the untethered remotely operated vehicle picasso-1 and its

P A P E R

The Untethered Remotely Operated VehiclePICASSO-1 and Its Deployment FromChartered Dive Vessels for Deep Sea SurveysOff Okinawa, Japan, and Osprey Reef,Coral Sea, Australia
A U T H O R SDhugal J. LindsayHiroshi YoshidaJapan Agency for Marine-EarthScience and Technology,Yokosuka, Japan
Takayuki UemuraKowa Corporation, Osaka, Japan

Hiroyuki YamamotoShojiro IshibashiJapan Agency for Marine-EarthScience and Technology,Yokosuka, Japan

Jun NishikawaAtmosphere and Ocean ResearchInstitute, University of Tokyo

James D. ReimerUniversity of the Ryukyus

Robin J. BeamanRichard FitzpatrickJames Cook University

Katsunori FujikuraTadashi MaruyamaJapan Agency for Marine-EarthScience and Technology,Yokosuka, Japan

20 Marine Technology Society Journa

A B S T R A C T

l

The untethered remotely operated vehicle (uROV) PICASSO-1, which is con-trolled in real time from a surface support vessel via a ϕ0.9 mm fiber optic cable,is capable of dives to 1,000-m depth at a duration of up to 6 h and yet is deployablefrom ships of sizes as low as 17 tonnes. The vehicle was developed at the JapanAgency for Marine-Earth Science and Technology, has carried out 63 dives to date,and is now operable by a team of four biologists and one technician. PICASSO-1 cancollect video (HDTV × 1, NTSC × 3) and environmental information (depth, temper-ature, salinity, dissolved oxygen concentration, fluorescence [chlorophyll a proxy],turbidity) concurrently, and this is output with vehicle heading, camera zoom, andother vital statistics via Ethernet. Acoustically obtained vehicle position information,deck and control room video, and sound data streams are also output via Ethernet,and the whole dive is recorded in a synchronous fashion on a logging/playback sys-tem that enables dives to be re-enacted in their entirety to facilitate analyses back inthe laboratory. Operations have been successfully carried out overseas using a char-tered dive boat, and the system represents a leap forward for exploration of theoceans to significant depths but at relatively low cost and with no loss in data quality.Keywords: midwater, mesophotic, untethered ROV, logging/analysis system

Introduction

Exploration of the oceans is gener-ally skewed toward regions close tomarine research stations, seaports,

and overseas territories of developednations, and this is particularly sofor in situ surveys using submers-ibles or remotely operated vehicles(ROVs). To accurately characterizethe biological diversity at any givensite, data must be collected on a vari-ety of spatio-temporal scales. This in-cludes the full spectrum from day andnight surveys to capture both diurnallyand nocturnally active or migratoryspecies through to seasonal and evendecadal oscillations in populationsand species compositions. At depthsbelow a few hundred meters, the num-

ber of sites for which such data existcan probably be counted on onehand, if not on one finger. This is cer-tainly true for the sessile benthic com-munities living on the sea floor but iseven more so for the denizens of thewater column.

Midwater research using submers-ible platforms suffers not only fromthe conundrum of needing to be inthe right water mass at the right timebut also having to fit into the cruiseschedules of large ocean-going researchvessels, whose primary research focus isusually on chemosynthetic benthic

ecosystems or geology (Armstronget al., 2004). The lead-up time fromthe submission of research proposalsto enactment of research cruises canbe anywhere from 8 to 20 monthsfor cruises within Japan, as an example(Lindsay, unpublished data), or 3 yearsor more for cruises in areas furtherthan about 1 week steaming distancefrom Japan’s Exclusive EconomicZone. The fluid nature of oceano-graphic currents and patchy nature ofthe spatio-temporal distribution oftheir inhabitants makes the planningof such cruises years in advance quitearbitrary in many respects, and thiscan negatively impact the success rateof submitted proposals. New tools,techniques and strategies for in situ re-search on midwater communities aredesperately needed.

Since 2005, the Japan Agency forMarine-Earth Science and Technology( JAMSTEC) has been developing adeep sea survey robot designed specif-ically for relatively rapid-response,worldwide deployment for midwatersurveys at low cost, specializing invideo image and environmental dataacquisition. The untethered remotelyoperated vehicle (uROV) PICASSO-1

(Plankton Investigatory CollaborativeSurvey System Operon-1) is the firststage in the development of a multiple-platform autonomous survey systemable to quantitatively characterizethe midwater environment, includingfragile components such as large par-ticulates and gelatinous plankton(Yoshida & Lindsay, 2007; Yoshidaet al., 2007a, 2007b; Maruyamaet al., 2009). The PICASSO-1 doesnot have a designated mother ship andhas been deployed successfully for seatrials from a variety of small, medium,and large boats and ships (Table 1).

The present paper introduces thesystem as it was deployed from char-tered dive boats under 16 m in lengthfor surveys in deep coral reef areas inOkinawa, Japan, and Osprey Reef,Coral Sea, Australia, as the first steptoward a bilateral collaborative deep-water survey program spanning boththe northern and southern hemispheres.

Deep coral reef areas were picked asthe remote or overseas test sites for thePICASSO-1 system because, as withthe midwater, research on the deep-water communities on and around coralreefs and on the geological–topologicaland/or oceanographic environment

July/Au

they inhabit is still in its infancy. As arule, depths between 60 and 300 m re-main largely unstudied, particularlynear reefs where the limited maneuver-ability of large ships precludes the useof tethered ROVs. Such areas can alsoexperience strong tidal runs and cur-rents leading to smaller ROVs beingunable to maintain position due tothe heaviness and large water resistanceof their tether compared to the smallpower output of their thrusters.Maneuverability of the ROV itself isalso quite limited and can be adverselyaffected by movements of the mothership due to currents and waves beingpropagated along the tether. A tetherserves to restrain the movement ofboth the vehicle and the mother shipbecause of the fixed nature of itsattachment. Even with a weight usedin lieu of a tether management systemto place the vehicle near the sea floor,away from the strongest currents, andto act as a damper on the effects of shipmovements, it is often impossible tohold station for scientific observationsfor more than a few minutes at a time.Due to these same restraints, researchon midwater communities in reefareas is even more sparse so as to be

TABLE 1

PICASSO cruises and support vessels.

Cruise ID
Dates Location Support Vessel Name
gust 2012

Length

Volume 46

Tonnage

Number 4

Dives

NT07-04
24 Feb–4 March 2007 Sagami and Suruga Bays Natsushima 67.3 m 1739 t 1-7
YK07-06
24–30 April 2007 Sagami Bay Yokosuka 105.2 m 4439 t 8-13
RM001
7–18 January 2008 Sagami and Tokyo Bays Rinkai Maru 18 m 17 t 14-22
N254
10–14 March 2008 off Nomomisaki Peninsula Nagasaki Maru 64 m 842 t 23-24
KH09-E03
14–18 Oct 2009 Tokyo Bay Hakuho Maru 100 m 3991 t 25-30
KT09-25
27 Nov–1 Dec 2009 Sagami Bay Tansei Maru 51 m 610 t 31-34
NT10-02
30 Jan–5 Feb 2010 Sagami and Suruga Bays Natsushima 67.3 m 1739 t 35-39
SG001
21–28 Sep 2010 off Ishigaki Island, Okinawa Sea Gull 15.23 m 19 t 40-47
CE001
29 April–16 May 2011 Osprey Reef, Coral Sea Coral Emperor 15.75 m 85 t 48-63
21

almost nonexistent. The untetherednature of the uROV PICASSO-1,where the communications fiber isfree to spool out from both the vehicleand mother ship fiber spoolers, makesit the ideal vehicle for surveys of suchareas because its movement in situ iscompletely decoupled from the move-ments of the ship. The superiority ofthis approach for in situ observationsin deep coral reef dropoff areas is high-lighted by comparing two recent pa-pers on the occurrence of the pygmyseahorse Hippocampus denise Lourieand Randall, 2003, in such areas(Nishikawa et al, 2011; Foster et al,2012). Nishikawa et al. (2011) usedthe uROV PICASSO-1 and were ableto measure swimming distances andbreathing rates in situ of the 21.5-mm-long fish, while Foster et al. (2012) onlybecame aware of the presence of thisfish after the sampled host gorgoniancoral was being examined in the shiplaboratory!

Vehicle SpecificationsPICASSO-1 is a small and light

(2.7 m long, 230 kg in weight) unteth-ered ROV (uROV) that, during op-erations, is only connected to itssupport ship via a thin optical fibercable (φ0.9 mm single mode nylon-coated fiber, Fujikura Ltd.). Becauseit has its own power supply in theform of batteries, much like an AUV,it is not constrained by a heavy umbil-ical tether and precious deck space nolonger needs to be taken up by a hugehydraulic winch. However, like anROV and unlike an AUV, real-timehigh-definition video data are availableto the pilot as the survey is carried out.Three thousand meters of cable arewound around a spooler inside atransparent acrylic case (φ157 mm,427 mm L, 5.2 kg in air, 0.7 kg inwater) with one case located inside the


vehicle and the other on the mothership locked into a pan-tiltable fiber de-ployment head. Fiber spools out ofeither or both the vehicle and shipsidespoolers at forces much weaker thanthe breaking strain of the cable allowingPICASSO-1 to travel unhindered bywater resistance to a distance of up to6,000 m from the support vessel. Thepower source of PICASSO-1 is a pairof oil-immersed pressure-compensatedbatteries, each composed of 56 sheettype lithium ion batteries (7 series ×8 parallel cells) with a total capacity ofabout 2 kWh and an energy density ofover 210 Wh/L. The pilot controlsthe vehicle in real time and has accessto high-quality, real-t ime videoimages using high data rate opticalcommunication tools. The color ofthe FRP fairing cover is mostly redbecause deep-sea organisms cannotsee light or reflections in the red spec-trum as a rule (Widder, 2005), yet redis an easy color to see at the sea surfacefor sighting of the vehicle before re-trieval after a dive (Figure 1). TheFRP cover incorporates several hatchesand holes to allow access to batterycables, switches and the like withoutneeding to remove the entire fairing.

PICASSO-1 has two lateral thrust-ers (100 W) with a 120° tilter and avertical thruster for highly maneuver-able cruising. The vertical thruster

l

also acts as a backup for ascent and de-scent if the tilt mechanism for thehorizontal thruster gets jammed dueto coral rubble or other material get-ting caught in the caterpillar trackthat rotates them. PICASSO has onevertical tail fin and two fins for stability,and a maximum cruising speed of1.6 knots. The vehicle is able to deployeither an HDTV camera or an auton-omous visual plankton recorder in itsmain payload space.

The main camera of PICASSO-1 isa broadcast quality HDTV camera.This high-resolution, high-sensitivitycamera allows scientists to identify or-ganisms to species level, in some cases,rather than only to generic or familiallevel, which is often all that is possi-ble with traditional NTSC cameras.PICASSO uses a wideband opticalcommunication system, developedin-house at JAMSTEC, with five inter-faces: one HD-SDI, three NTSC, fourRS-232C ports, two RS-485 ports,and eight-channel parallel I/O for thevehicle. The data from/to these portsare serialized/deserialized by a TLK3101gigabit transceiver. A 2.488 Gbps bitrate optical transceiver module pro-duced by Sumitomo Electric Indus-tries, Ltd., then converts the electricalsignal to/from optical signal. Thecustom-made optical communicationdevice installed in the vehicle is small(three printed circuit boards of 120 ×80 mm). On deck, communicationbetween PICASSO-1 and the shipsideMain Control System can also be es-tablished via Ethernet with the vehicleRS232 input/output being convertedto Ethernet protocol by an Ethernetconverter (EZL-50, Alpha ProjectCo., Ltd.). Ethernet communicationshave been incorporated widely intothe design of PICASSO to allow syn-chronous logging and display of allparameters, including video, both

FIGURE 1

The PICASSO-1 vehicle starting descentusing her vertical thruster.

within the vehicle control room andalso in remote locations such as theship bridge and the science room—

which can be over 100-m distant onsome large research vessels. Crowdingof the control room by eager scientists,students and other personnel cantherefore be mitigated to some degreeand the ship’s crew can check on thetemporal status of the pre-dive checkor check PICASSO’s underwater posi-tion without having to resort to fre-quent communications via handheldradio transceivers.

The main camera is a SONY com-pact high-definition camera system,HDC-X300K, with an original cameracontrol board employing a CAN (con-troller area network) interface in analuminum pressure hull. The CANbus is based on the broadcast commu-nication mechanism. Every messagehas a message identifier, which isunique within the whole networksince it defines content and the priorityof the message. The CAN bus also en-sures bit and frame synchronization.The maximum data transmission rateof CAN is 1 MHz. A special coaxialunderwater cable with pressure-tightSMB type RF connectors was madefor connecting between pressure hulls.The HDC-X300K has the followingspecifications: effective pixels 1,440 ×1,080, sensitivity of 2000 lx at F10,minimum luminance of 0.003 lx atF1.4, smear level of −120 dB, and sig-nal to noise ratio of 52 dB. Its imagesensor system consists of three 1/2″1.5M-pixel CCDs. It has a measuredcamera angle in air of 48.72° (H) ×25.9° (V). The focus, iris, zoom, shut-ter speed, gain, slow shutter mode,auto white balance and black balanceof this camera can be remotely con-trolled via a handheld control panel(M-AUDIO Xsession Pro). The HD-SDI output signal of the camera is

directly transmitted to the on-boardsystem as an optical modulation signalvia the optical communication system.

The NTSC cameras (WAT-240Vivid (G-2.5), Watec) on PICASSO-1have the following specifications: effec-tive pixels 768 × 494, 450 lines, mini-mum luminance of 1.6 lx at F2.0,a signal to noise ratio of >50 dB.The measured camera angle in air is80° (H) × 59.3° (V) and the image sen-sor system consists of one 1/4″ CCD.Two of the NTSC cameras areforward-mounted and positioned forstereoscopic object scale estimation.One more NTSC camera is positionedto face down at a 90° angle from theother cameras to image the seafloor ashorizontal transects are made. A finalNTSC camera (monochrome high-sensitivity version of the WAT-240)is positioned on the rear tailfin ofPICASSO pointing forward, allow-ing the thrusters and most protrudingparts of PICASSO that might catch ongorgonians, ropes, nets or other ob-stacles to be directly visualized. Thisforward-pointing camera is also usefulfor checking that PICASSO’s ownfiber optic cable is not in danger ofbeing wrapped around the thrustersand because of the wider angle ofview is scientifically useful also forplacing the stereo NTSC and HDTVvideo data in a wider context at a largerscale. Three NTSC signals can be sentfrom the vehicle to the main controlunit at one time so a switch is used toswitch within the vehicle betweensending the downward-pointing orrear-mounted camera video feed tothe surface.

Illumination is by HID lamps(three custom 30-W lamps divertedfrom car use) and/or two handmade20-W LED array lights. Two of theHID lights are positioned below andforward of the HDTV camera on a

July/Au

stainless steel bracket and angled suchthat the beams of light illuminate tar-gets approximately 50 cm from theHDTV camera lens at an angle ofabout 45°. This ensures that out-of-focus marine snow particles occurringcloser to the camera lens are not illumi-nated. The single forward-pointingHID light is used when searching fortargets and video quality is not para-mount. High-power white LEDs(NCCW023S, Nichia Corporation)have been combined with a copperbase plate, resistors, an underwaterconnector, and a 1/2″ clear tube filledwith oil to balance internal-externalpressure (Yoshida et al., 2007a, 2007b).One of these LED arrays incorporatesred LEDs to allow PICASSO-1 to op-erate in “stealth” mode due to the factthat most deep sea animals are thoughtto not behaviorally respond to red light(Widder, 2005; Raymond & Widder,2007). The other white LED array isused to illuminate the sea floor forthe downward-pointing NTSC cam-era. The LED arrays also serve as back-up should problems occur with theballast to the HID lights. Usuallyonly a single side-mounted HID lightand the downward-pointing LEDarray are used during surveys to pre-serve battery power—the amount ofillumination being more than ade-quate for the sensitivity of the cameras.

Environmental sensor packages in-clude a CTD-DO (XR-420CTDm,RBR Ltd. ) and a backup CTD(XR-420CTDm, RBR Ltd.) for mea-suring temperature and dissolvedoxygen concentration and for com-puting depth, salinity and water den-sity for real-time input into the loggingsystem, as well as a fluorometer-turbidity sensor (ECO FLNTU Puck,WET Labs, Inc.) to obtain data onchlorophyl l a and marine snowconcentrations.

gust 2012 Volume 46 Number 4 23

Other vehicle specifications for thePICASSO-1 are shown in Table 2.

Because PICASSO-1 is neutrallybuoyant at survey depth it has severallayers of safety built in to ensure its re-turn to the surface. In addition to themain computer onboard PICASSO, asecondary CPU has also been incorpo-rated into the main instrumentationpressure housing to allow monitoringof vehicle status from the main controlnotebook computer and jettisoning ofthe ascent ballast (45 × 100 × 200mm,density 7860 kg/m3, 8 kg SS400 car-bon steel block). If optical com-munications have been continuouslyinterrupted (<−20 dB) for 40 s, bothof these systems send a ballast jettisoncommand. Ascent ballast is also

T

S

aPtob

cC


jettisoned automatically by a timerrelay if dive length reaches 8 h dura-tion and/or by a low-voltage protec-tion circuit if main battery voltagefalls below 21.0 V, thereby triggeringa full system shutdown. The work-ing limit of PICASSO’s batteries is21.0–29.4 V and is the lower limitof overdischarge of 3.0 V/cell, withvoltage below approximately 2.8 V/cellleading to delamination on the sheetelectrode and the ensuing destructionof the battery. A circuit has been in-stalled to nullify residual magnetismin the electromagnet that holds theascent ballast and a spring protrudingfrom the central surface of the electro-magnet exerts a force of 0.6 kg to aidjettison.

l

Shipside SystemPower Supply System

Electrical power to the shipsidesystem is supplied via an isolationtransformer (1.5KVA, MD-3000,UNION Electronics Co. Ltd.) toan uninterruptible power supply(UPS1010HP [YEUP-101PA], YutakaElectric Mfg. Co., Ltd.). The UPS willtake single-phase electric power be-tween AC80 V and AC130 V to out-put at AC100 V and therefore incombination with the transformer,power supply to the shipside systemof PICASSO-1 can be anywhere fromAC80 V to AC260 V with a frequencyof 50 or 60Hz.Output is to amaximumof 700 W (7A) and the PICASSOshipside system draws a maximum ofapproximately 500 W (5A). The UPScan supply power for up to 7 min at700 W in the case of a power outage.

Main Control SystemThe ma i n c on t r o l un i t f o r

PICASSO-1 is incorporated into awatertight protective case (84.7 ×72.2 × 46.3 cm, model 1690NF,Pelican Products, Inc.) (Figure 2), forease of portability and setup, anddraws approximately 2.4 A (240 W).

The system was designed to be re-dundant with respect to both controland recording of data (Figure 3). Atopside control unit (TCU) communi-cates with PICASSO-1 via RS232Cprotocol and a custom controller boxallows the following functions ofPICASSO-1 to be controlled even ifthe main control notebook computerrunningWindowsXP (CF-S9JWECPS,Panasonic Corporation) is offline dueto hardware or software issues: thrusterpower [on/off], thruster tilt, forward-reverse-left-right on tiltable thrusters,up-down on vertical thruster, HIDlight and LED light on/off (each

ABLE 2

pecifications of PICASSO-1.

Item
Specification Remarks
Dimensions
2.7 m (L) × 0.94 m (W) × 0.68 m (H) without VPR
Weight
230 kg in air
Depth rating
1,000 m
Cruising speed
1.6 kt
Endurance
6 h
Operation mode
uROV
Propulsion
2 horizontal 100-W thrusters with tilt system,1 vertical 100-W thruster
descent and ascentballast (8 kg each)

Communicationsinstrumentation

2 G bps optical communication device, LAN,ARGOS transmitter, flasher

Navigationinstrumentation

MEMS gyro (AHRS400MB-100, Crossbow),Doppler velocity log (Explorer, TeledyneRD Instruments), backup CTD,a compass(TruePoint HMR3500, Honeywell),transponder (VLT-1, Desert Star Systems)

DVL measures altitudeand velocity vs. bottomb

Experimentpayload

CTD-DOc, fluorometer-turbidity sensor,4 × NTSC cameras, 3 × 35-W HID lamps,Visual Plankton Recorder*, high definitionTV camera*

*pick only one device

ressure sensor (circa 2005, Kowa Corporation) used during Okinawa and Australia missions and updatedCTD subsequently.

Velocity data not correct during Okinawan and Australian missions due to old firmware.TD with no DO deployed during Australian mission and used as backup CTD thereafter.

separately controllable), rear/forward-looking and downward-lookingcamera feed switch, communicationmode (optical or Ethernet), vehiclemode (boot/initial or dive/cruisemode), descent ballast jettison /remagnetization, ascent ballast jettison/remagnetization. A vacuum fluores-cent display panel on the TCU dis-plays the following: link [optical/Ethernet], uplink strength, downlinkstrength, vehicle depth, battery volt-age, compass heading, thruster powerstatus [on/off ], and an LED array dis-plays the following: receipt of commu-nications (ether or optical) packets

from PICASSO-1, vehicle mode (ini-tial or dive), receipt of optical commu-nications packets, ascent ballast dropcommand sent/not sent, descentballast drop command sent/not sent,main instrumentation pressure hous-ing leak status, power pressure housingleak status. The TCU communicateswith the main control notebook com-puter via Ethernet and the PICASSO-1

July/Au

is usually controlled and monitoredwithin the control software (NOU,Navigation Observation Unit) andgraphical user interface (NOU_view).

In addition to the equipment out-lined above, the main control unit alsocontains a display (FlexScan S2243W-HX LCD display, Eizo Nanao Corpo-ration) for multimonitor output fromthe control notebook computer on

FIGURE 2

PICASSO-1 being piloted by JN in the galley ofthe Coral Emperor at Osprey Reef, Australia.The main control unit is in the foregroundand the navigation unit in the background (top).The main control unit during operations (bot-tom). The graphic user interface of PICASSO-1displaying vehicle status parameters, supportvessel heading, and environmental data is onthe left, high-definition video on the right, NTSCvideo monitors above, and the main controlnotebook computer below, displaying the deck-deployed Ethernet camera feed and otherreal-time technical statistics for PICASSO-1that are not displayed on the GUI.

FIGURE 3

Schematic diagram of the main control system.


which a graphic user interface ofPICASSO-1 vehicle status parameterssupport vessel heading, and environ-mental data is displayed (Figure 2). Ajoystick (CH Fighter Stick) for drivingand a handheld control panel for cam-era control (M-AUDIO Xsession Pro)are connected via USB to the controlnotebook computer. All camera param-eters including shutter speed, gain,white balance and black balance canbe controlled.

The HD-SDI signal from theSONY compact high-definition cam-era system (HDC-X300K) aboardPICASSO-1 is demodulated and out-put f rom the on-board sys tem(SSDEV-0614), passes through a digi-tal to analog video converter (SDI toAnalog mini converter, BlackMagicDesign) where it is converted toNTSC format and then converted toan H.264 stream by a four-channelvideo encoder (AXIS Q7404, AXISCommunications) to be output viaEthernet to the logging system (seebelow and Figure 5). It is also through-put natively to a solid-state memoryrecording device (Panasonic AG-HPG20) to be recorded on PS2 cards,channeled natively to an HDCAMrecorder (Sony HDW-250), with timecode set to local time manually, andthen throughput to an SD-HDI toDVI converter (HDLink, BlackMagicDesign) and then to theHDTVdisplay(FlexScan S2243W-HX LCD display,Eizo Nanao Corporation) in the MainControl Unit. This parallel configura-tion ensures that even if a tape runs outor a memory card becomes full, novideo data are lost as they are alsologged via Ethernet, and if an Ethernetproblem occurs data are still recordedby the other devices. The time codeoutput from the HDW-250 deck isoutput to the logging system by co-axial cable, where it is converted to


USB by a USB-LTC/RDR LTCReader (Adrienne Electronics) (Fig-ures 3 and 5).

TheNTSC signals fromPICASSO-1are channeled through three videosplitters (VSP-2M, YKmusen) withthe throughput signal being re-corded by three MPEG-4 recorder/players (VP6060 [DMV-MP4],VOSONIC Technology Corporation)on SDI cards and then output to threeNTSC monitors (702TSV) in theMain Control Unit, while also beingoutput into the four-channel videoencoder (AXIS Q7404, AXIS Com-munications) to be output via Ethernetto the logging system (see below andFigure 5). Again, this ensures thatimage data are always recorded, evenif one or the other of the recording de-vices malfunctions during a dive.

Navigation SystemThe navigation system (Figure 4)

consists of a navigation unit incor-porated into a watertight protectivecase (63 × 49.2 × 35.2 cm, model1620NF, Pelican Products, Inc.) andperipherals and draws approximately1.1 A (110 W). Support vessel posi-tion within the GPS is monitored viaa DGPS Loop Antenna (GPA019,Furuno Marine Electronics) input intoaWAAS/DGPS receiver (GP37, FurunoMarine Electronics) and output viaRS232 protocol into the navigationcomputer—a notebook computer run-ning Windows XP (CF-S9JWECPS,Panasonic Corporation). The samemodel of notebook computer is alsoused for the control system and all soft-ware necessary for either use is installedon both computers to allow rapid re-placement of either computer in thecase of a malfunction during a dive.Ship heading is monitored via compass(C100, KVH Industries, Inc.) outputvia RS232 protocol to the navigation

l

computer. The XYZ position of thePICASSO-1 vehicle beneath thesupport vessel is monitored at 34-41 kHz using a short baseline acoustictracking system (PILOT STM-3VLT-1, Desert Star Systems) and out-put via USB to the navigation com-puter. The navigation computer runsDivebase PILOT software, whichthen outputs XYZ position data to anin-house software package via RS232protocol. This software uses shipGPS and vehicle XYZ location datato calculate the GPS position ofPICASSO-1 for real-time logging anddisplay in a window on the navigationmonitor and to be sent via Ethernet tothe logging system (see below and Fig-ure 5). GPS and uROV position datacan also be acquired from the hull-mounted acoustic positioning systemused by JAMSTEC vessels, loggedand displayed.

Logging SystemNetwork cameras monitor opera-

tions on the back deck (M1011-W,AXISCommunications) and in the con-trol room (M1054, AXIS Communica-tions) and provide a real-time H.264video stream to the logging com-puter (MacBook Pro6,1: Intel Corei5, 2.53 GHz, running Windows 7Professional under BootCamp) via anEtherne t hub (Swi tch-S8PWR[PN21089K], Panasonic Electric WorksNetworks Co. Ltd.). The M1054 net-work camera can be positioned torecord the navigation control unitwhen PICASSO-1 is being monitoredacoustically. Sound in the controlroom is recorded using the MacBookPro built-in microphone. The videofiles from the six cameras, GUI screenshowing PICASSO-1’s status, envi-ronmental parameter vertical profilescreen and sound file can be playedback in a synchronous fashion to

reconstruct the dive for trouble-shootingand scientific analysis through refer-ence to the time data recorded fromthe control notebook computer inter-nal clock and HDCAM time codedata. All vehicle status, environmental

parameter and position data can alsobe output to a csv file which includesfields in Darwin Core format for directimport into the Ocean BiogeographicInformation System. Darwin Coreformat is a standard agreed on by bio-

July/Au

geographical information databasesand defines the format in which datamust be supplied to such databases.For example, when the time at whichan observation is made is to be outputto the csv file, it must be a numericvalue in the field “Time of Day” andbe expressed as decimal hours frommidnight (e.g., 12.0 = mid-day, 13.5 =1:30 pm). For longitude information,the value should be numeric, in thefield “Longitude”, and expressed indecimal degrees (East & North = +;West & South = –), with GPS-deriveddata being referenced to the WGS/84da tum. The fu l l Da rw in CoreSchema can be found at http://www.scarmarbin.be/obisschema.php and/or http://www.iobis.org/data/schema-and-metadata.

Size Analysis SystemFor measuring the distance to an

object and estimating its size usingstereovision, triangulation is generallyused. In the present method, a dispar-ity map is prepared. The disparity mapis a depth map where the depth infor-mation is derived from offset images ofthe same scene. To measure disparityin the camera system, a given pixel lo-cation in either the right or left imagecoordinate frame is computed with astereo matching technique. The dis-parity map is constructed by imag-ing a calibration board with the twoforward-mounted NTSC camerasprior to a dive. The calibration boardis a white acrylic board covered in atleast 400 black circles, the center circlebeing 30 mm diameter and all others24 mm in diameter with the centerof each circle being 30 mm from thecenter of its nearest neighbors. Imagesare recorded at a distance of 1 m fromthe cameras with the board held per-pendicular to the vehicle axis, andalso sloped left, right, forwards and

FIGURE 4

Schematic diagram of the navigation system.


backwards. A further calibration imagepair is captured at a distance of 2 mfrom the cameras. Post-dive theNTSC video is played back using theMPEG-4 recorder/players and framepairs are captured using a PicoloAlert PCIe video capture card (Euresys


S.A., Belgium) in a Vostro 200 desk-top tower computer (Dell, USA) run-ning Windows XP. Objects (distancesand volumes) to be measured are man-ually selected using specially developedsoftware (AVSCalcXYZ_PICOLO,Applied Vision Systems Corporation,

l

Japan). Accuracy was measured tobe 2.7 ± 2.1% when measuring a10-mm-long object at a distance of1 m from the camera lenses. Sizemeasurement data are exportable as acsv file.

OperationsOkinawa, Japan

PICASSO-1 and all associatedequipment was sent in a 12-foot ship-ping container to Ishigaki Island,Okinawa, Japan. The diving boatSeagull (Figure 6) was not equipped

FIGURE 5

Schematic diagram of the logging system.

FIGURE 6

The chartered diving vessel Seagull withPICASSO-1 on the raised platform (top), low-ering PICASSO-1 into the water using handwinches and a pivoting-sliding platform attachedto the hydraulically operated dive elevator(middle), PICASSO-1 deployed from the char-tered dive vessel Coral Emperor off OspreyReef, Australia (bottom).

with a crane, so PICASSO-1 was low-ered into the harbor using a rentedcrane and dragged using ropes to thehydraulically powered lifting platformat the stern of the boat (Figure 6). Theplatform was unable to be lowereddeep enough for PICASSO-1 to bepulled directly on to it so a speciallymanufactured sliding deployment-retrieval rack was attached to theback deck and lifting platform suchthat with the use of hand-poweredwinches it was possible to use the lift-ing power of the hydraulic platform toraise the vehicle to the level of the backdeck where she could be draggedunder cover into the working andmaintenance space by hand. Deploy-ment needed a minimum of six peo-ple: one on each of the two handwinches, two holding guide ropes,one to pull the snap shackles to re-lease PICASSO-1 into the water atthe beginning of the dive, and one tohandle the fiber optic cable so it wasnot kinked or damaged during thedeployment—thereby ending thedive as optical communications wouldbe lost.

A total of eight dives were con-ducted between 21 and 28 September2010 (Table 1). Technical problemswith the SBL system (a loose con-nection within the STM-3 and theVLT-1 transponder battery not hold-ing a charge) prohibited dive surveysin regular survey mode, so dives wereconducted by thrusting down straightto the sea floor where it was possible toverify in the video feed that PICASSO-1was not being swept away from thevicinity of the support vessel by thecurrent.

Eight dives were made with thedeepest being only to 111-m depthdue to concerns about the where-abouts of the vehicle underwater. Ifoptical communications are suddenly

lost for 40 s the vehicle automaticallyjettisons the 8 kg steel ballast blockand floats to the surface. If waves arelarge it can be difficult to see thesmall vehicle from the roof of the sup-port vessel for retrieval so it is very im-portant to know the position of thevehicle as accurately as possible—notjust for science but to avoid losingthe vehicle to strong surface currents.

Osprey Reef, AustraliaPICASSO-1 and all associated

equipmentwas sent in a 20-foot shippingcontainer to James Cook University,Cairns, Australia, unpacked and ferriedto the Cairns Marina in a 2-tonne rent-a-truck with a hydraulically-powered lift-ing platform. The support vessel CoralEmperor was equipped with a smallcrane (560 kg load at maximum hori-zontal extension 4.36 m), which wasused to load PICASSO-1 onto theupper deck where a semidry spacehad been made using tarpaulins andplywood boards. A total of 16 dives, in-

July/Au

cluding two training dives at Lizard Is-land, were conducted between 29 Apriland 16 May 2011 (Table 1, Figure 7).

The SBL system performed quitepoorly with the maximum readingtaken during operations at only 293 mdepth (55 m to port, 135 m fromstern, 327 m total distance), far lessthan the 1,000m advertised by theman-ufacturer and closer to the 330 m rangeadvertised for the miniature TLT-3transponder. The extremely steepslope of the reef wall at Osprey Reefallowed us to zigzag down the walldirectly below the mother ship usingunderwater navigation techniqueshoned through SCUBA diving.

Midwater dives were done by firstanchoring a staple polyethylene (silver)rope to a mooring on the reef face atRaging Horn and running a zodiac in-flatable boat due west until the 500-m-long rope was taught. A weighted baittrap (ca. 20 kg) was then tied to the endof the rope before the trap was thrownover the side and left to swing down

FIGURE 7

Hillshade image of northwest Osprey Reef (right) developed using lidar and singlebeam bathym-etry data over depth ranges +3 to −815 m. PICASSO-1 dive sites 50-63 indicated on westernside of Osprey Reef—the location of which is indicated in the left panel.


until it landed on a sloping shelf on thereef wall at 365 m depth. PICASSO-1was equipped only with ascent and notdescent ballast for all dives at OspreyReef and was calculated to have nomore than 100 g positive buoyancy at400 m depth. A SCUBA diver guidedPICASSO-1 by swimming toward themooring at a distance of 5-10 m infront of the main camera while theuROV followed him under its ownpower. After the rope was identifiedon the main camera, PICASSO-1 wasthen piloted under the rope and turnedaround to face due west until the ropewas again in frame. It was then possiblefor PICASSO-1 to follow the ropethrough the midwater at altitudes ofbetween tens and hundreds of metersabove the bottom to the final targetdepth of 365 m. Observations of mid-water animals such as the ctenophoresThalassocalyce inconstans Madin &Harbison, 1978 (255 m depth, 17.2°C,35.36PSU) and Kiyohimea usagiMatsumoto & Robison, 1992 (353 mdepth, 14.1°C, 35.47PSU) were fleet-ing since the rope always had to bekept in frame to ensure our safe returnto the surface within sight of the mothership by once again following the ropeto themooring at 40mdepth. The baittrap, complete with cookie cutter sharkgouge marks on the bait, was retrievedby using the zodiac’s forward motionto lift it back to the surface.

Preliminary ResultsOkinawa

Elegant firefishNemateleotris decoraRandall & Allen, 1973 were observedduring dive 43 at 111mdepth (25.4°C,34.86 PSU) in a group, suggesting theymay have been juveniles (Figure 8a).This is a new depth record for this spe-cies as the previous maximum depth re-cord was at 70 m depth (Myers, 1991).


OspreyElegant firefish N. decora were also

observed at Osprey Reef on 8 May2011 during dive 50 at depths of 85 m(24.6°C, 35.8PSU) and 116m (23.7°C,35.9PSU), making 116 m depth thedeepest record for this species (Fig-ure 8b). At Osprey they appeared sin-gularly rather than in a group and werefound over topologically diverse car-bonate cliff substrate covered in redcoralline algae, gorgonians and patchesof coral sand. Their depths of distribu-tion, temperature and salinity environ-ments were therefore similar in boththe northern and southern hemi-spheres, although the substrate theyoccurred above differed. The green

l

calcareous algaHalimeda was observedjust shallower than the shallowest fire-fish at 82.5mdepth, as was the zoanthidEpizoanthus illoricatus Tischbierek,1930 (Figure 8c).

An attempt was made to constructa large-scale image map of the reefwall using the software packageMofixLIGHT v2.1.0 (View Max Ser-vice Emaki Co., Ltd.). The NTSCvideo recorded by the right-hand ste-reo camera was played back usingone of the MPEG-4 recorder/playersand the output converted to USB 2.0using a USB video capture cable(PCA-DAV2, Princeton Technology).The software was run on a HewlettPackard h8-1080jp Pavilion desktopcomputer (Intel Core i7-2600 CPUat 3.4 GHz, 4.00 GB RAM, NVIDIAGeForce GT 530Graphics Card). Theneutral buoyancy of the vehicle andthe resulting stable video allowedlarge sections of the reef wall to bemapped (Figure. 9), something thatwould be difficult using a tether-constrained ROV, but the propensityof biologists to try and capture inter-esting animals in the center frame onthe HDTV camera meant that the ve-hicle would suddenly face the reef wallfrom time to time and the software wasunable to combine these frames intothe map. Dedicated dives for mappingor a 360°-capable camera that cancapture images regardless of the vehi-cle’s heading are two possible ways toovercome this problem.

The broadcast -qual i ty high-definition television footage recordedby PICASSO-1 allowed species identi-fications to be made for well-knownand easily recognizable organismssuch as N. decora (Figure 8), the livingfossil cephalopod Nautilus pompiliusLinnaeus, 1758 (Figure 10a), and thepygmy seahorse Hippocampus deniseLourie and Randall, 2003 (Nishikawa

FIGURE 8

Elegant firefish Nemateleotris decora duringdive 43 off Ishigaki Island, Okinawa, at 111 mdepth (a), and at Osprey Reef during dive 50at 116 m depth (b, white arrow). Frame grabof reef wall at 82.5 m depth during dive 50(c) showing the calcareous alga Halimeda(white arrow 1) and the zoanthid Epizoanthusilloricatus (white arrow 2).

et al, 2011). Even the NTSC videofrom the other cameras allowed speciesidentifications to be made of charis-matic macrofauna such as a juveniletiger shark Galeocerdo cuvier Péronand Lesueur, 1822 at 360 m depth—a new depth record for a juvenile ofthis species. Even for lesser-knowngroups such as the squat lobsters, itwas usually at least possible to identifyanimals to genus level (e.g.Bathymunidasp. and Munida sp.; Figures 10band 10c) if not to species, such as forBabamunida javieri (Macpherson,

July/Au

1994)—a first record for Australianwaters (Figure 10d).

The red LED array was used to ob-serve organisms attracted to the baitedtrap on 13 May 2011 at 365 m depth.Large numbers of the ruby snapperEtelis carbunculus Cuvier, 1828 andone oblique-banded grouper Epine-phelus radiatus Day, 1868 came closeto the bait when illuminated with redlight but were quick to disperse whenthe HID lights were turned on (Fig-ure 11), suggesting that the use ofsuch red light illumination would bevaluable for “stealth” missions to ob-serve the natural behavior of deep-seaanimals.

ConclusionsThe untethered nature and small

size of the PICASSO-1 vehicle makesit an ideal tool for exploration of theocean’s midwater, mesophotic reefcommunities and other deep waterecosystems in hard-to-access localities.Perhaps the most revolutionary aspect

FIGURE 10

The living-fossil cephalopodNautilus pompilius(a), the squat lobsters Bathymunida sp. (b),Munida sp. (c) and Babamunida javieri (d).

FIGURE 9

A videomosaic of the reef wall at Osprey Reef.

FIGURE 11

The ruby snapper Etelis carbunculus and anoblique-banded grouper Epinephelus radiatusunder red light (top) and as the HID lightspowered on (bottom).


of the PICASSO system is the loggingand playback system, which allowsboth recreation of a dive duringpost-cruise analysis and data outputthat can be imported directly intoJAMSTEC’s Biological InformationSystem for Marine Life (http://www.godac.jp/bismal/e/) and the OceanBiogeographic Information System(http://www.iobis.org/). The abilityto take tissue or DNA samples usingthe PICASSO-1 would grea t lyincrease its value as a survey tool andthe development of such capabilitiesshould be actively pursued.

AcknowledgmentsThe quality of the HDTV images

recorded by PICASSO-1 is such thatmany of the sea trials and scientific sur-veys have been able to be subsidized bycollaborations with other parties. Seatrials in Japan were subsidized to an ex-tent by NHK (RinkaiMaru cruise) andTV-Asahi (Sea Gull cruise) and byDigital Dimensions, BBC, DiscoveryChannel and Channel 9 for the Aus-tralian cruise on the Coral Emperor.High-quality images are also of greatvalue for scientific research, and theauthors would like to thank the Aus-tralian Institute of Marine Scienceand James Cook University, who alsocontributed toward costs of the Aus-tralian survey. All other costs wereborne by JAMSTEC. The authorsexpress their appreciation to Mr.Kazuhiro Yagasaki, who constructedthe videomosaic in Figure 9, andwould also like to thank Ms. AsukaIto, Ms. Atsuko Watanabe, and Mr.Yoshifumi Noiri for logistical officesupport. This study is supported inpart by the fund for interdisciplinarycollaborative research by the Atmo-sphere and Ocean Research Institute,University of Tokyo.


Corresponding Author:Dhugal J. LindsayJapan Agency for Marine-EarthScience and Technology2-15 Natsushima-cho,Yokosuka 237-0061, JapanEmail: [email protected]

ReferencesArmstrong, J., Becker, K., Eagar, T.,Gilman, B.,

Johnson, M., Kastner, M.,… Solow, A. 2004.

Future Needs in Deep Submergence Science:

Occupied and Unoccupied Vehicles in Basic

Ocean Research. Committee on Future Needs

in Deep Submergence Science, Ocean Studies

Board, Division on Earth and Life Studies,

National Research Council of the National

Academies. Washington, DC: The National

Academies Press. 1-135 pp.

Foster, R., Bridge, T.C.L., & Bongaerts, P.

2012. The first record of Hippocampus denise

(Syngnathidae) from Australia. Aqua, Int J

Ichthyol. 18:55-7.

Maruyama, T., Kato, C., Yamamoto, H.,

Fujikura, K., Sato, T., Fujiwara, Y., …

Sasaki, T. 2009. Research activities of the

Research Program for Marine Biology and

Ecology from 2004 to 2008. JAMSTEC

Rep Res Develop. 9(1):13-74.

Myers, R.F. 1991. Micronesian Reef Fishes

(2nd ed.). Barrigada, Guam: Coral Graphics.

298 p.

Nishikawa, J., Fitzpatrick, R., Reimer, J.D.,

Beaman, R. J., Yamamoto, H., & Lindsay,

D.J. 2011. In situ observation of the Denise’s

pygmy seahorse Hippocampus denise associated

with a gorgonian coral Annella reticulata at

Osprey Reef, Australia. Galaxea, J Coral

Reef Stud. 13:25-6.

Raymond, E.H., & Widder, E.A. 2007.

Behavioral responses of two deep-sea fish

species to red, far-red, and white light.

Mar Ecol Prog Ser. 350:291-8.

Widder, E.A. 2005. Using red light for in

siu observations of deep-sea fishes. Deep-Sea

Res I. 52:2077-85.

l

Yoshida, H., Aoki, T.,Osawa,H., Tsukioka, S.,

Ishibashi, S., Watanabe, Y., … Ishikawa, A.

2007a. Newly-developed devices for two types

of underwater vehicles. In: Oceans 2007 Con-

ference Proceedings, 1-6. Aberdeen: IEEE.

Yoshida, H., & Lindsay, D.J. 2007. Devel-

opment of the PICASSO (Plankton Investi-

gatory Collaborating Autonomous Survey

System Operon) system at the Japan Agency

for Marine-Earth Science and Technology.

Jpn Deep Sea Technol Soc Rep. 54:5-10.

Yoshida, H., Lindsay, D.J., Yamamoto, H.,

Tsukioka, S., Shimura, T., & Ishibashi, S.

2007b. Small hybrid vehicles for jellyfish

surveys in midwater. In: Proceedings of the

17th International Offshore and Polar

Engineering Conference, p. 127. Lisbon,

Portugal: International Society of Offshore

and Polar Engineers (ISOPE).

the untethered remotely operated vehicle picasso-1 and its

Documents