See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/321530255

Automated mine detection, classification and identification with AUV

Conference Paper · June 2013

CITATION READS

1 764

1 author:

Per Espen Hagen

Kongsberg Maritime AS
78 PUBLICATIONS   608 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Synthetic Aperture Sonar View project

Terrain Navigation View project

All content following this page was uploaded by Per Espen Hagen on 05 December 2017.

The user has requested enhancement of the downloaded file.

 Automated mine detection, classification and
identification with AUV

Per Espen Hagen

Kongsberg Maritime
P O Box 111, NO-3191 Horten, Norway
Per.Espen.Hagen@kongsberg.com
http://www.km.kongsberg.com/auv

ABSTRACT

Autonomous underwater vehicle (AUV) technology is currently gaining acceptance as an

essential tool in mine hunting. Today’s AUVs perform detection and classification in a
single mission, followed by post-mission analysis before the identification and disposal
phase. Recent technological advances now allow the AUV to perform identification during
the same mission, substantially speeding up the efficiency of the mine hunting operation.
Such a system was demonstrated in an experiment led by NATO Centre for Maritime
Research and Experimentation (CMRE) in Italy last year, using a HUGIN AUV with a
HISAS 1030 synthetic aperture sonar and an optical camera.

This solution has required integration of a number of sophisticated technologies: Synthetic

aperture sonar to allow detection and classification over a very wide swath; real-time SAS
processing and automated target recognition (ATR) in order to automatically locate
possible mines; automated in-mission re-planning and high-quality optical imaging in order
to record images for identification.

The paper will discuss some of the technical challenges involved and the operational
consequences of such a system.
Introduction

Around the turn of the century, a few advanced navies started operational evaluation of AUVs as a
tool for mine countermeasures (MCM). It was quickly established that AUV technology offered
solutions to many pressing challenges in the MCM community; some of the most important being
reduced risk to personnel and own ships, low cost, high speed, capability for mine hunting in deep
water, deployability, and the possibility of doing covert or low-visibility MCM.

As a result, AUV use rapidly progressed from technology evaluation to regular “in service” use:
REMUS 100 AUVs were famously used in the Persian Gulf harbour of Umm Qasr during the US-
led Operation Iraqi Freedom in 2003. HUGIN 1000 AUVs have been operated in NATO exercises
and participated in NATO’s standing MCM forces since early 2004.

Left: REMUS 100 shortly after launch. Right: HUGIN 1000 during launch from an MCM vessel in
2004.

The AUV is now established as a tool of choice for detection and classification of mines, and
several navies have acquired or are in the process of acquiring fleets of AUVs for MCM vessels
(MCMVs) or clearance diver units. One example is the Finnish Navy, who equip each of their new
Katanpää-class MCMVs with HUGIN 1000 and REMUS 100 AUVs.

AUVs in mine hunting

A mine hunting operation consists of four phases: Detection/search, classification, identification,
and disposal/neutralisation. In a traditional mine hunting operation, a dedicated MCMV performs all
four phases in parallel: A detection sonar is used to search for mine-like echoes (MILECs). Once a
MILEC is detected and within classification range, a (usually separate) classification sonar is used
to classify the object as a mine-like contact (MILCO) or non-MILCO. If the object is classified as a
MILCO, the MCMV stops and dispatches an ROV or divers for visual identification. If the object is
identified as a mine, neutralisation follows immediately. Then the process returns to phase 1.

The introduction of AUVs to the MCM force has necessitated development of new concepts of
operations. The currently prevailing concept is to pre-program the AUV with a search pattern
covering the entire area of operations, using a sonar that provides classification quality data. For
smaller areas in very shallow water, this is typically a high-frequency side scan sonar. In shallow
and deep water, synthetic aperture sonar (SAS) can be used to achieve much higher area
coverage rates. Analysis is performed by human operators after recovery, and a decision is made
on what targets to inspect and potentially neutralise. In many cases, the options for identification
are limited to an ROV; an expendable mine disposal vehicle (EMDV); or clearance divers. In
cluttered areas where one can expect a high number of non-mine mine-like bottom objects
(NOMBOs), this can be a very inefficient operation.

Example responses from different cylindrical mines with HISAS 1030 on HUGIN 1000 (10x10 m
cut-outs from full-swath sonar images). Range to target 43 m, 80 m, and 130 m, respectively.

Using an AUV also for identification offers a number of advantages. The AUV can be sent out on a
new mission using an optical camera (and potentially also other sensors, such as a high-resolution
multibeam echo sounder, sniffers or magnetic sensors), passing the most likely contacts at low
altitude. AUV users including the Royal Norwegian Navy (RNoN) have experimented with this
concept in recent years, using a HUGIN 1000-MR AUV with a HISAS 1030 synthetic aperture
sonar for detection and classification, and a high performance still image camera system for
identification.

Operational experience has proven the AUV to be an efficient tool also for mine identification.
However, with today’s solutions, a considerable time delay is introduced by the need to recover the
AUV, process and analyse the SAS data recorded, select which targets to identify, plan the ID
mission, launch the AUV and transit back to the area of operations.

The natural next step is to make the AUV perform both tasks not in two, but in one mission.
Especially for a SAS-equipped system, this requires substantial additional on-board processing
capabilities:
• Real-time SAS processing
• Real-time automated target recognition (ATR)
• Automated in-mission re-planning for EOID

Kongsberg Maritime and the Norwegian Defence Research Establishment (FFI) have been working
together over the last several years on the technologies required to accomplish this. Last year, a
HUGIN 1000 AUV was upgraded with prototype versions of these solutions.

The real-time SAS processing is based on the same software as Kongsberg’s existing FOCUS
post-processing; the real-time ATR is similarly based on Kongsberg’s existing Sitar ATR product.
Adaptations have been made to achieve real-time performance on hardware suitable for an AUV,
and to make all components work seamlessly together in an in-mission environment. The
automated re-planning is part of the next-generation autonomy subsystem for HUGIN AUVs,
currently under development at Kongsberg Maritime and FFI.

The upgraded HISAS system also computes a reliable sonar performance measure in real time,
which can be used to automatically change line spacing in lawnmower patterns to ensure
maximum coverage without gaps.
Demonstrations
In October 2012, FFI with support from the RNoN and Kongsberg Maritime participated in an
experiment led by the NATO Centre for Maritime Research and Experimentation (CMRE) near
Elba, Italy. The prototype system mentioned above was operated from CMRE’s research vessel
NRV Alliance during the Autonomous Reactive Intelligence Sea Experiment (ARISE) 2012.

CMRE (formerly NURC and SACLANTCEN) is a research institute focusing on developing and
evaluating technologies for use by NATO nations in the marine environment. The ARISE 12
experiment was aimed at verifying and evaluating advanced autonomy concepts for AUVs under
realistic mine hunting conditions, and included a number of tests where AUVs needed to
dynamically re-plan missions and adapt to the local environment.

HUGIN AUV launched from NRV Alliance

In a single mission, a pre-defined search pattern was executed by the AUV in an area where
various exercise targets had been placed. The recorded HISAS data was processed during the
mission and the output fed to the SITAR ATR system. From the automatically detected and
classified targets, an extended mission plan for electro-optical identification (EOID) was computed.
The vehicle then surfaced and sent the list of targets and the new mission plan to the operator,
allowing the operator to adjust the EOID plan. After the plan was accepted by the operator, the
vehicle dove, turned on the camera system and recorded a series of still images of the targets.
When the mission was completed, all the data including SAS imagery and camera images were
instantly available for download from the AUV. The data was then played back in Kongsberg’s
Reflection PMA software package.
 SAS mosaic of the search survey with optical image of one of the detected targets

Consequences
With the above functions in the vehicle, the AUV can perform detection, classification and
identification during a single mission. This functionality provides substantial time savings in mine
countermeasure (MCM) missions and facilitates in-stride AUV based MCM operations.

In many cases, the AUV will not be able to capture images of every single potential mine-like
contact. This can be due to e.g. time constraints, imperfect sonar imagery, limitations in the ATR
performance, navigation uncertainty or other operational issues. However, even in such cases,
having optical imagery of a number of sonar contacts is very valuable. Interpreting sonar images –
even very high resolution SAS images – of previously unseen objects can be difficult. Having
“ground truth” in the form of optical images available for a number of contacts helps the operators
understand the environment.

About the Author

Per Espen Hagen received his MSc in Signal Processing from the Norwegian Institute of
Technology. In 1990, Hagen was employed as a scientist at the Norwegian Defence Research
Establishment (FFI), initially working on autonomous missile navigation and guidance problems. In
1994, he joined FFI’s AUV research group, working in areas including sonar image analysis, non-
traditional navigation, operator interfaces, and synthetic aperture sonar. From 1999 he served as
project manager for the AUV related projects at FFI. In 2008 Hagen joined Kongsberg Maritime,
where he is in charge of System Architecture at the AUV R&D Department.

View publication stats