current vehicles use are acoustic long baseline (LBL),
trol architecture to successfully carry out the assigned short baseline (SBL), or ultra short baseline (USBL)
 Design and Control of Autonomous Underwater Robots 15

Table 5. Doppler sonar sensors.

Litton RD Instruments EDO Corporation

(phone 804 974-2227 (phone 619 693-1178 (phone 801 486-7481
fax 804 974-2259) fax 619 695-1459) fax 801 484-3301)

Price (US $) 45490 (with valve) 24900 48000

Model SRD 500 Doppler velocity log; workhorse Model 3050 DSVS
navigator
Components Transducer, electronics, master Transducer, pressure case, Transducer, electronics
(parts) display electronics, power supply,
input/output cables,
manuals/software, spares/tool kit
Outputs Speed Speed, flux-date compass, pitch, roll, Speed
temperature
Bottom lock (m) 1.2 to 200 1 to 200 2 to 250
Speed range (m/s) Fore/Aft: −10/+25; Port/Stbd: +/−10 Fore/Aft: +/−20; Port/Stbd:
+/−5 +/−10; vertical: +/−5
Operating temperature −15 to 55 −5 to 45 0 to 35
range (◦ C)
Max Depth 600 KPA 2000 m Standard pressure: 1500 psi
Acoustic data Frequency: 307 kHz; four beams Frequency: 307 kHz; four beams Frequency: 287.5 kHz; four beams
Accuracy (bottom) Precision: <1 m/s +/−0.005 m/s; Precision: 1 m/s +/−0.003 m/s; Precision: 0.01 m/s; resolution:
>1 m/s +/−0.025 m/s 5 m/s +/−0.008 m/s; 0.004 m/s; accuracy: +/−0.1%
accuracy: +/−0.4%
Power requirements 100/115/230VAC (+/−10%); 20 to 60 VDC, 0.4 A; average 20 to 32 VDC; power at max
50/60 Hz, 100 vA power: 17 w range: 42 w
Interfaces (output) Digital: RS232, RS422, RS232, RS485, RS422 RS232, RS422
NMEA0183; analog: 1 mA
full scale or 0.1 volt/knot
Operating life (years) 10 10 N/A
Warranty (years) 2 1 1
Off shelf Yes Yes Yes
Delivery time (months) 0.5 (2 weeks) 3 4 to 6
Dimension (inches) Master display 14.4 × 10.7 × 8.5; Transducer 8.9 diameter (8.2 × 10.75 × 6 × 4.85 (without connector)
electronics 16.7 × 14.0 × 5.98; 8.9 × 8.9)
transducer: 8 diameter
Weight (lbs) Master display: 15; electronics: 15; Weight in air: 21; weight in Weight in air: 14.6; weight in
transducer: 60 water: 11 water: 8.5

methods requiring external transponders (Black and sensors, respectively. The purpose of this survey is not
Butler, 1994; Austin, 1994). However, signal attenua- to provide a complete list of commercial products with
tion varies with distance, frequency, and temperature, a best choice but to provide general information about
and positioning systems with acoustic beacons are ex- what is available commercially. This survey may be
pensive and often impractical (Vaganay et al., 1996). outdated within five years (if not sooner) since various
Sonar based navigation and localization have been efforts to improve efficiency of such products are being
studied in the literature (Auran and Silvan, 1995; Cristi made in research sectors.
et al., 1995, Marco, 1996; Vaganay et al., 1996; Nie
et al., 1998). 5. Communications
Some samples of commercial sensing systems
for underwater navigation were surveyed in 1998. The most common approach for ROV communica-
Tables 5–8 show Doppler sonars, ultra-short base- tions uses an umbilical line with coaxial cables or fiber
line sonars, inertia navigation systems, and pressure optics. This tether supplies duplex communications.
16 Yuh

Table 6. Ultra short baseline sonar sensors.

SonaTech (CA) Desert Star (CA)
(phone 805 683-1431 (phone 408 384-8000
fax 805 683-4862) fax 408 384-8062)

Price (US $) 350000 9985

Model NS-031 Pilot PT 3
Components SSBL processor unit; primary projector unit; STM-1 surface station w/ ‘over the side’
auxiliary projector unit; interconnect cable transducers; VLT-1 tracking transponder;
DiveBase-L2 multi-target tracking software
for window 95/NT includes GPS integration
and chart overlay capability
Housing material 7075-TG aluminum or 6AL4V titanium VLT-1 transponder: Al 6061T6, hard anodized
Max slant range 9000 m 1000 m (may vary between 800 m and 10000 m)
Accuracy Azimuthal bearing accuracy: <0.5 degree relative Range accuracy: +/−0.15 m (typical)
to processor unit housing for CW pulse signals;
<0.1 degree achievable with wide band signals
Operating temperature range (◦ C) −4 to 40 −4 to 50
Max depth (pressure) rating 6000 m aluminum; 9000 m titanium 1000 m
Transmitter characteristics, inter- Source level: >186 dB re 1 uPa @ 1 m @ 7 kHz; Source level: >=186 dB; multi-channel digital
rogate signals frequencies: four unique frequencies synthesized transmitter, 5 kHz–95 kHz
selectable; pulse width: 0–100 ms (4 channels in normal operation.)
programmable; max. interrogation rate: 1 per
10 seconds
Receiver characteristics Frequencies: up to 15 receive frequencies from 7 Digital synthesized superheterodyne receiver,
to 17 kHz; sensitivity, MDS (minimal operating freq. 5 kHz–95 kHz; sensitivity
detectable signal), 50% probability of <=80 dB
detection: <=80 dB re 1 uPa
Power requirements 18–36 VDC (28 VDC nominal), 3.5 A max. 9–16 [email protected] amp average during navigation
operation (1 amp peak during transmit)
Interfaces (output) ASCII RS-422 RS232C
Operating life (years) >7 Unlimited with every 3 yrs regular service
Warranty (years) 1 1
Off shelf Yes Yes
Delivery time (months) 6 months ARO 2 months
Dimension (inches) Processor unit length: 39.1 aluminum, 37.2 VLT-1: 3 D × 16.5 L
titanium, OD both 8.75
Weight (lbs.) In air: 120 aluminum, 123 titanium; in salt water: VLT-1: 3.9 lb in air, 0.5 lb in water
44 aluminum, 60 titanium

While coaxial cables would be effective for simple AUVs presents formidable challenges. The main ap-
operations with limited data transmission, fiber optic proach today for through-water transmission involves
cables can transmit more data with less electromag- acoustics in which transducers convert electrical en-
netic interference and are lighter, thinner cables. This ergy into sound waves. Since the ocean rapidly weakens
is important since cables cause substantial drag and the acoustic energy as the frequency is increased, re-
often become snagged. About ten percent of ROVs latively low frequencies are desirable for longer-range
are lost because of broken tethers. A tethered vehicle communications. But at very low frequencies, the re-
also requires an operating base and the surface mother quired transducer size is impractically large and the
ship, whose operating cost may be more than $20,000 data rates are lower. The speed and direction of sound
per day. signals vary depending on surface waves, tempera-
Research and development of untethered auto- ture, tides, and currents. Josko Catipovic and his re-
nomous vehicles is needed but communicating with search staff at Woods Hole Oceanographic Institution
 Design and Control of Autonomous Underwater Robots 17

Table 7. Inertia navigation systems.

Precision Navigation Inc. Systron Donner (BEI) Watson Industries

Price (US $) 699 13,489 12,636

Model TCM2-20 MotionPak IMU-BA604
Measurements (outputs) Heading, roll, pitch, magnetic 3 outputs for linear acceleration, Heading, roll, pitch, and their
field data, temperature 3 outputs for angular velocity, rates, acceleration in 3 axis
temperature
Max roll and pitch (degree) +/−20 N/A +/−180 roll; +/−90 pitch
Power consumption (mA) 15–20 (depends) <270 <600
Operating temperature range (◦ C) −20 to 70 −40 to 80 −30 to 60
Sampling rate (Hz) 1–30 >60 for rate; >300 for acce- 71 (depends)
leration
Operating life (years) 8.2 (72000 hours) 10 5.7 (50000 hours)
Accuracy (degree if not specified) Heading: when level: Threshold/resolution: Rate resolution: <0.025/SEC;
+/−0.5 RMS; when tilted: <= 0.004/sec for rate; rate bias: <0.5% of F.S.;
+/−1 RMS; resolution: 0.1; <= 10 ug for acceleration acceleration resolution: better
repeatability: +/−0.1 (RMS: than 2 mG’s; acceleration
Root mean square) bias: <10 mG’s
Accuracy (degree if not specified) Tilt: accuracy: +/−0.2; Bank and elevation accuracy:
resolution: 0.1; repeatability: 0.2 to 20, static, 2% dynamic,
+/−0.2; range: +/−20 1% F.S. static to 60
Input voltage (VDC) +5 regulated (or 6 to 18 + and −15 +/−10% +12
unregulated)
Interfaces (output) Digital: RS232, NMEA0183, Analog: +/−2.5 VDC for rate; Digital: RS232; analog +/−10
analog: 0–5 V linear; +/−7.5 VDC for acceleration VDC
0–5 V quadrature
Warranty (years) 1 1 1
Off shelf Yes Yes Yes
Delivery time (weeks) 4 to 6 6 to 8 4 to 6
Dimension (inches) 2.50 × 2.00 × 1.25 3.6 × 3.05 × 3.05 6.5 × 6.5 × 3
Weight (lbs.) 1.6/16 (1.6 ounces) 1.98 (900 g) <4

have estimated the characteristics of the water chan- for current AUVs rely on batteries that supply limited
nel through which a signal will travel and adjusted the energy. A typical battery type is lead-acid. Silver-zinc
signal accordingly (Fricke, 1994). Acoustic modems, offers roughly double the energy density of lead-acid
at a 1,200-baud rate were developed, are sufficient for batteries. However, silver-zinc batteries are expensive.
sending oceanographic data and transmitting video im- For example, a 325-kWh silver-zinc battery is about
ages. Herold and Johnson (1994) describe a compact US $400,000. Low-cost, high-density batteries, which
underwater acoustic modem for the shallow water envi- provide the vehicle with more than 24-hours endurance,
ronment. Chappell et al. (1994) describe acoustic com- are desired. Fuel cells or fuel-cell-like devices which
munication between two AUVs. Table 9 shows some are more energetic than silver-zinc batteries are being
of the recent acoustic modem systems. considered. Active research and development in the
area of batteries has been in progress, especially with
recent attention on electric vehicles that has accelerated
6. Power Systems the development of more efficient and safer batteries.
In the near future, the underwater robotics community
While tethered ROVs can be powered by the mother is expected to receive a great benefit from this develop-
ship, operating hours of untethered robots are limited ment. Specific energy comparisons of current batteries
by the on-board power system. Most power systems and fuel cells are listed in Table 10.
18 Yuh

Table 8. Pressure sensors.

Data Instruments, Inc. BEI Electronics, Inc. Tavis Corp. Parascientific, Inc.
(tel: 800 333-3282 (tel: 818 362-0300 (tel: 800 842-6102 (tel: 425 883-8700
fax: 508 263-0630) fax: 818 362-2487) fax: 209 966-4930) fax: 425 867-5407)

Price (US $) 1108 2720 2099 5900

Model DS (option 2) 6-142-1325-01 P110C 8B7000-L
Submersible in seawater, Yes Yes Not completely Yes
max one month
Temperature Yes Yes Yes Yes
compensated
Operating temperature −18 to 93 −4 to 85 −18 to 85 −25 to 65
range (◦ C)
Compensated temper- −1 to 54 −1 to 71 −1 to 54 −2 to 40
ature range (◦ C)
Pressure (psi) Yes Over range 15,000 Yes Yes
Accuracy Overall: 0.5%; linearity: Overall: +/−0.5% Overall: +/−0.5% Each: +/−0.01%
0.34%; hysteresis: 0.15%
Material Housing: 316 stainless 316 ss External: 17-4 PH Varies (titanium
steel (ss); diaphragm: and 300 ss possible)
15-5 PH stainless steel;
connector: elastomer
Excitation voltage 14–50 6–26 17–50 6–25
(VDC)
Output 16+/−0.16 mA into 0 to 4–20 mA 4–20 mA RS-232
2000 ohm loop resistance
(4–20 mA)
Warranty (years) 1 1 1 5
Off shelf Yes Yes Yes Yes
Delivery (weeks) 6 8 16 8
Dimension (inch. (mm)) 7.25 × 1.375 (184 × 36) 4.11 × 1.75 (104 × 44) 3.65 × 2.2 (93 × 56) 10.48 × 2.2 (56 × 27)
Weight (lbs.) 2.5 2 (max) 3 (max) 6

Table 9. Acoustic modem systems.

Developer Data rate (kbps) Range (km) Modulation

Datasonics (Fang et al., 1987) 5 1 Incoherent

WHOI (Eastwood et al., 1996) 5 5 Incoherent
WHOI (Johnson et al., 1994) 5 Shallow, under ice Coherent
JAMSTEC (Sasaki and Suzuki, 1992) 16 6.5 vertical Coherent
IFREMER/ORCA (Ayela and Coudeville, 1991) 19.2 2 vertical Coherent
Oki Elec. (Kaya and Yauchi, 1989) 500 0.06 Coherent

7. Pressure Hulls and Fairings the use of special materials that have rigidity, strength,
and environmental resistance. Many ROVs use open-
Water pressure on the vehicles can be enormous. The frame structures with a few pressure hulls while many
deep oceans range from 6,000 to 11,000 m. At a mere AUVs have torpedo-shaped fairings that include a few
10 m depth, the pressure will be twice the normal pressure hulls for on-board electronics and batteries.
atmosphere pressure of 203 kPa. The chemical envi- Table 11 shows the pros and cons of different shapes of
ronment of the sea is highly corrosive, thus requiring fairings. The most common materials for pressure hulls
 Design and Control of Autonomous Underwater Robots 19

Table 10. Specific energy comparison of batteries and are limited to survey-type applications. Unlike
and fuel cells. stationary industrial manipulators in factories, under-
System Energy/weight (Watt-hr/lb.) water manipulators are attached to vehicles that are
constantly moving. Therefore, it is quite difficult and
Lead-acid 10–18
tedious to operate these manipulators with accuracy.
Ni-Cd 12–20
Teleoperation using a master/slave system is a com-
Ni-Fe 20–25
mon approach for ROVs. In the offshore oil indus-
Ag-Cd 18–45 try, teleoperated manipulators are used on the tethered
Ag-Zn 40–48 ROVs. These vehicles often use two arms—one to latch
Hi-H2 80–90 onto the structure for stability and the other to per-
Acid fuel cells 70–460 form tests and maintenance. For multi-task operations,
Alkaline fuel cells 110–430 more than one type of manipulator end-effector may be
needed. To change the end-effector with the current ve-
hicle system, the vehicle must be brought to the surface
are aluminum or titanium. Recently, composite mate- and the end-effector changed for each task. This proce-
rials are being considered. The potential advantages dure is time-consuming and expensive. A flexible and
of composite materials for undersea pressure hulls are dexterous design of the end-effector and workpackage
well known and numerous research and development is necessary to carry out multi-task and sophisticated
projects are underway (Walton, 1991; Du Pont, 1991; operations. Lane (1995) and Davies et al. (1998) pre-
Anderson et al., 1992; Davies et al., 1993). Pressure sented a new design of the underwater end-effector that
hull materials are summarized in Table 12. has flexible fingers for dexterous operation.
As mentioned above, most underwater arms were
designed for ROVs and they are actuated by hydraulic
8. Mechanical Manipulators drivers. They are also designed for teleoperation rather
than robotic operation. Hydraulic drivers are disad-
Mechanical manipulators are needed for underwater in- vantageous for AUVs or underwater robots because of
tervention missions. While many ROVs are equipped high power requirements and noise generation. Elec-
with one or two arms, most AUVs do not have arms trical drivers are preferred for AUVs that have limited

Table 11. Comparison of various vehicle (fairing) shapes.

Advantages Disadvantages

Single sphere Low weight/vol. (W/V) ratio, excellent Low optimum vehicle length/diameter
for deep diving vehicles (L/D) ratio
Cylinder Ease of fabrication, high optimum High W/V ratio, end closures
vehicle L/D ratio
Saucer Improved hydrodynamics in horizontal Inefficient structure, low controllability,
plane, ease of hovering in currents limited to shallow depths
Egg Good hydrodynamics, good W/V ratio Difficult to design & fabricate

Table 12. Comparison of pressure hull materials.

Steel alloy Aluminum alloy Titanium alloy C/peek composite Ceramic

Ultimate stress (Kpsi) 60 73 125 300 100

Density (lb/in3 ) 0.283 0.1 0.16 0.056 0.13
Fabrication Excellent Very good Good Fair Fair
Corrosion resistance Poor Fair Very good Excellent Excellent
Magnetic susceptibility Very high Medium High Very low Very low
Relative cost Very low Very low Moderate Moderate Moderate
 20
Yuh

Table 13. Commercial underwater manipulators.

International Submarine
Tecnomare, Italy Western Space and Marine, Inc. Kraft TeleRobotics, Inc. Schiling Robotics Systems, Inc. Engr. Ltd., Canada

Model Telemanipulator The Arm-66 Predator Titan III S ISE 7F

D.O.F. 6 plus gripper 6 plus gripper 6 plus gripper 6 plus gripper 6 plus gripper
Master/slave Master/slave Master/slave Master/slave Master/slave Master/slave
Power source 220 V-50 Hz/110 V-60 Hz, 110–240 VAC, 50/60 Hz, 47–63 Hz, 105–250 VAC 50/60 Hz, 90–260 VAC and Hydraulic power @ 1250 Psi,
optional hydraulic power-2 GPM and hydraulic power hydraulic power @ 3,000 Psi, 10 LPM, 25 micron filter
@ 3000 Psi, 5–25 Micron @ 2000 Psi, 5 GPM, 1.5–5.0 GPM, 10–200 cSt.
absolute 25 micron absolute
Material Aluminum alloy type 6000 Aluminum, stainless steel Aluminum with teflon 6–4 titanium & 316 stainless 6061-T6 Aluminum
composites, corrosion coating
isolation system
Joint sensors Resolver at each joint and Position, velocity, and torque Position and force feedback Resolver Potentiometers
torque sensor at the output
shaft
Force/torque Jaw closure force: 700 N Jaw force controls and sets Jaw closure force: 300 lbs., Gripping force: 1,000 lbf., wrist Gripping force: 330 lbs.,
grip 0–350 lbs. (120 lb/ft) wrist torque: 100 [Link] torque: 125 [Link] wrist torque: 140 ft/lb.
Actuator DC motor - brushless Hydraulic cylinders Hydraulic cylinders Hydraulic cylinders Hydraulic cylinders
Price $350,000–410,000 $280,000+ $119,000 $149,500 $70,000–250,000
Max reach 2.07 m 66 inches 80 inches 75.4 inches 59 inches
Payload 40 kg 145 lbs. 200 lbs. 250 lbs. 650 lbs. @ 1.4 m
 Design and Control of Autonomous Underwater Robots 21

power and use noise-sensitive sensors for the robotic power consumption and low noise operation rather than
operation. The University of Hawaii, the University teleoperation since AUVs have limited on-board power
of Genoa, and Ansaldo (Italy) jointly developed an resources and various noise-sensitive sensors.
electro-mechanical underwater manipulator. Table 13 This paper surveyed the current state-of-the-art in
lists samples of underwater manipulators currently key areas of the underwater robotics for underwater
available. robots or AUVs. While not providing a complete sur-
vey, it is hoped that this survey can help provide a
9. Summary and Evaluation direction for future advancements in the subject area
and attract more researchers and potential users of such
Underwater robotics represents a fast growing research robots. One of the key areas that this paper did not cover
area and promising industry as advanced technologies is motion planning. There have been many methods
in various subsystems develop and potential applica- proposed for land mobile robots but only a few pro-
tion areas are explored. As shown in Table 2, common posed for underwater robots (Vasudevan and Ganesan,
operating systems include V × Works and OS9. As 1996; Heart et al., 1996; Sugihara and Yuh, 1997). The
PC technologies quickly become outdated, more low- new millennium will bring advancements in techno-
cost solutions such as Linux are emerging in AUV re- logy that will enable the development of more practical,
search. New vehicles may take an advantage of this reliable AUVs.
new development while many existing vehicles con- More information about recent development in au-
tinue their development using current operating sys- tonomous underwater vehicles (robots) can be ob-
tems. However, different operating systems would not tained from various resources. The technical commit-
affect much on the direction of the basic research in tee on Underwater Robotics of the IEEE Society of
other subsystems such as vehicle servo-level control. Robotics and Automation continually updates its World
Research activities will increase in multi-vehicle oper- Wide Web homepage ([Link]
ation area, including communication between vehicles, ME/Research/URTC/[Link]). It lists recent re-
intelligent control architecture, and intelligent motion search and development activities such as conferences
planning. Virtual reality is also a very attractive field and workshops, and the page provides links to research
for underwater robotics. Most of the time, visual im- institutions worldwide that are involved in underwater
ages of the vehicle operation may not be available due robotics. Photos of various underwater robots can be
to a lack of light in the deep ocean. It is often preferred found throughout this web site. Related technical soci-
to have a synthetic environment with actual images (if eties include the Marine Technology Society (MTS),
available), graphic images created by sensor feedback IEEE Oceanic Engineering Society, IEEE Robotics
signals, and a world model from stored data. Introduc- and Automation Society, and IFAC Technical Com-
ing virtual reality technology will enhance interaction mittee on Marine Systems. Technical meetings spon-
between the vehicle and operator. sored by these societies include the IEEE Symposium
Advanced development in navigation sensors is nec- on Autonomous Underwater Vehicle Technologies;
essary for more reliable and accurate performance re- International Symposium on Unmanned Unthethered
quired by many potential applications. Current naviga- Submersible Technology; International Conference on
tion sensors, as discussed in Section 4, still have a low Robotics and Automation, Underwater Technology,
performance-to-cost ratio. Because of the hazardous Underwater Intervention; ROVs; Control Applica-
and noisy environment, it is difficulty to get robust tions in Marine Systems (CAM); and Oceans. Reg-
outputs from sensors. It would be desirable to have on- ular journals and magazines include the IEEE Jour-
board navigation sensors, avoiding physical constraints nal of Oceanic Engineering and Sea Technology. Two
of some current sensors such as LBL. Advancement in books in underwater robotics were recently published:
underwater communication, high-density battery, and Underwater Robotic Vehicles—Design and Control,
new materials for pressure vessels will also enhance TSI Press (1995) and Underwater Robots, Kluwer
the development in underwater robotics. Publisher (1996).
As the ROV industry continues to grow, develop-
ment of hydraulic-driven arms with high telepresence Acknowledgments
will continue. Table 13 shows only a subset of what
each company produces and all of their products are This research was sponsored in part by the NSF PYI
hydraulically driven teleoperating arms. As mentioned Award (BES91-57896), NSF (BES97-01614), ONR
earlier, AUVs need robotic manipulators with low (N00014-97-1-0961), SNU ERC-ACI, and a grant
22 Yuh

agreement (NA86RG0041) from NOAA of the Dept. Conference on Robotics and Automation, Leuven, Belgium,
of Commerce (R/ES-4). The views expressed herein pp. 452–457.
Chappell, S.G., Jalbert, J.C., Pietryka, P., and Duchesney, J. 1994.
are those of the authors and do not necessarily reflect
Acoustic communication between two AUVs. In Proceedings on
the views of funding agencies. UNIHI-SEAGRANT- IEEE AUV Technology, pp. 462–469.
JC-99-04. Choi, S.K. and Yuh, J. 1993. Design of advanced underwater robotic
vehicle and graphic workstation. In Proceedings IEEE Int’l Conf.
on Robotics and Automation, Vol. 2, pp. 99–105.
References Choi, S.K. and Yuh, J. 1996. Experimental study on a learning control
system with bound estimation for underwater vehicles. Int’l J. of
Adakawa, K. 1995. Development of AUV: Aqua explorer 1000. In Autonomous Robots, 3(2/3):187–194.
Underwater Robotic Vehicles: Design and Control, J. Yuh (Ed.), Cohen, J.E., Small, C., Mellinger, A., Gallup, J., and Sachs, J.
TSI: Albuquerque. 1997. Estimates of coastal populations. Science, 278(5341):1211–
Adam, J.A. 1985. Probing beneath the sea. IEEE Spectrum, 1212.
22(4):55–64. Cox, R. and Wei, S. 1994. Advances in the state of the art for AUV
Adam, J.D. 1991. Using a micro-sub for in-vessel visual inspection. inertial sensors and navigation systems. In Proceedings on IEEE
Nuclear Europe Worldscan, 10:5–6. AUV Technology, pp. 360–369.
Anderson, S.M., Newman, K., Lamontia, M.A., and Olson, B. 1992. Cristi, R., Caccia, M., Veruggio, G., and Healey, A.J. 1995. A sonar
Design, analysis and hydrotesting of a composite-aluminum cylin- approach to AUV localization. In CAMS’95, May, Trondheim,
der joint for pressure hull applications, ASTM/STP on Compres- Norway, pp. 291–298.
sion Response of Composite Structures. Cristi, R., Papoulias, F.A., and Healey, A.J. 1991. Adaptive sliding
Antonelli, G. and Chiaverini, S. 1998. Task-priority redundancy res- mode control of autonomous underwater vehicles in the dive plane.
olution for underwater vehicle-manipulator systems. In Proceed- IEEE J. of Oceanic Engineering, 15(3):462–470.
ings of IEEE International Conference on Robotics and Automa- Dane, A. 1993. Robots of the deep. Popular Mechanics, 170(6):104–
tion, Leuven, Belgium, pp. 768–773. 105.
Ashley, S. 1993. Voyage to the bottom of the sea. Mechanical Engi- Davies, P., Rannou, F., Cantwell, W.J., Pomies, F., and Carlsson,
neering, 115(12):52–57. L.A. 1993. Durability of composite materials in a marine
Auran, P.G. and Silven, O. 1995. Ideas for underwater 3D sonar environment—a fracture mechanics approach. In Proceedings of
range sensing and environmental modeling. In CAMS’95, May, ICCM-9, Madrid, Spain, Vol. II, pp. 308–315.
Trondheim, Norway, pp. 284–290. Davies, J.B.C., Lane, D.M., Robinson, G.C., O’Brien, D.J., Pickett,
Austin, T.C. 1994. The application of spread spectrum signaling M., Sfakiotakis, M., and Deacon, B. 1998. Subsea applications of
techniques to underwater acoustic navigation. In Proceedings on continuum robots. In UT98, pp. 363–369.
IEEE AUV Technology, pp. 443–449. DeBitetto, P.A. 1994. Fuzzy logic for depth control of unmanned
Ayela, G. and Coudeville, J.M. 1991. TICA: A long range, high undersea vehicles. In Proceedings of Symposium of Autonomous
baud rate image/data acoustic transmission system for underwater Underwater Vehicle Technology, pp. 233–241.
applications. In Underwater Defense Technology Conference. Dougherty, F., Woolweave, G. et al. 1990. At-sea testing of an un-
Babcock, P.S. and Zinchuck, J.J. 1990. Fault detection design opti- manned underwater vehicle flight control system. In Proceedings
mization: Application to an autonomous underwater vehicle nav- of Symposium of Autonomous Underwater Vehicle Technology, pp.
igation system. In Proceedings IEEE AUV’90, Washington, D.C. 65–73.
Bellingham, J.G. and Chryssostomidis, C. 1993. Economic ocean Du Pont Co. 1991. Advanced submarine technology—thermoplastic
survey capability with AUVs. Sea Technology, 34:12–18. materials program. Phase IIA Final Report, DARPA Contract #
Bellingham, J.G. and Willcox, J.C. 1996. Optimizing AUV oceano- MDA972-89-0043.
graphic surveys. IEEE Sym of AUVT, pp. 391–398. Dunn, S.E. and Rae, G.J.S. 1992. On-line damage detection for
Black, M.R. and Butler, B. 1994. Arctic ocean trials of trackpoint autonomous underwater vehicles. In IEEE AUV’94, pp. 383–
ultrashort baseline acoustic positioning system. In Proceedings on 392.
IEEE AUV Technology, pp. 297–302. Eastwood, R.L., Freitag, L.E., and Catipovic, J.A. 1996. Acoustic
Blidberg, D.R. and Jalbert, J. 1995. AUV mission & system sensors. communication system for the AMMT program. IEEE Oceans,
In Underwater Robotic Vehicles: Design and Control, J. Yuh (Ed.), pp. 87–92.
TSI: Albuquerque. Fang, C., Eck, C., and Porta, D. 1987. High speed digital acous-
Blidberg, D.R. and Turner, R. 1995. Mission planner. In Under- tic telemetry system. In International Symposium on Unmanned
water Robotic Vehicles: Design and Control, J. Yuh (Ed.), TSI: Untethered Submersible Technology, pp. 348–362.
Albuquerque. Fossen, T.I. 1994. Guidance and Control of Ocean Vehicles, John
Blidberg, D.R. 1991. Autonomous underwater vehicles: A tool for Wiley & Sons Ltd.
the ocean. Unmanned Systems, 9(2):10–15. Fossen, T.I. 1995. Underwater vehicle dynamics. In Underwa-
Brutzman, D.P., Kanayama, Y., and Zyda, M.J. 1992. Integrated sim- ter Robotic Vehicles: Design and Control, J. Yuh (Ed.), TSI:
ulation for rapid development of autonomous underwater vehicles. Albuquerque.
In IEEE AUV’92, Washington, D.C. Fricke, J.R. 1994. Down to the sea in robots. Technology Review,
Canudas-de-Wit, C., Diaz, E.O., and Perrier, M. 1998. Robust 10(1):46.
nonlinear control of an underwater vehicle/manipulator system Goddard, R.P. 1989. The sonar simulation toolset. In Proceedings of
with composite dynamics. In Proceedings of IEEE International Oceans ’89, pp. 1217–1222.
 Design and Control of Autonomous Underwater Robots 23

Goheen, K. 1995. Techniques for URV modeling, In Underwa- of an underwater vehicle and robotic manipulator. J. of Robotic
ter Robotic Vehicles: Design and Control, J. Yuh (Ed.), TSI: Systems on Underwater Robotics, 8(3):339–370.
Albuquerque. Marco, D.B. 1996. Autonomous control of underwater vehicles and
Goheen, K.R. and Jeffery, R.E. 1990. Multivariable self-tuning local area maneuvering. Ph.D. Dissertation, Naval Post Graduate
autopilots for autonomous and remotely operated underwater ve- School, Monterey, CA.
hicles. IEEE Journal of Oceanic Engineering, 15(3):144–151. McLain, T.W., Rock, S.M., and Lee, M.J. 1996. Experiments in the
Girard, A.R., Smith, S.M., and Ganesan, K. 1998. A convenient form coordinated control of an underwater arm/vehicle system. Under-
for discrete event control of autonomous underwater vehicles. In water Robots, J. Yuh, T. Ura and G.A. Bekey (Eds.), Kluwer:
IARP’98, pp. 125–130. Boston, pp. 139–158.
Herold, D. and Johnson, M. 1994. A compact underwater acoustic McMillan, S., Orin, D.E., and McGhee, R.B. 1995. DynaMechs: An
modem. In Proceedings on IEEE AUV Technology, pp. 393–398. object oriented software package for efficient dynamic simulation
Healey, A.J. and Lienard, D. 1993. Multivariable sliding mode con- of URVs. In Underwater Robotic Vehicles: Design and Control,
trol for autonomous diving and steering of unmanned underwater J. Yuh (Ed.), TSI: Albuquerque.
vehicles. IEEE Journal of Oceanic Engineering, 18(3):327–339. Nakamura, Y. and Savant, S. 1992. Nonlinear tracking control of au-
Healey, A.J. and Marco, D.B. 1992. Slow speed flight control of tonomous underwater vehicles. In Proceedings of IEEE Int. Conf.
autonomous underwater vehicles: Experimental results with NPS on Robotics and Automation, Vol. 3, pp. A4–A9.
AUV II. In Proceedings ISOPE, pp. 523–532. Nie, J., Yuh, J., Kardash, E. and Fossen, T.I. 1998. Onboard sensor-
Healey, A.J., Rock, S.M., Cody, S., Miles, D., and Brown, J.P. 1994. based adaptive control of small UUVs in the very shallow water.
Toward an improved understanding of thruster dynamics for un- In IFAC-Control Applications in Marine Systems, Fukuoka, Japan,
derwater vehicles. In Proceedings of the 1994 IEEE Symposium pp. 201–206.
on AUV Technology, Boston, MA, pp. 340–352. Orrick, A. and McDermott, M. 1994. Failure detection in an
Healey, A.J. 1992. A neural network approach to failure diagnostics autonomous vehicle. In Proceedings of IEEE Symposium on
for underwater vehicles. In IEEE AUV’92, pp. 131–134. Autonomous Underwater Vehicle Technology, AUV ’94, pp. 377–
Heart, S., Tiwari, S., and Lumelsky, V. 1996. A terrain-covering 382.
algorithm for an AUV. Underwater Robots, J. Yuh, T. Ura, and Pappas, G., Shotts, W., O’Brien, M., and Wyman, W. Spring, 1991.
G.A. Bekey (Eds.), Kluwer: Boston, pp. 17–45. The DARPA/NAVY unmanned undersea vehicle program. Un-
Intel Corp. 1992. iNNTS Neural Network Training System. User’s manned Systems, 9(2):24–30.
Guide. Intel. Robinson, R.C. 1986. National defense applications of autonomous
Ishii, K., Fujii, T., and Ura, T. 1998. Neural network system for on- underwater vehicles. IEEE J. Oceanic Engineering, OE-11(4).
line controller adaptation and its application to underwater robot. Roman, H.T. 1990. Robot applications in nuclear power plants.
In Proceedings of IEEE International Conference on Robotics & Newsletter of the IEEE Robotics and Automation Society, pp. 8–9.
Automation, pp. 756–761. Sagatun, S.I. 1992. Modeling and control of underwater vehicles:
Johnson, M., Herold, D., and Catipovic, J. 1994. The design and per- A Lagrangian approach. Dr. Ing. Dissertation, The Norwegian
formance of a compact underwater acoustic network node. IEEE Institute of Technology, Norway.
Oceans, pp. 467–471. Sasaki, T. and Suzuki, M. 1992. Digital acoustic image transport
Judge, Jr. J. 1992. Remote operated vehicles—a driving force for system for deep sea research submersible. IEEE Oceans, pp. 567–
improved outages. Nuclear Engineering International, 37:34–36. 570.
Kato, N. 1995. Applications of fuzzy algorithm to guidance and Shoults, G.A. 1996. Dynamics and control of an underwater robotic
control of underwater vehicles. In Underwater Robotic Vehicles: vehicle with an N-axis manipulator. Ph.D. Thesis, Washington
Design and Control, J. Yuh (Ed.), TSI: Albuquerque. University.
Kirchhoff, G. 1869. Uber die Bewegung eines Rotationskorpers in Smith, S., Dunn, S., Betzer, P., and Hopkins, T. 1995. Design of
einer Flussigkeit, Crelle’s J. No. 71, pp. 237–273. AUVs for coastal oceanography. In Underwater Robotic Vehicles:
Kok, K., Law, T., Bartilson, B., Renner, G.F., and Rosen, K. 1984. Design and Control, J. Yuh (Ed.), TSI: Albuquerque.
Application of robotic systems to nuclear power plant maintenance Sugihara, K. and Yuh, J. 1997. GA-based motion planning for un-
tasks. In Proceedings of the 1984 National Topical Meeting on derwater robotic vehicles. In Proceedings of 10th International
Robotics and Remote Handling in Hostile Environments, pp. 161– Symposium on Unmanned Untethered Submersible Technology
168. (UUST-10), Durham, NH, September 7–10, pp. 406–415.
Kuroda, Y., Aramaki, K., Fujii, T., and Ura, T. 1995. A hybrid envi- Tabaii, S.S., El-Hawary, F., and El-Hawary, M. 1994. Hybrid adap-
ronment for the development of underwater mechatronic systems. tive control of autonomous underwater vehicle. In Proceedings of
In IECON. Symposium of Autonomous Underwater Vehicle Technology, pp.
Kaya, A. and Yauchi, S. 1989. An acoustic communication system 275–282.
for subsea robot. IEEE Oceans, pp. 765–770. Takai, M. and Ura, T. A model based self diagnosis system for
Lane, D.M. 1995. Subsea robotics for the offshore industry. In IEEE autonomous underwater vehicles, Institute of Industrial Science,
Workshop on Robotic Technologies in Oceanic Engineering, Yuh, University of Tokyo.
(Ed.), IEEE Robotics and Automation. Tarn, T.J., Shoults, G.A., and Yang, S.P. 1996. A dynamic model of
Lewis, D.J., Lipscomb, J.M., and Thompson, P.G. 1984. The simu- an underwater vehicle with a robotic manipulator using Kane’s
lation of remotely operated underwater vehicles. In ROV’84. method. Underwater Robots, J. Yuh, T. Ura, and G.A. Bekey
Lorentz, J. and Yuh, J. 1996. A survey and experimental study of (Eds.), Kluwer: Boston, pp. 195–209.
neural network AUV control. In IEEE AUV’96, Monterey, CA. Triantafyllou, M.S. and Amzallag, A.M. 1984. A new generation
Mahesh, M., Yuh, J., and Lakshmi, R. 1991. A coordinated control of underwater unmanned tethered vehicles carrying heavy equip-
24 Yuh

ment at large depths. Technical Report MITSG 85-30TN, MIT Yuh, J. 1996. An adaptive and learning control system for under-
Dea Grant, Boston, MA. water robots. In 13th World Congress International Federation of
Tsukamoto, C.L., Yuh, J., Choi, S.K., Lee, W.C., and Lorentz, J. Automatic Control, San Francisco, CA, Vol. A, pp. 145–150.
1999. Experimental study of advanced controllers for an under- Yuh, J., Ura, T., and Bekey, G.A. (Eds.), 1996. Autonomous Under-
water robotic vehicle thruster system. International Journal of water Robots, Kluwer: Boston.
Intelligent Automation and Soft Computing, 5(3):225–238.
Tucker, J.B. 1986. Submersibles reach new depths. High Technology,
pp. 17–24.
Vaganay, J., Bellingham, J.G., and Leonard, J.J. 1996. Outlier rejec-
tion for autonomous acoustic navigation. In Proceedings of IEEE
International Conference on Robotics and Automation, pp. 2174–
2181.
Valavanix, K.P., Gracanin, D., Matijasevic, M., Kolluru, R., and
Demetriou, G.A. 1997. Control architectures for autonomous un-
derwater vehicles. IEEE Control Systems Magazine, pp. 48–64.
Vasudevan, C. and Ganesan, K. 1996. Case-based path planning for
AUVs. Underwater Robots, J. Yuh, T. Ura, and G.A. Bekey (Eds.),
Kluwer: Boston, pp. 5–15. Dr. J. Yuh is currently a Professor and Graduate Chair of the De-
Walton, J.M. 1991. Advanced unmanned search systems. In partment of Mechanical Engineering at the University of Hawaii
Oceans’91, pp. 1392–1399. in Honolulu, Hawaii, U.S.A. He is also Director of the Autonomous
Whitcomb, L.L. and Yoerger, D.R. 1995. Comparative experiments Systems Laboratory. His major research focus is underwater robotics.
in the dynamics and model-based control of marine thrusters. In Dr. Yuh received a 1991 National Science Foundation Presidential
CD Proceedings of the 1995 IEEE Oceans, San Diego, CA. Young Investigator Award from U.S. President George Bush, a 1991
Yang, K.C., Yuh, J., and Choi, S.K. 1999. Fault-tolerant system Boeing Faculty Award, a UH Fujio Matsuda Fellow award, and a
design of an autonomous underwater vehicle—ODIN: An experi- 1989 DOW Outstanding Young Faculty Award from the American
mental study. International Journal of Systems Science, accepted. Society for Engineering Education. He is listed in Who’s Who in the
Yoerger, D.N. and Slotine, J.E. 1985. Robust trajectory control of World, Who’s Who in the West, and Men of Achievement. He has
underwater vehicles. IEEE J. of Oceanic Engineering, OE-10(4): published over 70 technical articles and edited/co-edited six books
462–470. and two journal special issues on underwater robotics. He serves
Yoerger, D.N., Cooke, J.G., and Slotine, J.E. 1990. The influence of as an Associate Editor for the IEEE Transaction on Robotics and
thruster dynamics on underwater vehicle behavior and their incor- Automation, the International Journal of Engineering Design and
poration into control system design. IEEE J. Ocean Engineering, Automation, and the International Journal of Intelligent Automa-
OE-15(3):167–178. tion & Soft Computing. Dr. Yuh has served as program chair of the
Yoerger, D.N., Bradley, A.M., and Walden, B.B. 1991. The au- 1993 International conference on Computer Applications in Indus-
tonomous benthic explorer. Unmanned Systems, 9(2):17–23. try and Engineering; organizing chair of the 1994 World Automation
Yuh, J. 1990a. Modeling and control of underwater robotic vehicles. Congress (WAC); program co-chair of the 1996 WAC; and co-chair
IEEE Trans. Sys., Man and Cyber., 20(6):1475–1483. of an NSF Workshop on “Future Research Directions in Underwater
Yuh, J. 1990b. A neural net controller for underwater robotic vehicles. Robotics,” August, 1994. He has been invited to serve as the or-
IEEE J. Oceanic Engineering, 15(3):161–166. ganizing chair of the 2001 IEEE/RSJ International Conference on
Yuh, J. 1994. Learning control for underwater robotic vehicles. IEEE Intelligent Robots and Systems and program co-chair of the 2001
Control System Magazine, 14(2):39–46. IEEE International Conference on Robotics and Automation. He
Yuh, J. (Ed.), 1995. Underwater Robotic Vehicles: Design and Con- founded and chairs the technical committee on Underwater Robotics
trol, TSI: Albuquerque. of the IEEE Robotics and Automation Society.