Autonomous underwater mappingusing terrain- and sonar navigation

Ove Kent HagenForsvarets forskningsinstitutt, FFI

Geodesi- og hydrografidagene 2014Sola, 13. november 2014

Outline

• HUGIN AUV and underwater mapping– Underwater navigation– Mapping sensors

• Data-driven navigation techniques– Terrain navigation– Sonar navigation

• Autonomous mapping concepts– Autonomous survey– Autonomous mine hunting

• Plans

AUV research at FFI

Batteries

SAS

Autonomy ATRNavigation

Camera

Underwater mapping

Global referenced vehicle position and orientation+

Sensor data referenced in vehicle reference frame– Sensor processing: acoustic signals to bathymetry– Lever arms and reference systems alignments– Time synchronization

= Global referenced sensor data

Ormen LangeNorsk Hydro

Sigsbee, Gulf of MexicoC&C Technologies

Sensors near sea floor- High resolution- Increased stability

< 800 m < 2400 m

HUGIN AUVKongsberg Maritime

NavPReal-time system on HUGIN AUVs

NavLab (commercially available)Post-processing, simulation, analysis and development

HUGIN Navigation System

AUV Navigation

Problem: GPS does not work under water!

Solution:- Inertial navigation system (INS)- Aided by any available sensor

State-of-the-art DVL-aided INS for AUVs

AUV trajectory

Elliptic Position Uncertainty

Honeywell HG9900• 0.8 nm/h• 0.003 °/h

Doppler Velocity Log (DVL) Inertial Measurement Unit (IMU)

WHN 600• 0.2% • ±1mm/s

- Orientation found by processing accurate accelerometers and gyros - Position drift:

~ 0.1% travelled distance, straight line~ 0.01% travelled distance, lawn mower pattern, 1km lines

IMU

+DV

L

DVL

Navigation post-processing (NavLab): Simulated Survey

Trajectory

Estimated postionaccuracy (1σ)

Green Real-time pos acc (1σ)Red Post proc pos acc (1σ)

GPS-fix

AUV navigation autonomy levels

• Supervised– AUV may be commanded– Requires infrastructure for positioning– E.g.: Mother ship with GPS and acoustic tracking

of AUV (USBL)

• Semi-autonomous– AUV operates autonomously– Requires infrastructure for positioning– E.g.: GPS-fix, GPS-buoy transponder navigation,

sea floor transponder navigation, …

• Autonomous– AUV operates autonomously– No infrastructure for positioning– E.g.: terrain navigation, sonar navigation, …

GPS

USBL

Transpondernavigation

TerrainNavigation

Mapping sensors

EM 3000Multibeam echosounder

HISASSynthetic Aperture Sonar Combined

SAS imaging and bathymetry

Courtesy Royal Norwegian Navy

Courtesy Royal Norwegian Navy

Resolution > 2 cm Resolution ~ 6 cm

SAS Image SAS Bathymetry

SAS image fused with bathymetry

Holmengraa sunk 1944 outside Horten

Terrain navigation with loose INS coupling• Search area based on INS estimate• Correlate bathymetric

measurements with an a priori known digital terrain model (DTM)

• Find best position estimate within search area

• Send position estimate and accuracy to the INS

HUGIN Terrain Navigation SystemCommercially available for HUGIN AUVs

TerrLab: simulation, development, analysis and post-processingTerrP : real-time system

Point Mass Filter

Hagen & Ånonsen, MTS/IEEE OCEANS 2010, Seattle

HUGIN AUV, Barents Sea, 2009Test of terrain navigation during long covert transit in open sea

• DTM from EM710 with 10 m resolution

• Straight line transit ~50 km or ~27 nm

• HUGIN navigated autonomously, monitored on USBL

• ~ 4 m difference at end of transit

East [km]

Nor

th [k

m]

-18 -16 -14 -12 -10 -8 -6 -4 -2 0

-45

-40

-35

-30

-25

-20

-15

-10

-5

0

5

300

310

320

330

340

350

360

370

380

390

East [m]

Nor

th [m

]

-5 0 5 10

-15

-10

-5

0

5

10

HUGIN INS

USBL

HUGIN AUV, Oslo fjord, 2010Test of covert transit in fjord

10.5 10.55 10.6 10.65 10.7

59.18

59.2

59.22

59.24

59.26

59.28

59.3

59.32

59.34

59.36

0

5

10

100

350

Bolærne

←Åsgårdstrand

• DTM created from EM710: 10 m resolution

• TerrP used only DVL (4 beams)

• HUGIN navigated autonomously for ~5 hours (about 36 km), monitored on USBL

• The difference between HUGIN’s GPS and the INS after resurfacing was ~ 5 m

Ånonsen & Hagen, MTS/IEEE OCEANS 2010, Seattle

East

Nor

th

-155

-150

-145

-140

-135

1.42 1.43 1.44 1.45 1.46 1.47 1.48 1.49 1.5 1.51

x 104

0

100

200

300

400

500

600

700

Eastings [m]

Nor

thin

gs [m

]

Navigation solution and TerrNav position fixes, meters

NavPNavLabTerrP RQTerrP OK

Mappingmode

Navigation mode

HUGIN AUV, Larvik, 2011Test of local covert terrain navigation

• Map built in-mission at start• Used for navigation upon return

East

Nor

th

-160

-155

-150

-145

-140

-135

EM 710H.U. Sverdrup II

EM2000HUGINin-mission

Ånonsen & Hagen, MTS/IEEE OCEANS 2011, Kona

H.U. Sverdrup II, West Coast fjords, 2013Test of terrain navigation against simulated GPS jamming/spoofing

Hagen & Ånonsen, MTS Journal, vol 48(2), 2014

Line-by-line terrain navigation• Use of sparse maps - search area only partially covered• Estimate position difference along the line• Simulation results (conservative)

– Side scan bathymetry (~1 m res) with SAS DTM (~ 0.18 m res)

10.426 10.427 10.428 10.429 10.43 10.431 10.432 10.433 10.434 10.43559.4675

59.468

59.4685

59.469

59.4695

59.47

59.4705

59.471

59.4715

0 20 40 60 80 100 120 140 1600

2

4

6

8

10

12

14

Time [s]

RM

S S

tdD

ev[m

]

20% overlap30% overlap50% overlapSAS Line DTM SSB measurements

Overlap

Sonar ATR navigation

Automatic target recognition (ATR) – association - invariant to poseEstimates position difference

0 0.5 1 1.5 2 2.5 3 3.5-10

-5

0

5

Time [h]

y [m

]

DVLDVL+FBNFeature Observations

0 0.5 1 1.5 2 2.5 3 3.5-10

-5

0

5

Time [h]

Horizontal position error and std (1σ)

x [m

]

DVLDVL+FBNFeature Observations

Feature-based sonar navigation

Feature tracker, similar pose, line-by-lineEstimates position and heading difference

Resolution 4 cm x 4 cm

Autonomous mapping: survey

Supervised mappingwith GPS+USBL

Cross-line autonomous navigation and mapping: - DVL-aided INS- Terrain navigation- Sonar ATR navigation Supervised mapping

with GPS+USBL

Line-by-line autonomous navigation and mapping:- DVL-aided INS- Terrain navigation- Sonar navigation

• Accurate real-time navigation required to ensure coverage • Navigation accuracy requirements to be met by post-processing• Navigation post-processing inherently increases mapping data consistency

Autonomous mapping: mine hunting

Transit 1

DVL-aided INS

Line-by-line terrain- and sonar navigation

Detection

Transit 2

DVL-aided INS

Independent global position error

Detect & Classifyby SAS image

Identify byoptical image

FFI Plans 2014/2015

• Finish development of line-by-line terrain navigation delta measurements– SSB measurements vs SSB DTM line– SSB measurements vs SAS line DTM– SAS line bathymetry vs SAS line DTM

• Finish development of line-by-line feature tracker based sonar navigation delta measurements

• Integrate the delta measurements with NavLab4, navigation post-processing

• Test on dedicated data set collected in Barents Sea 2014

Can also be used in real-time

Questions?http://www.ffi.no/no/Forskningen/Avdeling-Maritime-systemer/hugin