Proceedings of the Eleventh (2001) International Offshore and Polar Engineering Conference

Stavanger, Norway, June l 7-22, 2001

Copyright © 2001 by The International Society of Offshore and Polar Engineers
ISBN 1-880653-51-6 (Set); ISBN 1-880653-55-9 (VoL IV); ISSN 1098-6189 (SeO

The HUGIN 3000 Survey AUV

Karstein Vestgard and Roar Hansen

Kongsberg Simrad AS
Horten, Norway

Bjorn Jalving and Odd Arild Pedersen

F F I - Norwegian Defence Research Establishment
Kjeller, Norway

ABSTRACT understanding that underwater robotics and in particular AUVs will

play an important role in future survey and subsea engineering work.
The HUGIN 3000 AUV (Autonomous Underwater Vehicle) is
The AUV as a free swimming underwater survey sensor carrier has
intended for detailed seabed surveying in water depths ranging
several advantages compared to cable controlled remotely operated
from 100 m to 3000 m. HUGIN 3000 has an extensive payload suit.
vehicles (ROV) and deep towed systems:
The power source is a 40 kWh aluminum-oxygen fuel cell, sufficient
for 40-50 h of operation. A new high torque motor allows direct
• High payload sensor data quality: AUVs are ideal survey sensor
carriers for obtaining high quality survey d~ta due to low acoustic
drive of a single large three bladed propeller, providing a
self-noise, hydrodynamic stability and effective control of optimal
significant improvement in electrical to-hydrodynamic efficiency.
altitude above the seafloor.
Titanium, glass- and carbon fiber laminate material and high
performance syntactic foam are used in the vehicle design. The • High positioning accuracy: The operational freedom in relative
distributed subsystems are placed in separate pressure containers. position between the AUV and the survey vessel allows the survey
Emphasis has been put on flexibility in payload sensor vessel to be directly above the AUV. This will minimize the
configuration and easy maintenance and replacement of effects of acoustic ray-bending errors on the acoustic positioning
components with respect to mechanic, electronic and software. The accuracy. In addition minimal horizontal distance minimizes the
navigation system is an aided inertial navigation system integrating error contribution from survey vessel heading sensor installation
a number of surface and subsea navigation sensors allowing the misalignment and acoustic transducer installation misalignment.
vehicle to be run either in autonomous or surface support mode. For mapping of a smaller area or in shallow water, an AUV can
This paper includes test reports on endurance and navigation operate autonomously with good position accuracy provided by the
accuracy. aided inertial navigation system.
• High survey speed: An AUV relieved from an umbilical can
KEYWORDS: Autonomous underwater vehicle, seabed survey, vehicle typical operate with a speed of 4 knots independent of water depth,
design, battery endurance, navigation accuracy. providing a significant increase in survey effectiveness compared
to other survey tools.
INTRODUCTION • Curvature changes and line turns: Heading changes along a
survey line and turns from one survey line and onto the next can be
During the last few years autonomous underwater vehicle (AUV) accomplished within minutes opposed to hours for deep-towed
technology has evolved from concept demonstrators developed by systems.
research institutes and universities to commercial products. In the short- • Portability: An AUV system can be maintained in standard cargo
term perspective, the driving forces are the move for energy containers and can be air freighted throughout the world. One
exploitation towards deeper waters; naval applications and the Internet container can hold the vehicle and the launch-and-retrievai system.
driven need for more intercontinental underwater communication A second container can act as a combined maintenance shop and
cables. Other applications emerging for the future will be within battery service facility.
environmental research and monitoring and deepwater exploration.
Deepwater developments beyond the continental shelf will require THE HUGIN AUVS
the same level of survey data quality and intervention access as
established for shallow water. The potential for cost savings is two The HUGIN series of AUVs have primarily been designed for
folds. Firstly, there is a reduction in the survey cost itself and secondly detailed seabed surveying. The HUGIN I and H vehicles have been used
there is a considerable cost saving by avoiding over design of subsea in a wide range of seabed surveying operations down to 1200 m, both
installations due to lack of sufficient documentation of the field, for tests and commercial work (St~rkersen et al. 1998, Kristensen and
Vestg~rd et. aL 2000. In this scenario, there is an increasing Vestg~u'd, 1998, Vestg~'d et al., 1998). The Norwegian company NUI
operates HUGIN II on a commercial basis in the North Sea.

679
 This paper describes the HUGIN 3000 AUV. HUGIN 3000 is a provides also for an open internal structure for installation of instrument
further development towards deeper water, extended payload suit and containers, battery and transducers, see Fig. 1.
improved navigation and autonomy. The first HUGIN 3000 has been The pressure containers are made from titanium or glass. Vehicle
sold to C&C Technologies Inc and has now a busy schedule of survey control system is hosed in a large titanium sphere in the aft section.
work in the Gulf of Mexico (Chance et al., 2000, Vestg~d et al. 2001). Another large titanium sphere in the vehicle front section contains most
of the payload system. In addition there are additional dedicated
VEHICLE OPERATION containers for other payloads and control functions. The design
philosophy is based on a distributed and standardization approach
Launch and recovery allowing easy installation, maintenance and replacement of sub units.
A special transducer bay design philosophy has been adopted, allowing
The vehicle is deployed from the launch/recovery system located at payload reconfiguration either by replacing individual transducers or
the stem of the survey vessel. The vehicle executes the survey complete transducer bays.
according to the programmed mission plan. The mission plan can be The complete propulsion and rudder control system is installed in an
modified throughout the survey using the acoustic command link. When oil filled pressure compensated section.
initiated by the mission plan or commanded by the operator, the vehicle A new developed high torque synchronous motor allows direct drive
ascends to the surface using its propulsion system. Upon reaching the of a single large three bladed propeller, providing a significant
surface, the vehicle drops its nose cone and releases a retrieval line. improvement in electrical to-hydrodynamic efficiency.
This line is used to pull the vehicle onto the launch/recovery ramp The overall size and volume of HUGIN allows standard ISO
system projecting into the water from the stem of the survey vessel. transport containers to be used for storage, transportation and
UUV mode
maintenance. A dedicated launch and recovery system based on an
extendable stinger principle has been developed and verified in a large
In UUV mode HUGIN 3000 is operated close to the survey vessel. number of offshore operations with weather conditions up to sea state 5.
The acoustic links function as an "acoustic tether", which enables the
operator to supervise operation and control the vehicle in general.
Operator interaction could for instance be optimization of the survey
sensor parameters or reprogramming of the mission plan. For example, /~:ent and 8etteffeeegm
reoo~7eeee~ [ eotatamd
should hazards be located along a pipeline route, the mission can be / I 0ro0e~
altered for closer inspection. The mission plan is a set of waypoints and I ~ | ~ /
altitude, depth and propeller speed references as well as commands for
controlling mission modes, payload sensors etc.
In the survey vessel, the AUV position is tracked by combining
DGPS and SSBL data. Combined DGPS-SSBL position data is
transmitted to the AUV navigation system to bound the position drift.
During the survey, subsets of the payload and the vehicle data is
compressed and transmitted to the surface and displayed as real-time
plots and status windows, providing on-line quality control of the Inter~n~le "~ ~ ~ l e ¢ m ~ e r t at
collected data. The HUGIN Operator Station also includes a chart view,
displaying the "mission plan", survey vessel position and vehicle
position and heading.
AUV mode
In AUV mode HUGIN 3000 operates independently of the survey
vessel. In this mode the operator does not have real-time supervision or Fig. 1 HUGIN 3000 modular concept
control of the vehicle. The navigation system position drift is bounded
by DGPS fixes at the surface at regular intervals (for shallow water POWER SYSTEM
operations) or operation within a pre-calibrated transponder array The power source is the aluminum/oxygen fuel cell, extended to 40
(LBL) navigation system). kWh, sufficient for 40-50 hours of operation, depending on payload
For surveying of a limited area (for example a site survey), the AINS configuration and vehicle speed. Power is generated by a chemical
can maintain good position accuracy with only DVL aiding. A reaction between aluminum and oxygen in an alkaline electrolyte. The
prerequisite is that the dive phase is limited (when the DVL has no oxidant concentration is maintained during discharge by adding
bottom track) and that the vehicle mission geometry is optimized to hydrogen peroxide fluid proportional to the power load. The battery is
cancel out most of the systematic error growth. A favorable geometry is operated at ambient pressure with a performance independent of water
the "lawnmower" pattern. depth. No battery pressure container is needed, as with most other types
Basically, the difference between UUV mode and autonomous mode of power sources. The facilities for battery recharging and maintenance
is the presence of a survey vessel transmitting position updates. are installed in a 20 feet ISO transport container, including storage
tanks for clean water, hydrogen peroxide, fresh electrolyte and used
MECHANICAL DESIGN electrolyte.
HUGIN 3000 has a depth rating of 3000 n~ The weight is Recharging of the battery involves exchange of battery fluids and
approximately 1500 kg in air, the overall length is 5.3 m and the aluminum anodes:
A new highly efficient DC/DC converter has been developed
displacement is 2.5 m 3" Carbon fiber laminate material and high providing a significant gain in power efficiency house holding.
performance syntactic foam are used for the vehicle flame and hull. The
two main for-aft frames provide a stiff and torsion free hull structure,
ideal for precise sensor installation and alignment accuracy. The design

680
ELECTRONICS ARCHITECTURE NAVIGATION SYSTEM
The HUGIN 3000 electronics architecture is schematically presented In UUV mode the survey vessel tracks HUGIN 3000 with a super
in Fig. 2. The Control Processor (CP), Navigation Processor (NavP) and short baseline acoustic position system (SSBL). The survey vessel is
Payload Processor (PP) are connected together in a Vehicle Network equipped with the following navigation instrumentation:
implementing a TCP/IP network over a 10BaseT Ethernet. CP is the • SSBL
main processor. All sensors necessary for controlling the vehicle and • DGPS
the communication links are connected to CP. NavP runs the Aided • Heading/roll/pitch attitude sensor
Inertial Navigation (INS) system described in Section Navigation • Acoustic communication links
System. NavP provides the real-time navigation solution to CP for By combining DGPS with SSBL data compensated for attitude, an
guidance and control and to PP for sensor stabilization. PP interfaces AUV position estimate in global coordinates is obtained. This position
the payload sensors. PP is described in more detail in Section Payload estimate is sent to the AUV on the acoustic communication link.
System. HUGIN 3000 is equipped with an aided inertial navigation system
When HUGIN 3000 is on-deck, the HUGIN Operator Station and any (AINS). The inertial navigation system (INS) calculates position,
other computer can connect to vehicle using the Vehicle Service velocity and attitude of the vehicle using high frequency data from an
Network. This enables easy downloading of sensor data using FTP, inertial measurement unit (IMU). An IMU consists of three
flexible SW maintenance etc. accelerometers measuring specific force and three gyros measuring
angular rate. A Kalman filter utilizes in a mathematical optimal manner
a wide variety of navigation sensors for aiding the INS. The Kalman
filter is based on an error-state model and provides a much higher total
navigation performance than what is obtained by the independent
navigation sensors. The structure of the aided inertial navigation system
is shown in Fig. 3.

,~.~.~ I I~ , ok- ~ ,
m

lelld iiim

I me oOl~ h l m l m ~lllmlm E~Wmtm

J ~ee0~ m
Fig. 2 Schematic illustration of the HUGIN 3000 hardware
architecture. Abbreviations: POS - Payload Operator Station, NOS - Fig. 3 Kaiman filter structure for the HUGIN 3000 aided inertial
Navigation Operator Station, AEL - Acoustic Emergency Link, Cmd navigation system. DGPS-SSBL position and DVL aiding is illustrated.
link - Acoustic Command Link, RF link - Radio Link. The other aiding sensors are integrated similarly.

HUOIN 3000 is equipped with the following navigation sensors:

SOFTWARE DESIGN • Honeywell HG 1700 inertial measurement unit
• Oetans fiberoptic gyrocompass and motion sensor
The processors in HUGIN 3000 have very flexible and powerful
• Digiquartz pressure sensor
communication mechanisms. The design of the system is such that
• RDI WHN-300 Doppler velocity log (DVL)
customers may implement then" own Payload Processor, payload
sensors and payload operator stations. An option chosen by C&C • DGPS receiver
Technologies, capitalizing on theh" development works in the Orca • Kongsberg Simrad HiPAP SSBL transponder
Project. Alternatively, customers can easily add sensors and • Optional long baseline navigation transceiver (LBL)
functionality to a payload suit delivered by Kongsberg Simrad. Position updates are necessary to bound the position error drift in the
All inter-processor communication takes place via CORBA navigation system. When the AUV is operating in UUV mode, the
interfaces. For instance, all vehicle data from the sensors connected to AINS regularly receives combined DGPS - SSBL measurements from
CP are made available through a flexible publish/subscribe pattern. the survey vessel on the acoustic link. When the AUV is working in
Similar mechanisms exist for transmitting commands and for reporting autonomous mode, position updates can be provided by the AUV
events and errors, CORBA is an acronym for Common Object Request DGPS receiver or from a LBL or SSBL system.
Broker Architecture and is an international standard developed to fill
the need for interoperability between the vast number of hardware and PAYLOAD SYSTEM
software platforms available today. CORBA is fundamentally object
The HUGIN 3000 payload system consists of the Payload Processor,
oriented. Care has been taken to deal properly with the hard real-time the Payload Network, the Payload Power Distribution and the different
demands in the system. payload sensors.
The processors nm a modern a real-time operating system. The The payload sensor interface software is written as separate software
program systems have been developed using object oriented analysis, modules for each type of payload sensor, implementing all sensor
design and implementation. specific functions. The communication between a surface Payload

681
Operator Station (POS) and its subsea payload sensor is "generic"; i.e. with their error covariance matrices are saved for step two. In the
the data transfer is transparent and common for all sensor types. This second step, an optimal recursive-smoothing algorithm adjusts all the
ensures that new payload sensors can be added without the need to estimates, starting at the last estimate and running backwards in time,
change the basic system software. Signal interfacing of the payload improving the accuracy further.
sensors is either Ethemet 10/100MBit or serial line RS232 or RS422. Fig. 4 illustrates the data loggin and post-processing process. Wild
Payload sensor power on/off is controlled from the Control Processor. points in the data from the different payload sensors and navigation
Connectors for distribution of signal and power from the Payload sensors are removed and the data is filtered, automatically or manual.
Container to the individual payloads are standardized. The depth sensor measurements are compensated for datum, tidal water,
HUGIN 3000 can be equipped with a range of payload sensors and atmospheric pressure and water density. Finally the processed data
depending on application and user needs: is gridded, a DTM constructed and contour plots etc. produced.
• Multibeam echosounder (MBE)
• Side scan sonar (SSS) HUGIN 3000 FIELD EXPERIENCE
• Sub-bottom profiler (SBP)
HUGIN 3000 has been subjected to extensive sea trials. A customer
• C r D sensor
acceptance test program (CAT) has been concluded in the Gulf of
• Magnetometer Mexico. The CAT test program can be summarized as follows:
• Fishery research echosounder
• Others Water denth Main objective
As explained in Section Software System, C&C Technologies has 150 m System verification
tailor-made its own payload suit consisting of the following sensors: Calibration of payload sensors
• Kongsberg Simrad 2000 MBE 150 - 1200 m Endurance test
• Edgetech AUV SBP/SSS sonar Acoustic link tests
• Seabird SBE 19 CTD 1200 m Navigation accuracy
2200 m Deep- water operation
DATA LOGGING AND POST-PROCESSING Acoustic link tests
Navigation accuracy
During a mission, survey sensor and navigation sensor data is stored Altitude control in very steep terrain
on the 2 GB vehicle hard disk, while DGPS, SSBL, water level 3000 m Repetition of the 2200 m test
recorder, CTD, sound velocity and atmospheric pressure data is stored
on the survey vessel. All data is time-tagged to GPS UTC. Proper time In this paper we will concentrate on results on endurance and
tagging is ensured by synchronization of the AUV with GPS UTC prior navigation accuracy.
to a mission. A low drift clock in the AUV maintains an accurate time
reference for the whole mission. Endurance test
Already 19 - 21 July, 2000, HUGIN 3000 passed a 40 h endurance
test. The results are shown in Fig. 5. The system power is
approximately 1000 W when all the sensor systems are running. The
,.._____~ Sour.d~k=k'y~..,~moepher~ pr~re. battery system delivered a total of 40 kWh. The average speed was 4
knots. When the AUV turned, the rudder deflection caused increased
Po~-ptc~Nmd drag and thus reduced speed. This is clearly seen as small drops on the
Nlec;gllk~imllordellu graph. Since the power consumption of the propulsion system is cubic
IMU,Fluurl II~II01".
post- to the AUV speed, the AUV speed very much determines the endurance
p¢oc~8~
of the vehicle. HUGIN 3000 nominal performance parameters are 40
hours endurance at 4 knots with all payload sensors running.
M 6 E data. p r o c u r e d a t a

HUGIN 3000 endurance test

- - - . . . . . . . . . . . , ---, . . . . . . . .

Fig. 4 Illustration of the data logging, data flow and data processing ., , , , , , ,
u0~ ' ~ ' ~ 5 10 15 20 25 30 35 "40 45
required to produce a digital terrain model. Depending on the DTM ~" 1500 , , , , , , , , /
accuracy specification, the DTM can be based on either the real-time I I I I I I I I

t
navigation solution or the post-processed navigation solution. 1000
/

I I I I I I I I
500 r - - -I . . . . I-- T - -I . . . . I. . . . t- - - -I . . . .
I I I I I I I I
In the post-processing all data logged in the survey vessel is
u)
combined with the data logged in the AUV to produce the final digital 5 10 15 2O 25 3O 35 4O 45
terrain model (DTM) and to make an accurate georeference of the i r I T F 3 r "1
survey sensor data. The navigation solution stored by the AINS can be
used to produce the DTM. However, if maximum positioning accuracy
~ 4 - - -

is required, post-processing should be applied. The post-processing is a ~2 2 .... ,'-- - -1 .... :.... l .... : .... 1-- -I ......
o')
two-stage process. The first step is Kalman filtering using all AUV
5 10 15 20 25 30 35 40 45
navigation sensor data and all surface vessel stored DGPS-SSBL 11rne (hours)
position measurements. This is similar to the real-time AlNS that runs
in the AUV, except that the DGPS-SSBL position measurements are
Fig. 5 Results from the 40-h endurance test
available at higher rate with no time delays. All a posteriori estimates

682
Accuracy test We are now in the process of analyzing the navigation system
accuracy in sea trials in varying water depths in the Gulf of Mexico.
In detailed seabed mapping, it is the position accuracy of the The method used is mapping a known object, typically a well head,
produced maps, the digital terrain models (DTM), which is of primary multiple times with reciprocal lines in different directions. Figure 7
interest to the user. A complete DTM position error budget has been shows the "wagon wheel" pattern when mapping a well head in 1300 m
presented in Jalving et al. (2001). The DTM position accuracy is a water depth. The well head, hardly visible in this coarse figure, is in the
function of AUV navigation system accuracy, survey vessel attitude middle of the intersected lines.
accuracy, survey vessel systems installation accuracy, multibeam In Table 1, results from two accuracy tests in 1300 m are shown.
cchosounder position accuracy, AUV systems installation accuracy and "Ref difference" is the difference between the mean value of the DTM
AUV clock drift. In Fig. 6 the predicted AUV navigation system observation and the pre-surveyed position of the well head. The well
accuracy and DTM position accuracy for varying depths are shown. head was deployed from the BP drill platform Ocean America in 1999
AUV heading accuracy is in the order of 0.2 °. and was positioned by determining the surface position of the drill
string using DGPS. The drill string had a heavy well guide attached to
DIM and AUV n ~ i t i o ~ system accuracy the end. This made the drill string to be almost vertical but uncertainty
12 i
) - - DIM accuracy ] ~ /
of the effect of water current on the drill string is the largest source for
11 . . . . . I___
I
__
,
NlwiOatmsystenlaccuracy
, , -
h__L___/_- I
position uncertainty.
. . . . .
I
I. . . . . .
I
I. . . . .
I
.-I . . . . . A
I
. . . . . . . .
I
. .
The position accuracy can be improved by post-processing the dat_a
10
I I I I with a forward/backward smoother, compare the "KF Std dev" in Table
t I I I i
. . . . . I. . . . . . I . . . . . -t . . . . . .-t . . . . ~ . . . . . . 4. The smoother produces an optimal estimate based on all logged
I i I I I
measurements from both history and future. Furthermore, the full set of
~s . . . . .
I
I. . . . . .
I
I. . . . .
I
-I . . . . "T . . . . . T
I
. . . . . .
position measurements stored on the survey vessel is utilized.
I I I I l
The Kalman filter s t a n d a r d deviations are well in accordance with the
..... ,. . . . . . ). . . . . . ,. . . . "~. . . . . T~-" ....
1 standard deviation of the well head observations (population standard
. . . . . I. . . . . J . . . . J . . . . . A - ~ - - -- J- . . . . . .
b ~ I I I ~d S I deviation), though the March 2001 post-processed r e s u l t s a r e a little
. . . . . !. . . . . high. Given the uncertainty of the verticality of the drill string,the "Ref
difference" results are consistent.
I ~q" I " I I
Mission No. of Ref difference Pop.Std dev ! KF Std dev
.... C=7---~ ..... -~ . . . . . "T . . . . . T ...... meas. North East Nocth East North East
March 2001 6 t.2m 6.1m 2Am 4.1m 4.4m 4.5m
0 500 1000 1500 2000 2500 3000 real-lime
A U V depth (m)
March 2001 6 0.5m 3.1m 3.7m 4.2m l.Tm 1.7m
post-processed
Fig. 6 AUV navigation system position accuracy (dotted line) and October 2OO0 II 2.7m 4.0m 1.2m 1.7m l.Tm 1.7m
resulting DTM position accuracy (solid line) for varying water depths. post-processed
Table 1 Position accuracy
i
CONCLUSIONS

The HUGIN 3000 AUV has been presented. The vehicle is optimized
' "~'~ ' ~= "~i '. for detailed seabed surveying in water depths down to 3000 n~ A key
design goal as been to have a flexible payload system providing easy
integration of new payload sensors.
HUGIN 3000 has demonstrated 40 hours endurance at 4 k n o t s w i t h
all payload sensors running. It has also demonstrated very accurate
navigation. HUGIN 3000 is currently (September - October 2000)
subjected to extensive testing in the Gulf of Mexico in order to prove
robust, viable and economical operation and very high da_t~_quality.
.,%9 ~. ', ,2
When the test program is completed, C&C Technologies will bring
HUGIN 3000 into active survey service.
HUGIN 3000 is based on the well-proven HUGIN I and II vehicles.
HUGIN II has provided commercial survey services in the North Sea
for 3 years.

REFERENCES
Jalving B and Gade K (1998). "Positioning Accuracy for the HUGIN
Seabed Surveying Untethered Underwater Vehicle", eroc Oceans 98,
Nice, France.
Jalving B, Vestg~l K and Storkersen N (2001). "Technology and
Fig. 7 Lines with multibeam echosounder data collected for
Applications of Autonomous Underwater Vehicles". To be published
navigation system accuracy verification in 1300 m water depth. A well
by Gordon&Breach in their series Ocean Science and Technology.
bead with known position in the middle of the intersected lines is
mapped multiple times.

683
Kristensen J and Vestg~h-'d K (1998). "HUGIN - An Untethered Vestg~rd, K, Klepaker, RA and Sterkersen N (1998). "High Resolution
Underwater Vehicle for Seabed Surveying," Proc Oceans 98, Nice, Mapping with the HUGIN UUV," Proc UUVS98, Southampton UK.
France. Vestg~rd K., Hansen R., Jalving B., Pedersen O. A. (2001). "The HUGIN 3000
Chance TC, Kleiner AS and Northcutt JG (2000). "A High-Resolution survey AUV - Design and Field Results," Proc Underwater Intervention
Survey AUV," Sea Technology, voL 41, no 12, December 2000. 2001, Tampa, Florida.
StCrkersen, N, Kristensen, J, lndreeide, A, Seim, J and Glancy, T
(1998). "HUGIN - UUV for seabed surveying", Sea Technology, vol.
39 no. 2, February 1998.

684