Alternatives for Landmine Detection

RAND 11

Science & Technology Policy Science & Technology Policy InstituteInstitute

Alternatives for Landmine DetectionAlternatives for Landmine Detection

Jacqueline MacDonald

J.R. Lockwood

November 14, 2002

RAND 22


Briefing OutlineBriefing Outline

1. Origins of this project

2. Project tasks and study method

3. Background on the scope of the landmine problem

4. Limitations of conventional mine detection technologies

5. Alternative mine detection technologies: capabilities and limitations

6. Recommendations for developing an advanced mine detection system

RAND 33


Project OriginsProject Origins

October 2001 memo from Gary Ellis to Helga Rippen, director of the Science and Technology Policy Institute (STPI):

“With the number of buried mines exceeding 50 million worldwide, OSTP seeks an update on knowledge of how to clear these hazards. What new technologies are available? In specific, what is the best area of research and development to support, to make an order-of-magnitude improvement in mine-clearing?”

RAND 44


Project TasksProject Tasks

1. Identify antipersonnel mine detection technologies currently in the R&D stage.

2. Evaluate the potential for each to improve the reliability and safety, increase the speed, and decrease the costs of demining.

3. Identify any barriers to completing development of new technologies.

4. Recommend options for federal investments to speed development of key technologies.

5. Provide information on funding requirements to complete development of the new methods.

RAND 55


Study MethodStudy Method

Review literature on landmine detection technologies

Identify leaders in the mine detection field

Appoint Landmine Detection Task Force

Identify innovative technologies (literature search and task force interviews)

Identify two lead researchers on each technology

Ask researchers to submit papers describing the potential of each technology (received 23 papers)

RAND 66


Study Method (continued)Study Method (continued)

Meet with task force to review capabilities and limitations of each technology (two-day meeting held in May 2002)

Work with task force to refine evaluations and recommendations

Submit report to task force members for review

Submit report to additional peer reviewers who were not task force members

RAND 77


Landmine Detection Task ForceLandmine Detection Task Force

Dr. John McFee (chair) Canadian Centre for Mine Action Technology

Dr. Tom Altshuler DARPA

Dr. Tom Broach Army Night Vision and Electronic Sensors Directorate

Dr. Larry Carin Duke University

Dr. Russell Harmon Army Research Office

Dr. Cary Rappaport Northeastern University

Dr. Waymond Scott Georgia Tech

Mr. Richard Weaver Army Night Vision and Electronic Sensors Directorate

RAND 88









RAND 99


Scope of the Landmine ProblemScope of the Landmine Problem

45-50 million mines worldwide

100,000 mines cleared each year => 450-500 years to clear all existing mines

1 million new mines emplaced annually => 19 years of additional mine clearance time added each year

15,000-20,000 victims each year in 90 countries

RAND 1010


American Mine VictimsAmerican Mine Victims

Fred Downs: lost left Fred Downs: lost left arm to mine in Vietnamarm to mine in Vietnam

Robert Washburn: lost leg to Robert Washburn: lost leg to mine in Bosniamine in Bosnia

RAND 1111


American Victims (continued)American Victims (continued)

Marianne Holtz: civilian Marianne Holtz: civilian nurse, lost both legs to a nurse, lost both legs to a mine in Zairemine in Zaire

Jerry White: lost a leg Jerry White: lost a leg while a student on a while a student on a backpacking trip in Israelbackpacking trip in Israel

RAND 1212


Case Study: AfghanistanCase Study: Afghanistan

One of the world’s most heavily mined countries

More than 11 percent of land is mined

150-300 mine victims per month, half of them children

17 in 1,000 children injured or killed by mines

Most mines left from Soviet occupation; some emplaced during civil wars that followed

Mine presence interfering with restoration of stability

RAND 1313


Mine Victims, AfghanistanMine Victims, Afghanistan

Mine victims at Kabul limb-fitting center.

RAND 1414


Blast MineBlast Mine

Blast mines cause the affected object (e.g., foot) to blast upward into fragments

RAND 1515


Fragmentation MineFragmentation Mine

Fragmentation mines throw fragments radially outward and can cause casualties at large distance (100 m)

RAND 1616









RAND 1717


Reliable, Safe, Efficient Reliable, Safe, Efficient Detection Methods Are LackingDetection Methods Are Lacking

Detection technologies have advanced little since World War II:

“Today, highly trained, scared soldiers use all of their senses, augmented with a coin detector and a pointed stick.”

Col. Robert Greenwalt

RAND 1818


Mine Detection ProcessMine Detection Process

Divide mined area into grids (e.g., 100 m2)

Split grid into 1-m-wide lanes

Slowly traverse each lane while swinging a metal detector low to the ground

Investigate each item signaled by the metal detector, using pointed stick

Variations: mechanical flails, mine-sniffing dogs

RAND 1919


““Highly trained, scared soldiers Highly trained, scared soldiers augmented with a coin detector …”augmented with a coin detector …”

Mozambiquan Mozambiquan deminer, deminer, KosovoKosovo

RAND 2020


“… “… and a stick” and a stick”

Deminer probing a detected object to determine whether it is a mine

RAND 2121


Dogs sometimes lend a handDogs sometimes lend a hand

RAND 2222


Metal Detector ConceptsMetal Detector Concepts

Operate via “electromagnetic induction” (EMI)

Electric current from detector coil creates magnetic field in ground

Induces an electric current in buried metal

Current in buried metal creates secondary magnetic field

Receiver coil detects voltage change

Detector converts voltage change to audible signal

RAND 2323


EMI LimitationsEMI Limitations

1. Imperfect probability of detection: Not all mines are detected

2. High false-alarm rate:

- 100-1,000 inert metal objects excavated for every mine

- Increases deminer fatigue and likelihood of carelessness

- Trade-off between false-alarm rate and probability of detection

RAND 2424


EMI Limitations (continued)EMI Limitations (continued)

3. Slow:

- Deminer needs 5-20 minutes to investigate each declaration, whether scrap or clutter

- Most of deminer’s time is spent investigating false alarms

4. Dangerous:

- Deminers must work close to mines; must excavate or prod to confirm mine presence

- 1 deminer killed for every 1,000-2,000 mines cleared

RAND 2525


RAND 2626


RAND 2727


False Alarms Make Mine Detection False Alarms Make Mine Detection Extremely SlowExtremely Slow

Type of Item Number Found

Time Spent (hours)

Percentage of Total Time

Antitank mines 961 240 0.0074

Antipersonnel mines

89,327 22,000 0.068

Unexploded ordnance

452,770 110,000 0.34

Scrap 191,737,707 32,000,000 99.6

Mine Clearance Data from Cambodia, 1992-1998

RAND 2828


RAND 2929









RAND 3030


Electromagnetic MethodsElectromagnetic Methods

Technology Principle Limitations

Ground-penetrating radar (GPR)

Reflects radio waves off mine/soil interface

Roots, rocks, water pockets; extremely moist or dry environments

Electrical impedance tomography

Determines electrical conductivity distribution

Dry environments; can detonate mine

X-ray backscatter

Images buried objects with x-rays

Slow; emits radiation

Infrared/ hyper-spectral

Assesses temperature, light reflectance differences

Cannot locate individual mines

RAND 3131


Low-metal mines

Metal mine

Landmine Images from GPRLandmine Images from GPR

RAND 3232


Acoustic/Seismic MethodsAcoustic/Seismic Methods


Acoustic/ seismic

Reflects sound or seismic waves off mines

Deep mines; vegetation cover; frozen ground

RAND 3333


Explosive Vapor Detection MethodsExplosive Vapor Detection Methods


Biological (dogs, bees, bacteria)

Living organisms detect explosive vapors

Dry environments; environmental confounders

Fluorescent Measure changes in polymer fluorescence in presence of explosive vapors

Dry environments

Piezoelectric Measure shift in resonant frequency of various materials upon exposure to explosive vapors

Dry environments

Spectroscopic Analyze spectral response of sample

Dry environments

RAND 3434


Bacteria Fluorescing in the Presence of Bacteria Fluorescing in the Presence of TNTTNT

RAND 3535


Prototype Fluorescent Polymer SystemPrototype Fluorescent Polymer System

RAND 3636


Bulk Explosives Detection MethodsBulk Explosives Detection Methods


Nuclear quadrupole resonance (NQR)

Induces radio frequency pulse that causes chemical bonds in explosives to resonate

TNT; liquid explosives; radio-frequency interference; quartz-bearing and magnetic soils

Neutron Induces radiation emissions from the atomic nuclei in explosives

Not specific to explosives molecule; moist soil; ground surface fluctuations

RAND 3737


Prototype NQR SystemPrototype NQR System

RAND 3838


Summary of Innovative Detection Summary of Innovative Detection Technology PotentialTechnology Potential

Unlikely to Yield Major Gains or Not Suitable

Basic Research Needed

Promising Established

Electrical impedance tomography

X-ray backscatter

Infrared/ hyperspectral

Neutron

Biological (bacteria, bees)

Fluorescent

Electrochemical

Piezoelectric

Spectroscopic

Acoustic/ seismic

NQR

EMI

GPR

RAND 3939


ConclusionsConclusions

Some individual technologies warrant further research

However, no single mine sensor can detect all mine types in all environments

All sensors are limited by false alarms (specific to the sensor type) and/or environmental interference

RAND 4040









RAND 4141


Multi-Sensor Approach Is NeededMulti-Sensor Approach Is Needed

Multi-sensor system would overcome limitations of single sensors:

Multiple sensors with different false alarm sources would decrease false alarm rate

Multiple sensors with different environmental confounders would increase probability of detection

Advanced signal processing and/or decision algorithms would optimize operator decisions about whether or not item is a mine

Design from first principles is needed

RAND 4242


Multi-Sensor System Would Exploit Multi-Sensor System Would Exploit Different False Alarm SourcesDifferent False Alarm Sources

Detection Technology Primary Source of False Alarms

EMI Metal scrap, natural soil conductivity and magnetism variation

GPR Natural clutter (roots, rocks, water pockets, etc.)

Acoustic/seismic Hollow, man-made objects (e.g., soda cans)

Fluorescent polymers Explosive residues

NQR Radio frequency interference

RAND 4343


Army Dual-Sensor System: Hand-Held Army Dual-Sensor System: Hand-Held Standoff Mine Detection System Standoff Mine Detection System

Combines GPR and EMI

Production scheduled for 2004

Does not represent the type of advanced muti-sensor system we envision:

Relies on established electromagnetic sensors

Does not use innovative methods for detecting explosives or acoustic properties

Does not use advanced signal processing or multi-sensor decision algorithms; operator receives two distinct signals

RAND 4444


Prototype HSTAMIDS SystemPrototype HSTAMIDS System

RAND 4545


U.S. Funding Is Not Optimized for Multi-U.S. Funding Is Not Optimized for Multi-Sensor System DevelopmentSensor System Development

Total 2002 funding for humanitarian mine detection R&D: $13.5 million

Only $4.9 million of this was allocated for detection technologies

Nearly half the $4.9 million was allocated for wide-area (not close-in) detection

Total available for close-in detection: $2.7 million

Most of the $2.7 million was focused on established technologies

RAND 4646


Distribution of U.S. Funds for Distribution of U.S. Funds for Humanitarian Mine Detection R&DHumanitarian Mine Detection R&D

0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2

Hand-held detection(EMI, GPR)

Explosive vapordetection

Radars for wide-area detection

Infrared sensors forwide-area detection

Funding ($ millions)

RAND 4747


How Much Faster Could We Clear Mines How Much Faster Could We Clear Mines with a Multi-Sensor System?with a Multi-Sensor System?

At project outset, we were asked to evaluate whether order-of-magnitude decreases in time is possible

Across-the-board order-of-magnitude decrease in time is not possible in foreseeable future:

Vegetation, trip-wire clearance are time consuming

Thus even a perfect detector could not cut clearance time by a factor of 10

Current research predicts 60-300% decrease in clearance rates with elimination of 99% of false alarms

RAND 4848


Benefits of Multi-Sensor SystemBenefits of Multi-Sensor System

Savings of billions to tens of billions of dollars in world-wide cost of mine clearance:

• Estimated total cost to clear all mines is $14-50 billion

• Most of cost is personnel cost

• Thus time savings translate almost directly into cost savings

Improvement in probability of detection

Improvement in demining safety

Spin-off benefits

RAND 4949


What Would It Take to Develop an What Would It Take to Develop an Advanced, Multi-Sensor Detector?Advanced, Multi-Sensor Detector?

R&D Stage HSTAMIDS Time

HSTAMIDS Cost

Predicted Advanced System Time

Predicted Advanced System Cost

Basic 4 years $5 million 5-8 years $50 million

Prototype 2 years $8 million 2 years $10 million

Demonstration 5 years $33 million 5 years $40 million

Manufacturing 4 years $27 million 4 years $35 million

RAND 5050


5-Year Research Plan for 5-Year Research Plan for Multi-sensor SystemMulti-sensor System

Research Area Researcher Years

Cost Results

Algorithms for sensor fusion

40 $10 million

Minimal set of sensor-level fusion algorithms for specific sensor suite

Integration of component sensors

25 $6.2 million

Multi-sensor prototype detector with three to four sensor technologies

Explosives detection

50 $12 million

Set of sensors suitable for use in multi-sensor prototype

Environment effects on sensors

10 $2.5 million

Simple tests that can be performed to improve or predict sensor performance

RAND 5151


SummarySummary

Antipersonnel mines are a significant problem with effects on U.S. interests world-wide

Existing mine detection technologies are primitive: coin detectors and sticks

A number of promising mine sensors are in the R&D stage

No single sensor can overcome all sources of false alarms and find all mine types in all environments

An advanced, multi-sensor system is needed

Current U.S. R&D would need to be refocused to develop such a system

Alternatives for Landmine Detection

Documents

Transcript of Alternatives for Landmine Detection