doi:10.3723/ut.28.

091 International Journal of the Society for Underwater Technology, Vol 28, No 3, pp 91–98, 2009

Technical Paper
Autosub6000: its first deepwater trials and science
missions
S McPhail, M Furlong, V Huvenne, J Perrett, M Pebody and P Stevenson
National Oceanography Centre, Southampton, UK

Abstract Ever since its conception and first trials in 1996,

In September 2007 on RRS Discovery, the we have emphasised the ‘Auto’ in the Autosub
Autosub6000 autonomous underwater vehicle (AUV) programme name. Great importance has always
completed its first deepwater engineering trials and, been placed upon freeing up the support ship
fitted with a multibeam bathymetric mapping sonar, to carry out other operations, as well as having
carried out its first science missions less than a year routinely operated the vehicle ‘over the horizon’,
later as part of a geology and geophysics science beyond communication or acoustic tracking range.
cruise onboard the RRS James Cook. This paper For example, during the Arctic and Antarctic
describes how the issues of energy storage, navigation under-ice missions, as part of the UK Natural
and buoyancy control were tackled that specifically Environmental Research Council (NERC) funded
affect a deep-diving AUV, capable of operating with Autosub Under Ice programme, we operated the
true autonomy independently of the mother ship. AUV for a 24-hour missions under sea ice in north
east Greenland (Wadhams et al., 2006) and a
Keywords: autonomous underwater vehicle (AUV), 30km run under an Antarctic ice shelf (Nicholls
deep-diving, lithium polymer battery, underwater et al., 2006). In both of these missions, the AUV
robotics, navigation, ultra short baseline (USBL) operated well beyond any communications range
or hope of rescue if anything went wrong. For these
under ice and other science missions, we developed
1. Introduction the control, collision avoidance and navigation
Worldwide there are several scientific survey au- systems for the AUV, and gained experience
tonomous underwater vehicles (AUVs) that are of operating an AUV in extreme environments
either operational or in advanced stages of devel- (Pebody, 2008).
opment. For example, the Woods Hole Oceano- We have continued with this philosophy of true
graphic Institution (WHOI) Autonomous Benthic autonomy for the Autosub6000 AUV. For example,
Explorer (ABE) has been carrying out pioneering while the AUV is carrying out a high-resolution
work in high-resolution mapping of mid-ocean bathymetric survey, we may wish to use the mother
ridge environments for over ten years (German ship for coring, perhaps guided by data previous
et al., 2005). This will soon be supplemented by the recorded by the AUV.
Sentry AUV. There are several challenges to achieving the
Monterey Bay Aquarium Research Institute required performance. For useful autonomy, the
(MBARI), with the Seafloor Mapping AUV energy storage technology needs to be considered
(Henthorn et al., 2006); Altium, with Bluefin-21 and developed. Another potential problem is the
(Wilson and Bales, 2006); and the University of buoyancy variation of the AUV as it descends.
Tokyo, with R2D4 (Ura et al., 2007), all manufac- This could cause an increase in the hydrodynamic
ture AUVs with deep-sea science capabilities. Kongs- drag, and hence decrease the useful range of
berg Maritime, with the HUGIN series of AUVs the AUV.
and the recently introduced HUGIN 4500m, has Unassisted navigation of a deep-diving AUV is
also been particularly active in the deepwater com- another challenge. An AUV fitted with currently
mercial survey market. There has been significant available multibeam sonars is capable of bathymet-
take-up by academic and research organisations of ric surveying at a resolution of 1–5m (depending on
both the Kongsberg, Hydroid REMUS 6000, and the AUV flying altitude). The value of this data can
the International Submarine Engineering (ISE) be maximised if we can position the vehicle with
Explorer AUV classes of AUVs (for example by corresponding accuracy, particularly if the AUV is
IFREMER with the AsterX AUV) (Opderbecke and being used to identify interesting seabed features
Laframboise, 2007). for later, more detailed investigation by itself or

91
 McPhail et al. Autosub6000: its first deepwater trials and science missions

2.2. Batteries
We achieved the usually mutually exclusive
characteristics of deep-diving capability and good
range by developing a pressure-tolerant lithium
polymer battery technology, thereby eliminating
the need for expensive and bulky pressure-resistant
housings. This approach was first pioneered for use
in AUVs by Bluefin Robotics (Wilson and Bales,
2006). Using the deep-pressure facilities at National
Oceanography Centre, Southampton (NOCS), up
to 68MPa, extensive pressure cycle testing of the
batteries has been carried out.
Within each battery box are 405 Kokam lithium
polymer cells, storing a total of 16.2M joule
(4.5kW hr) of energy at a nominal 57 volts at up
to 15 Amperes discharge rate. The batteries are
protected against over charge, over discharge and
Fig 1: Layout of Autosub6000: the forward and aft over current through failsafe, redundant circuitry.
sections are free flooded, with an aluminium space Each battery is monitored via an I2 C bus for
frame and glass fibre reinforced plastic (GFRP) currents, voltages, temperature, leaks and pressure-
fairings; the centre cylindrical section provides compensating oil level. Charge monitoring and
most of the buoyancy with syntactic foam and also control is integrated into the battery, so charging is
contains the pressure tolerant batteries relatively simple, using a commercial 1.2kW, 60-volt
(Glassman) power supply for each battery.
Autosub6000 is currently fitted with four pres-
another vehicle (for example a remotely operated sure balanced batteries, which with a multibeam
vehicle, or ROV). sensor payload (120W power) give an autonomy
Hence, there are three issues for an AUV of 36 hours and a range of 180km. There is
platform which are specific to the deep-diving and capacity in the vehicle to increase this to 12
true autonomy: batteries, with a proportional increase in range
and endurance (longer ranges are possible at
• Energy storage at high ambient pressures slower operating speeds). Autosub is propelled by
• Accurate autonomous positioning of the vehicle a direct-drive brushless direct current (DC) motor
throughout its mission and two-bladed propeller.
• Buoyancy change due to compressibility effects.
2.3. Navigation, communications and tracking
For dead-reckoned navigation, the AUV uses a
2. The Autosub6000 vehicle 300kHz Teledyne RDI Workhorse Acoustic Doppler
Current Profiler (ADCP) to measure its velocity
2.1. Mechanical design relative to the seabed when within the 220m seabed
Autosub6000 is 5.5m long, with a 2.8m3 dis- tracking range, and an Ixsea Oceano PHINS fibre
placement and a 6000m depth rating. The main optic gyro (FOG) based Inertial Navigation System
difference between it and its predecessor, the (INS). These are housed together in a titanium
1600m-depth-rated Autosub3 described elsewhere pressure case. The experience that NOCS and
(Stevenson et al., 2003), is the centre section another operator (Bjerrum, 2002) has had with
(Fig 1). Whereas the Autosub3 uses seven 3m-long similar navigation systems indicates that, when
carbon fibre pressure cases containing up to 600kg calibrated, the drift rate of this system is 5 miles per
of primary manganese alkaline cells, Autosub6000 hour or less. Two problems remain for the accurate
uses a completely different approach. The centre navigation for a deep-diving AUV:
section contains no pressure cases and is essentially
• Initial position: positioning of the AUV after its
a cylinder made from sections of syntactic Emerson
initial descent to the seafloor
and Cuming foam (EL34 – density 580kg m−3 ), with
• Drift: controlling and correcting the navigation
slots cut out for up to 12 batteries. The navigation
drift error for long missions.
and control systems are contained within titanium
pressure cases in the free-flooding tail section. The Both of these problems could be tackled by the
1.5m-long free-flooding nose section is free for the use of a seabed moored acoustic transponder net-
science payload. work, however, this approach is relatively expensive

92
 Vol 28, No 3, 2009

in ship time. It has been reported (Griffiths, uncontrolled navigation error growth during area
2007) that to deploy, position and recover five survey type missions.
transponders at 4800m water depth for acoustic We use a combined USBL and bidirectional
navigation of the Isis ROV within a square box of acoustic messaging system, the LinkQuest Track-
only 1km side, it took a total of 27 hours of ship Link 10000 for real-time tracking of the AUV
time. The AUV could survey 40km2 in this time with from the mother ship and telemetry for health
high-resolution multibeam bathymetry. monitoring and AUV command functions.
Use of an ultra short baseline (USBL) system
from the mother ship is another option, and it 2.4. Buoyancy change
has been used successfully for tracking deepwater If the materials used in the construction of an
survey AUVs, but it is less attractive for our needs AUV do not compress at a similar rate to that
as it commits the ship to continuously monitoring of seawater itself (2.8% over 6000m), then the
the AUV. buoyancy of the vehicle will change substantially as
Instead we use a technique of range-only it descends. The largest solid item on the vehicle
navigation. A set of ranges from the ship to an is the syntactic foam used for buoyancy. From
acoustic transponder on the AUV – when combined manufacturers data and laboratory tests, we were
with the AUVs own dead-reckoned navigation and able to get an approximate estimate of the bulk
the ships navigation – provides the information compressive and thermal moduli of this material, as
to accurately position the vehicle after its descent. well as account for the other materials used in the
This approach avoids the main problem with construction of the vehicle.
USBL based systems – their need for extremely An absolute worst-case buoyancy change estimate
high pointing and attitude reference accuracy, was made prior to the trials based on manufac-
necessitating a costly and very precisely calibrated turers’ maximum compression and quoted water
system (Jalving et al., 2002). absorption (the filling of micro cracks and voids
By combining the AUV’s self-navigation as it with water with repeated pressure cycles), showing
executed the 1km side, square course around the we ought to build in some contingency in case there
ship at 4556m depth and an altitude of 120m was a reduction in buoyancy. So, for the first trials
above the seafloor, with the ships navigation and we conservatively ballasted the vehicle with a surface
the ranges from the ship to the transponder on buoyancy of 20kg, rather than the typically used
the AUV, we were able to calculate the amount 10kg for Autosub3, which is a larger vehicle. We also
(1X , 1Y ) by which the AUV navigation had drifted installed two independent emergency weight drop
during its descent to the seabed. systems, each able to increase the vehicle buoyancy
The task is a least mean squares minimisation by 10kg in an emergency situation (e.g. when over
problem based upon Pythagoras’ equation (X 2 + depth is detected).
Y 2 + Z 2 = R 2 ). The problem is to find the values This conservative level of surface buoyancy,
of the AUV navigation error: 1X , 1Y , 1Z , which together with the expected increase in buoyancy
minimises the mean squared error (ξ ) over the N with depth, could create a problem. The vehicle
measurements of ranges (R ), AUV (xAUV, etc.) and might have difficulty controlling its depth, or might
ships (xSHIP, etc.) positions (Equation 1). suffer significantly increased drag due to the need
2 to produce large hydrodynamic downward force.
(xSHIPj − xAUVj + 1X 2 ) 

N

X + (ySHIPj − yAUVj + 1Y )2 

  The solution was to install small wings on the body
ξ= (1) set slightly pitch down. These help by producing
 + (zSHIPj − zAUVj + 1Z )2 
j =1  down force more efficiently than the vehicle body
−(Rj )2
 

can alone.
There remains the challenge of controlling the Table 1 illustrates the best estimate of pre-
drift of the AUV positioning system during the dicted change in buoyancy for the trials condi-
mission. For missions where an area is to be tions (20 ◦ C laboratory temperature at which the
surveyed by the AUV, there is much interest vehicle was ballasted, falling to 2 ◦ C and 4700m
in approaches such as simultaneous localisation operating depth). The most significant element
and mapping (SLAM) (Kinsey et al., 2006). We is the syntactic foam compression, accounting for
are planning to use a similar approach with some 80% of the change; data provided by the
Autosub6000. Using data from the Kongsberg manufacturers describes the bulk modulus as 1–2%
Simrad EM 2000 multibeam bathymetric mapping volume change at maximum rated depth of 7315m.
system that is fitted to Autosub6000, we are A mean volume change of 1.5%, pro rata with
developing algorithms based on correlation, or actual pressure was used for the best estimate, but
terrain contour mapping (TERCOM) approaches, the rather nebulous bulk modulus quoted suggests
with the aim of substantially eliminating the the buoyancy value could be ±3L either side of

93
 McPhail et al. Autosub6000: its first deepwater trials and science missions

Table 1: The predicted change in displacement of the AUV’s major components: note that almost 80%
of the predicted changes in displacement are due to the thermal contraction plus compression of the
syntactic foam, however, there are large uncertainties associated with the syntactic foam compressive
and thermal moduli
Item Volume Contraction due to Compression at Contraction of
(litre) temp change from 47MPa (litre) cells due to 50%
20 ◦ C to 2 ◦ C (litre) discharge (litre)
Syntactic foam EL36-A 919 1.28 8.86 NA
Lithium polymer cells 40.1 0.48 0.33 0.53
Carnation pressure 33.5 0.40 1.64 NA
balancing oil
Titanium logger case 9.3 0.005 0.11 NA
Titanium power case 12.1 0.006 0.15 NA
Titanium ADCP/PHINS case 20.7 0.010 0.25 NA
Totals (litre) — 2.18 11.34 0.53
Total reduction in displacement: 14.1 Litre

Table 2: Summary of predicted AUV buoyancy change due to pressure and temperature effects as the
vehicle descends from the surface to 4700m depth: the overall change in buoyancy is due to the
increase in seawater density (which increases the vehicle buoyancy), added to the decrease of the
vehicle displacement (predicted in Table 1); the AUV buoyancy was predicted to increase from 20kg
at the surface to 36kg at 4700m depth
Sea water density Total vehicle Vehicle buoyancy, resulting from
(kg/m3 ) displacement increase in water density and loss
(litre) of displacement (kg)
At the surface 1.023 1212 20 (set in the laboratory)
At 4700m depth 1.049 1198 36 (i.e. increase of 16kg)

the estimate. Furthermore, it could be more or tolerance on the properties is large and we have
less depending on the non-Hookean nature of the assumed that there is no water absorption occurring
material. A technical note suggests water absorption at these intermediate depths. As new missions take
occurs as a logarithmic response, levelling off the vehicle deeper, we will need to exercise caution
at about 2.5% increase in foam weight, but is and monitor this characteristic.
not completely understood (Cuming Corporation,
2000). This phenomenon of losing buoyancy with 3. Results
successive dives was not observed in the trials. 3.1. Autosub6000 sea trials in September 2007
The lithium polymer cells are described as Autosub6000’s first test cruise and first time in
having ‘20% proprietary electrolyte’, and it was water was in September 2007, onboard the RRS
assumed they would behave in a similar way to the Discovery. We headed for a flat part of the deep
pressure compensation oil. In addition, the cells are abyssal Atlantic, with a water depth of 4680m near
known to expand during charging, which has been 47◦ N, 11◦ W, 250 miles from Falmouth, UK, the
approximately measured during charge/discharge embarkation port.
cycles, and a 50% state of discharge was assumed for We needed to plan the first deep mission with
the buoyancy estimate. Table 2 indicates that these some care for its first dive to 4556m. We could
assumptions are not particularly important given not assume that any system (such as the acoustic
the dominant factor of the foam compression. It is communications system, which had never been
seen that the compression of the titanium housings tested on the vehicle) would work as designed.
is negligible. Shortly after daybreak on 22 September 2008,
The actual buoyancy changes were measured we launched Autosub6000. Following system checks
by comparing the rates of ascent at depth and via the radio link, we sent the command to start
near the surface. Assuming the buoyancy to be the mission. The vehicle dived and spiralled down
proportional to the square of the ascent rate and to 1000m depth and then began circling beneath
knowing these two rates and the surface buoyancy, the ship. We were relieved that the acoustic com-
the ‘at depth’ buoyancy was measured to be 31kg, munications system worked faultlessly, telemetering
an increase of 11kg over the surface buoyancy. This engineering data as the AUV descended.
compares reasonably well with the best estimate We were particularly interested in monitoring
of 16kg, although it needs to be stressed that the the AUV pitch, forward speed and stern plane

94
 Vol 28, No 3, 2009

RADIAL HORIZONTAL RESIDUAL (m)

3.5
–1000
0 3
1000
Depth [m]

2.5
2000
2
3000
4000 1.5
5000
1
1000
1000 0.5
0 500
0
–500
–1000 –1000 0
X position (North) [m] Y position (East) [m] 0 20 40 60 80 100 120 140
SAMPLE NUMBER

Fig 2: The 3D navigation plot for the first deep

Autosub6000 mission: the AUV spiralled to 4556m Fig 3: The absolute values of horizontal range
depth, then executed a 1km side box while data residuals for all of the ranges measured between
was gathered for the range-only positioning the ship and the AUV as the latter executed a 1km
side box around the ship at 4556m depth

angles. When flying at constant depth, these

variables are related to the vehicle buoyancy
(Equation 1) and hence can be used to monitor
the change of buoyancy as the vehicle descends with
Equation 2:
1
ρU 2 (Clbody φ + Clsplane δ)
B= (2)
2
where B = AUV buoyancy, ρ = density of water,
U = AUV speed through the water, Clbody = lift
slope of AUV body, Clsplane = lift slope of stern-
plane, φ = pitch angle and δ = sternplane angle.
Having collected enough data to be satisfied
that the vehicle was operating correctly and safely,
we sent an acoustic command for the vehicle
to continue its descent to 2500m. If the AUV Fig 4: Autosub6000 being recovered into its
had not received this ‘continue’ command within launch and recovery platform following its first deep
a period of 1 hour since reaching 1000m, it survey mission; note the EM2000 multibeam sonar
would have automatically aborted its mission, mounted on the underside of the front section
dropping its 20kg ballast weights, and surfaced.
This mode of behaviour is inherently failsafe.
If the vehicle had gone out of control and we this case was: east, −368m; north, 848m. One
couldn’t establish acoustic communications, the approach to estimating the robustness and accuracy
vehicle would abort the mission and start to ascend of this method is to evaluate the solver residuals,
before it could collide with the seabed. This failsafe calculated for the horizontal radial error, for each
approach, together with acoustic communications of the 130 range measurements (Fig 3). The low
feedback, made it possible for us to safely dive scatter and maximum values of these residuals is a
the vehicle to 4556m on its first dive beyond 20m. strong indicator of the robustness of this method.
Without such an approach, on only the AUV’s The solver uses all of the data to produce a
second ever dive, such a mission would have been single position estimate. Very similar results are be
arguably reckless. obtained by using only 10 ranges.
One of the objectives of the trials was to test After its 7-hour first dive, the vehicle surfaced
the performance of the TrackLink 10000 telemetry and was recovered onto the launch and recovery
and USBL system, as well as our procedures and system (Fig 4).
algorithms for navigating the AUV using range-only On subsequent dives we programmed the AUV
measurements. Fig 2 is a plot of the AUV navigation to head south for 5km, while the ship remained
(uncorrected) for the first deep mission. more of less stationary. The USBL and telemetry
The amount by which the AUV navigation system tracked and continued to reliably send and
had drifted during its descent to the seabed in receive telemetry messages up to a horizontal range

95
 McPhail et al. Autosub6000: its first deepwater trials and science missions

–1000
0
1000

Depth [m]
2000
3000
4000
5000
2000
1000 2000
0 1000
–1000 0
–1000
X position (North) [m] –2000 –2000 Y position (East) [m]

Fig 6: The 3D navigation trace for the first deep

Autosub mission, starting at the right dot, ending
on the left dot; note the spiral for descent and
ascent, the 1km boxes run at the start and the end
of the mission, and the actual ‘lawnmower’ survey
runs at 100m altitude

Fig 5: A 10m piston core being deployed with important that the positioning of the Autosub was
Autosub6000 is 35km away carrying out a accurate, both in real time (so that the correct
bathymetric survey of the seafloor area would be surveyed), and after post-mission
processing (so that the navigation corrected data
could be used to guide coring operations).
of 7km, confirming the suitability of the system for
Each of the missions was based on the same
AUV operations to its depth limit of 6km.
template, allowing us to develop standard mission
planning tools, which very much simplified (and
3.2. RRS James Cook Cruise 027
hence made more reliable) the mission planning
Following the successful engineering trails, it was
process. The only parameters which needed setting
arranged for Autosub6000 to be taken onboard for
for each mission were:
RRS James Cook Cruise 27. Lead by Dr Russell Wynn
of the NOCS, the objective of the cruise was to • Position centre of the survey area
investigate potential threats to coastal communities • Width and length of the survey box
along the Western European margin from tsunamis • Line spacing (200m initially, increased to 300m
generated by giant landslides and earthquakes. after the first mission)
For this cruise, Autosub6000 was fitted with • Survey altitude (always 100m)
a Kongsberg Simrad EM 2000 multibeam bathy- • Minimum water depth in the survey area (by
metric sonar. Mounted on the underside of the taking this into account, we could increase the
nose section of Autosub behind a hydrodynamic safety of the mission by being able to set a
fairing, the system isonifies the seabed with a 120◦ depth for over-depth triggered mission abort
swath of 111 beams. Flying at 100 altitude, the total at less than the minimum water depth in the
swath width is 346m. For the first mission, we con- survey area).
servatively set the line spacing of the swaths at 200m. Following the dive, every mission included a stop
Results from the first mission showed, however, that and circling at 1000m depth, and then again at
the full swath width was measured 100% of the time, 100m less than the water depth while we checked
and the navigation drift was negligible compared the AUV engineering systems via the acoustic
to the line spacing. For subsequent missions, we telemetry link. Only when we were content with the
set the line spacing at 300m. With a vehicle speed AUV performance did we send the command for it
of 5km per hour, this produced an area coverage to continue. A risk analysis indicated that by doing
rate of 1.5km2 per hour. The bathymetric data was this we could very significantly reduce the risk to the
displayed at a resolution of 2m. AUV due to system failure. The AUV descent rate
The intention was that Autosub6000 would carry was 0.9ms−1 , taking 1.5 hours to reach the typical
out a high resolution multibeam survey of an area, operating depth of 4600m.
and once recovered, the data would be used to Following the descent, the vehicle executed a
guide locations for seabed sampling using piston 1km box around the ship’s position, while we gath-
cores (Fig 5). For this to be successful, it was ered data for a range-only navigation fix (Fig 6).

96
 Vol 28, No 3, 2009

–4238.56 meters

–4280.37 meters

Fig 7: First bathymetric survey image produced by the Autosub6000 system – a 4 × 4km survey of part of
a submarine canyon, north of the Canary Islands: the maximum depth (bottom) is 4280m, the minimum
depth (top) is 4238m and the scour marks, which are hundreds of metres across, were formed by
turbulent submarine flows that ripped up huge volumes of seafloor sediment. Within a few hours of the
recovery of the AUV, this image was used in helping to plan coring operations. Image reproduced by
permission of Dr Russell Wynn (NOC), principle scientist of RRS James Cook cruise 027

The range-only navigation was implemented in rate was 1.5ms−1 , taking 50 minutes to ascend
‘near real time’, with the AUV self-navigation from 4500m.
telemetered back to the ship for each range we With the USBL system still working with the AUV
obtained to the AUV. Once the AUV navigation surfaced, it was generally easy to locate the AUV.
error had been calculated and quality checked (this Flashing lights and two UHF Argos transmitters
took typically 5 minutes), the offset to the AUV on the vehicle were available, if necessary, to aid
navigation was sent via the acoustic downlink. We recovery. The engineering and multibeam data
then sent the acoustic command for the vehicle could then be downloaded from the vehicle via a
to begin its survey. Typically, it took 3 hours from WiFi link with an effective range of 500m.
launch to start of the survey. With the survey For this cruise we experimented with compressed-
underway, we were able to recover the tow fish used air-powered grappling-line launcher (manufactured
for communicating with the AUV, and the ship was by ResQmax) to recover the lines used for lifting
free to carry out other operations, usually to transit Autosub into its launch and recovery. After some
to another area and take piston cores. practice we found this to be a useful system, allowing
The AUV surveys took about 18 hours to execute. us to reliably recover the AUV lines when the vehicle
At the end of the survey, the AUV returned to was floating significantly further from the ship
the centre of the box and waited while circling at than had been possible when using the traditional
100m altitude (in case of unexpected delays with method of a crew member throwing the grappling
the other ships operations), waiting for our acoustic line. This is an important improvement, as damage
‘continue’ command. It then carried out another to the AUV through collision with the mother ship
1km box around the ships position, for a post-survey is a serious risk.
range-only navigation fix. With the AUV safely recovered onto the ship, the
After sending yet another acoustic command to next anxious wait was for the data to be processed.
the AUV, it began its ascent. For each mission, Had the system worked? After an hour we had
we set the AUV to ascend at alternately high our answer, and Autosub6000’s first bathymetric
power and very low power, and at a steep angle images (Fig 7).
(60 degrees). The ascent rate at the low power
is sensitive to the vehicle buoyancy, giving an
accurate indication of any changes in buoyancy 4. Conclusions and future work
as the AUV ascends and, more importantly, The trials in September 2007 and the science mis-
between missions. We did not detect any changes sions a year later were a success. The Autosub6000
in buoyancy during the four missions of the AUV controlled, navigated and communicated as
cruise – a reassuring result. The average ascent designed, and the batteries worked without any

97
 McPhail et al. Autosub6000: its first deepwater trials and science missions

problems. The multibeam data was of good quality. Henthorn R, Caress D, Thomas H, McEwen R, Kirkwood W,
The range-only navigation algorithms were tried Paull CK and Keaten R. (2006). High-resolution multi-
and tested in both post-processing mode and beam and subbottom surveys of submarine canyons
during mission time. The results looked promising and gas seeps using the MBARI mapping AUV. In:
as one solution to the initial positioning problem Proceedings IEEE Oceans ’06, 18–21 September, Boston,
MA, 1–6.
and as a way of estimating navigation drift during
Jalving B, Vestgard K and Storkersen N. (2002). Detailed
the mission. Seabed Surveys with AUVs. In: Griffiths G. (ed.)
In the near future, we plan to further develop Technology and Applications of Autonomous Underwater
and test our navigation algorithms, with the aim of Vehicles. London: Overseas Publishing Associates OPA
implementing TERCOM-like navigation using the (UK) Ltd, 179–201.
AUV multibeam data. Through this method, we Kinsey JC, Eustice MR and Whitcomb LL. (2006). A Survey
hope to be able to routinely navigate the AUV to of Underwater Vehicle Navigation: Recent Advances and
an accuracy as good as standard GPS, without the New Challenges. In: Proceedings of the 7th Conference on
need for the deployment of acoustic transponder Maneuvering and Control of Marine Craft (MCMC’2006).
network. We also plan to develop the collision- Lisbon: IFAC.
Nicholls KW, Abrahamsen EP, Buck JJH, Dodd PA,
avoidance capability of the AUV. This, together
Goldblatt C, Griffiths G, Heywood KJ, Hughes NE,
with improvements of the collision detection and
Kaletzky A, Lane-Serff GF, McPhail SD, Millard NW,
crash tolerance of the vehicle, will give the AUV the Oliver KIC, Perrett J, Price MR, Pudsey CJ, Saw K, Stans-
capability to operate more safely close to the seabed field K, Stott MJ, Wadhams P, Webb AT and Wilkinson JP.
(e.g. to take colour photography) and in areas with (2006). Measurements beneath an Antarctic ice shelf
rugged seafloor terrain. using an autonomous underwater vehicle. Geophysical
Research Letters 33(8): L08612.
Acknowledgements Opderbecke J and Laframboise JM. (2007). AUVs for
Autosub6000 is being developed at the National oceanographic science at IFREMER, project progress and
operational feedback, In: Proceedings Oceans ’07, 2–4
Oceanography Centre, Southampton, with funding
October, Vancouver BC, Canada, 1–5.
from the UK Natural Environment Research Coun-
Pebody M. (2008). Autonomous underwater vehicle colli-
cil. We would like to thank the Master and crew sion avoidance for under-ice exploration. In: Proceedings
of RRS Discovery Cruise 323 and RRS James Cook IMechE, Part M: J. Engineering for the Maritime Environment
Cruise 027. 222: 53–66.
Stevenson P, Millard NW, McPhail SD, Riggs J, White D,
References Pebody M, Perrett JR and Webb AT. (2003). Engineering
Bjerrum A. (2002). AUVs for Survey in Complex Environ- an autonomous underwater vehicle for under ice oper-
ments. In: Griffiths G. (ed.) Technology and Applications ations. In: Proceedings of the 22nd International Con-
of Autonomous Underwater Vehicles. London: Overseas
Publishing Associates OPA (UK) Ltd, 203–215. ference on Offshore Mechanics and Arctic Engineering
Cuming Corporation. (2000). Technical Note 100-2. Pre- (OMAE 2003), Cancun, Mexico: OMAE2003-37038.
dicting Hydrostatic Performance of Syntactic Foam. Ura T, Tamaki K, Asada A, Okamura K, Nagahashi K,
Accessed on 27 April 2009 at http://www.cumingcorp. Sakamaki T, Gamo T, Okino K, Obara T, Nakane K,
com/pdf/cumingtechnicalnote100-2.pdf. Obata T, Ooyabu Y, Yamaoka N, Orihashi Y, Jun Han
German CR, Connelly DP, Prien RD, Yoerger DR, Jakuba M, Koyama H and Sugimatsu H. (2007). Dives of AUV
Bradley A, Shank T, Nakamura K-I, Langmuir C and
‘‘r2D4’’ to Rift Valley of Central Indian Mid-Ocean Ridge
Parsons LM. (2005). New techniques for hydrothermal
plume investigation by AUV. In: European Geosciences System. In: Oceans 2007 – Europe, 18–21 June: 1–6.
Union (ed). Geophysical Research Abstracts 7: EGU05-A- Wadhams P, Wilkinson JP and McPhail SD. (2006). A new
04361. view of the underside of Arctic sea ice. Geophysical Research
Griffiths G. (2007). RRS James Cook Cruise JC009T. Letters 33: L04501.
Trials of the Isis Remotely Operated Vehicle. National Wilson RA and Bales JW. (2006). Development and
Oceanography Centre Southampton Cruise Report, Experience of a Practical, Pressure-Tolerant, Lithium
No. 18. Southampton, UK: National Oceanography Battery for Underwater Use. In: Proceedings IEEE
Centre Southampton, 42pp. Oceans ’06, 18–21 September, Boston, MA, 1–5.

98