Mobile & Marine Robotics Research Centre University of ... · PDF fileChallenges of ROV System Integration for Deep Water Habitat Mapping ... Sonar & INS Simulator Development......

Challenges of ROV System Integration for Deep Water Habitat Mapping

Mapping cold water corals at 800 - 1,200 m on the margins of the continental shelf off west coast Ireland

Dan Toal

Mobile & Marine Robotics Research Centre University

ofLimerick

Overview of TalkMobile & Marine Robotics Research Centre

HEA PRTLI 3 - Deep Ocean Habitat Mapping Project

High Resolution Imaging with ROV Deployed Sensors, Positional Accuracy v Resolution Trade Off and Importance of Nav/positioning

Systems Integration ROV, Ship, Sonar & INS

Simulator Development. VR Command Centre, & Real-time Imaging Sonar Simulator

Research Cruise CE0505 June 2005

Main Thrust of ResearchIntervention Autonomous Underwater Vehicles

Dan Toal, Edin (dinO) Omerdic, James Riordan, Levente MolnarPostgrads: S. Nolan, E. Thurman, J Horgan, H Ahmad

Mobile & Marine Robotics Research Centre

Robot operation close to the seabed and marine structures Navigation & controlArtificial intelligence, neural-networks

Imaging sonars Sidescan & Multibeam.

Simulators, Immersive VR Command CentreVision in vehicle control.Robot arms for intervention.Proximal object detect sensors for operation near seabed

Facilities & Equipment: Robots

• Tethra & Mini-Sub AUVs • Wheeled robots, robot arms• Marine survey equipment

– Reson 7125 Multi-beam Sonar– Sound Velocity Profiler– Tritech SS sonar– Tritech Colour Camera.– Navigation / positioning

Ixsea Fibre Gyro INSRDI Doppler Velocity Log Microbath precision depth &

altimiter sensor• Obstacle avoidance

6 Tritech altimeters

• Thrusted Pontoon

Mobile & Marine Robotics Reseaech Centre

• 4.5m deep test pool

• 50m Olympic pool

Test Pool and 50m Arena Pool

HEA PRTLI 3 Deep Ocean Habitat Mapping

Using a Remotely Operated Vehicle

HEA PRTLI 3 project In collaboration with Department of Earth & Ocean Science, NUIGalway,

Capital for ROV Toolskid(Multibeam + Precision Navigation)Funding delayed by 18 month

Year 1 Tender, design, integrateYear 2 Shakedown cruise + ScienceYear 3 Deep ocean mapping

Year 1 -Year 2 TenderYear 3 Design Integrate, Do or Die Cruise.

Operations Near Seabed

Celtic Explorer ROV Deployment

‘Arc’ Mounds

Belgica Mounds

Sonar Imaging of the Seabed (Sidescan, Multibeam),

Positional Accuracy v Resolution Trade Off

&Importance of Navigation / Positioning

Experimental Sidescan Data Logger

Geoacoustics 100/410 kHz sidescan sonar image acquired with a NI DAC card interfaced to a laptop PC

Custom Labview application as the data acquisition software

Echo data fused with concurrently generated GPS and IMU data in XTF format.

Screen shot of data gathered in Survey of Clewbay, Ireland

Dynamically positioned ship (DGPS)

Time to reception of echo for each beam gives 120 range / angle pairs across the track of the ship.120 range angle pairs per ping as the vessel moves builds up the Bathymetry (shape of seabed).Assumes velocity of propegation, position and attitude of vessel/sonar are known

Sonar Imaging of the Seabed Multibeam

Sonar Imaging of the Seabed Sidescan & Multibeam

Necessary for building up image1. Sonar returns2. Knowledge of position & orientation of sonar / vehicle3. Measure of motion disturbance of platform4. Refraction of beams due to Sound Velocity Profile (SVP)

Errors in measurement and/or approximations all result in degradation of imagery.

‘Fuzzy out of focus images’Sonar processing / filtering result in averaging (smoothing).

Sonar Imaging of the Seabed Effects of Platform Motion

Heave

YawRoll

SurgeSway

Pitch

Roll – beam footprint shifts laterally

Yaw – beam footprint rotate

Heave afects dimension of footprint

Surge & Sway - afect position of footprint

Pitch – beam footprint shifts forward or back

Combination of all six disturbances?Beam refraction and range / delay?Unknown geometry of bed,

Instrument misalignment - patch test.

Motion Disturbance of Platformsome figures on the effects

Assume: 1,000m depth, depression angle 30, Fan beam - 1° along track, 120 beams – equiangular beam spacing - 1° across track

1° 5°

Pitch offset

17m along track, 1 beam width

87m, 5 beams

Rolloffset

Outer beam 72m across track

412m across track

Yawoffset

Outer beam 30m along track

150m along track

Variation in Sound Velocity Refraction

Sound Velocity

Error at 500m depth

Marine Survey Ship Mounted Multibeam

Dynamically positioned ship (DGPS)

Marine Survey ROV Mounted Multibeam

Higher resolution imagery – Lower positional accuracy, narrow swath.Wide area surveying and image mosaicking?

SBL/USBLtransponder

Dynamically positioned ship (DGPS)

Tethermanagement

system, depressoor weight

AUV

ROV

Position accuracy on sea surface GPS

Source Uncorrected With DifferentialIonosphere 0-30 meters Mostly RemovedTroposphere 0-30 meters All RemovedSignal Noise 0-10 meters All RemovedEphemeris 1-5 meters All RemovedClock Drift 0-1.5 m All RemovedMultipath 0-1 meters Not RemovedSA 0-70 meters All Removed

DGPS much higher level of accuracy than GPS, RTCM corrections.

Real Time Kinematic (RTK) GPS positional accuracies to ~ 10 – 20 cm. Roving receiver limited to within 10km radius of fixed receiver - in shore applications.

Networked RTK GPS for offshore use e.g. Fugro multiple fixed receivers with satellites can extend the range of rover GPS receiver. Sub meter accuracy possible.

Position error for sub sea:INS < 3m / hour drift bounded by 1% slant range USBL - (10 – 15m error at 1,000m depth)

Image resolution of 0.1m?

Position Accuracy Subsea

GAPS USBL – 1% Slant range

Fibre Gyro INS + Kalman FilterDoppler Velocity LogPrecision Depth

Motion Reference Unit

Nav sub systemsDGPS, Ship MRU, USBL(GAPS), INS, DVL, Precision Depth

Positional Accuracy / Resolution Trade Off.

High resolution multibeambathymetry/backscatter mappingROV mounted sonars.Reson 7125 Multibeam

0.1 - 1m seabed feature size typically distinguishable

Low resolution survey

High resolution surveyNarrow swath/field of view

Video mosaic mapping from ROV

Wide swath/field of view

Regional scale multibeam bathymetry/ backscatter mapping (National Seabed Survey)Hull mounted SonarsKongsberg-Simrad EM120, EM1002

0.01 - 0.1 m seabed feature size typically distinguishable

1 - 10m seabed feature size typically distinguishable

Ship mounted SonarsHigh accuracyLow resolution

ROV mounted SonarsLow accuracyHigh resolution

Systems Integration ROV, Ship, Sonar & INS

Mobile & Marine Robotics Research Centre University

ofLimerick

RV Celtic Explorer•GPS1•GAPS USBL

Navigation Equipment

ROV WetBottle

InterconnectWet Bottle

PHINSWet Bottle

RDIDVL Microbath

Precision Depth &Altimeter

SVP

UmbilicalTether

7LCU

7125 Sonar System

Fibre converter MUX

Rs232split DC

PSU

GPS2

Toolskid

Bathysaurus ROV•PHINS•Mobile GPS2•Depth sensor•DVL•USBL transponder

GPS1

GAPS

Technical challenges

GAPS – PHINS Interface (incompatible protocols)

•Send depth sensor data to GAPS

GAPS Holder

GAPS

Interface

GAPSPHINS

Depth sensor

Technical challenges

Depth sensor – PHINS Interface (incompatible protocols)

Depth(Microbath format)

PHINS

Depth sensor

Onboardmicrocontroller

Depth(PAROSCIENTIFIC format)

Hardware solution

Depth(Microbath format)

PHINS

Depth sensor

Softwareon mothership

Depth(PAROSCIENTIFIC format)

Software solution

Technical challenges

DVL (Real-time monitoring & Real-time change of mode of operation)

•Send corresponding configuration scripts during different stages of mission

Software on mothershipReal-time monitoring

Configuration scripts:Water track onlyBottom track onlyWater & Bottom track (combined)DVL

PHINS (Real-time monitoring & Real-time configuration)

•Electrical interface (communication parameters, protocols etc.)

•Mechanical interface (lever arms of sensors, geometric configuration etc.)

Software on mothershipReal-time monitoring

Real-time configuration

PHINS

Preparation

•Interconnection diagrams (two variants)

•Quick Reference Book (collection of key information from different manuals in compact form)

•MATLAB software to determine lever arms and make easy PHINS configuration

•LabView software for GAPS-PHINS-DEPTH interface

•DVL test trials in UL diving pool in order to test sign of DVL outputs

Interconnection Design

Problems during mobilisation / shakedown

Leak in the interconnection bottle•Cause: faulty connector•Responsibility: OceanTools (manufacturer)•Solution: Running repare of faulty connector

GAPS data were not always accepted by PHINS•Cause: wrong setting in GAPS software•Responsibility: IxSea engineer & confusing GAPS software•Solution: Set parameter D to correct value

BOSLevel

LevelSea

DropKeel

GAPS

Holder

SLz

BOSz

DKz

Hz

CGDistance Value Unit TypezSL 1.363 m VariablezBOS 7.262 m FixedzDK 3.000 m VariablezH 0.920 m FixedD = zSL+zBOS+zDK+zH = 12.545 m

Wrong operation of PHINS (instrument rotated by 180°inside enclosure & bugs in firmware)•Cause: error in technical drawing of enclosure•Responsibility: system integrator (RESON)•Solution: Rotation of PHINS and firmware update

Multibeam (7125) Integration. Day before embarking advised to swap sonar.•Cause: New product, DOA•Responsibility: (Reson)•Solution: Panic & swap to mature 8125

Results

•Successful integration GAPS-PHINS-DEPTH for the first time!

•Accurate determination of lever arms (all sensors were accepted by PHINS).

•High accuracy of navigation data.

•Succesful DVL real-time reconfiguration.

•Systematic approach for system integration.

•Discovery of errors in Bloom report.

Identified weak points

•Lack of leak detectors

•Inability to remotely switch on/off individual components

•Full separation between control and navigation

•Difficult navigation with existing displays in severe conditions

Simulation Solutions

Virtual Laboratory (A talk for another day)Modelling of pools, marine environment, vehicles (ship AUVs and

ROVs), navigation & payload sensors with signal level compatibility. Hardware in the loop testing and progressive porting of developed code

to the realtime target.Shift the load of integration and debugging to offline simulation rather

than during mobilisation with daily ship cost.

VR Command Centre for Offshore Ops(Augmented Reality) Video Clip

Real-Time High Resolution Imaging Sonar Simulator

Dr EdinOmerdic

Dr James Riordan

VR Command Centre for Offshore Ops

(Augmented Reality)

Lunar Module calling Heuston Control, Come in HeustonRoger niner, this is Heuston…

Features• Library of Vehicles (Ships, ROVs, AUVs)• Full dynamic models with signal-level compatibility• vR models integrated with real-time sensor input

- augmented reality during ops.• Configuration support tools for new vehicles.• Fault accommodation capabilities• Background seabed map import facility (INSS)• Models of waves, 3 D current models • and more.

VR Command Centre for Offshore Ops

(Augmented Reality)

Real-Time High Resolution Imaging Sonar Simulator

As discussed Imaging the Deep – Many issues affecting the quality of images, maps, 3D generated geometry model

Without draining the ocean it’s impossible to qualify?

Seafloor Terrain Synthesis ModelMulti-Fractal Surfaces – Hybrid Model

Combination of:Perlin noise

synthesisFourier synthesisWavelets

Permits local control over terrain features.

Multiple geologic features can be incorporated in a single terrain environment.

Fractal Dimension varying with altitude

Peaks and outcrops rough, valleys smooth – erosion and deposition

Low lying regions consist of sediment ripple beds – influence of currents

Acoustic Propagation ModelPath of acoustic pulse determined

by ray theory solution to Helmholtz Eq.

Seafloor Scattering Model

Output of Jacksons backscattering model for a typical seafloor configuration

View Dependant Refinementof Fractal Model of Seabed.

Full resolution terrain mesh viewed with adaptive refinement disabled

Multi-resolution representation of the terrain mesh, adaptive refinement enabled.

View Dependant Refinement

Full high-resolution mesh. The sonar position and fan beam footprint are represented by the superimposed

white sphere and blue wedge respectively.

Multi-resolution abstraction of the seafloor mesh. The geometric detail

within the fan beam footprint is preserved while non-contributing facets

are decimated to the maximum possible.

Incremental View Dependant Refinement

Detail information reserved for recovering Mi+1 from Mi.The structure queriedefficiently online duringsimulation to obtain adaptivemeshes on-the-fly byreconstructing the vertexdependency relations.Multiple sonar footprints can be accommodated.Real-time view dependant Level Of Detail

control of terrain mesh

Results – Multi Fractal SS Imagery

Results - Image Generation Performance

Based on Simulator Experimentation

• 4 -5 Hz ping rate while imaging seafloor terrain ~ 2 million facets.

• 40-fold increase in simulation throughput.

• Achieved principally by use of video rendering techniques within sonar simulation process.

• No degradation on integrity of imagery.

• Improved speed and quick results Simulation a more useful tool Allows user interaction on the fly as simulation progresses.

• Performance capability extended to survey scale maps.

Integration of Simulators

•Virtual Laboratory

•vR Command Centre for Offshore Ops

•Real-time High Resolution Imaging Simulator

•Work Underway

R.V. Celtic Explorer

Research Cruise CE0505 June 2005

First Irish-led deep water ROV habitat survey – technical and scientific partnership NUIG/UL

ROV multibeam and video surveys – high resolution mapping and non-destructive observations of habitat/fauna

Use INSS data to identify survey targets and plan surveys– essential data for cost effective survey

Bathysaurus ROV

HEA PRTLI 3 MSR3.2

Margaret Wilson, February 2006

ROV & INSS multibeam