detection of obscured targetspeople.ee.duke.edu/~lcarin/scott1.pdf · 2003. 1. 28. · – the...
TRANSCRIPT
Detection of Obscured Targets
Waymond R. Scott, Jr. and James McclellanSchool of Electrical and Computer Engineering
Georgia Institute of TechnologyAtlanta, GA 30332-0250
MURI Kickoff/Progress 1-24-03 Scott and Mcclellan, Georgia Tech 2
Outline
• Objectives• Sensor Systems
– Buried Structures– Buried Landmines
• Material Parameter Measurements• Near and Far Term Goals
MURI Kickoff/Progress 1-24-03 Scott and Mcclellan, Georgia Tech 3
Objectives for the Georgia Tech Effort on the Obscured Targets MURI
The objective of this research is to use a combination of theoretical simulation, experimental measurements, and signal processing to develop and understand innovative techniques for detecting obscured targets such as buried landmines and buried structures.
MURI Kickoff/Progress 1-24-03 Scott and Mcclellan, Georgia Tech 4
The Components of the Research are all Interrelated
Sensor System:Mine Detector,Buried StructureDetector, etc.
Probing Signals:Electromagnetic,Seismic, HybridPassive/Active.
Models: Theoretical,Large scale Numerical and Experimental
Underlying Physics:Wave Interactions,Material Properties,etc.
Signal Processing:Detection, Inversion, etc.
MURI Kickoff/Progress 1-24-03 Scott and Mcclellan, Georgia Tech 5
Outline
• Objectives• Sensor Systems
– Buried Structures– Buried Landmines
• Material Parameter Measurements• Near and Far Term Goals
MURI Kickoff/Progress 1-24-03 Scott and Mcclellan, Georgia Tech 6
Sensor Systems
• Buried Structure Detection– Is it feasible to use either active or passive seismic
techniques and/or electromagnetic techniques to detect buried structures?
– Many of the issues are similar to those for mine detection.
• Soil properties• Seismic/Electromagnetic wave interactions• Configuration• Signal processing• Ambient/target noise.
– Numerical and experimental models are also similar
MURI Kickoff/Progress 1-24-03 Scott and Mcclellan, Georgia Tech 7
Sensor Systems• Possible configurations for a sensor to detect buried structures.
– These can be independent or interdependent sensors.
Buried Structure
NoiseSource
AirSoil
Seismic wavesRadiated from Structure
Sensors to detect waves radiated from structure
Passive Seismic Active EM to Sense Vibrations
Buried Structure
NoiseSource
Air
Structure Vibrating due to Noise
Soil
VibrationSensing Radar
Electromagnetic Waves
MURI Kickoff/Progress 1-24-03 Scott and Mcclellan, Georgia Tech 8
Buried Structure: 2m X 3m X 3m room with 14cm thick concrete walls :
Internal Source
MURI Kickoff/Progress 1-24-03 Scott and Mcclellan, Georgia Tech 9
Buried Structure: 2m X 3m X 3m room with 14cm thick concrete walls :
External Source
MURI Kickoff/Progress 1-24-03 Scott and Mcclellan, Georgia Tech 10
Models• Large scale numerical and experimental models
will be developed for these systems.– Extensions of the models developed under Demining
MURI and ONR projects.– Used to develop an understanding of the underlying
physics.• Wave interactions• Material parameters
– Used to generate synthetics and test ideas• Robust signal processing algorithms• Physical theories• Measurement Configurations
MURI Kickoff/Progress 1-24-03 Scott and Mcclellan, Georgia Tech 11
Outline
• Objectives• Sensor Systems
– Buried Structures– Buried Landmines
• Material Parameter Measurements• Near and Far Term Goals
MURI Kickoff/Progress 1-24-03 Scott and Mcclellan, Georgia Tech 12
Sensor Systems• Mine Detection
– Extension of Seismic/Electromagnetic Sensor developed as part ofthe Demining MURI and ONR projects
– Improve signal processing• In situ characterization of the subsurface velocity profile• Better mine detection algorithms
– Improve agreement between and experimental and numerical models
• Better understand/measure elastic properties of the soil• Better understand/measure seismic wave interactions with mine
– How is the best way to configure such a system?– How is the best way to sense the seismic vibrations?– Can ambient seismic noise be used to detect mines?
MURI Kickoff/Progress 1-24-03 Scott and Mcclellan, Georgia Tech 13
Possible Configurations
S SN
Mine
Rayleigh Wave
Elastic Wave Source
Displacements
AirSoil
Sensor
MURI Kickoff/Progress 1-24-03 Scott and Mcclellan, Georgia Tech 14
Support FrameSeismic Source
Elastic Wave
Sensor
Possible Configurations
MURI Kickoff/Progress 1-24-03 Scott and Mcclellan, Georgia Tech 15
Photograph of the Experimental Model
MURI Kickoff/Progress 1-24-03 Scott and Mcclellan, Georgia Tech 16
Photograph of the Uncovered Mines and Rocks.
MURI Kickoff/Progress 1-24-03 Scott and Mcclellan, Georgia Tech 17
Single AT Mine Surrounded by Multiple AP Mines30 dB Scale
Experimental Model Numerical Model
The differences between these results are due to the inaccurate values of the material parameters used in the model.
MURI Kickoff/Progress 1-24-03 Scott and Mcclellan, Georgia Tech 18
Speed?• One of the most significant issues that must be overcome
to make a practical seismic mine detection system is measurement speed.
• We have been using a 4 second measurement time to maximize the signal to noise ratio in our laboratory measurements. – This is overkill for a practical system
• Lower signal to noise ratios are adequate to find mines.• A mine field will probably be much less noisy than our lab.• Real soils will be more linear than the sand in the laboratory.
• What are reasonable measurement times?– Data from an experiment at a US Government test facility– Synthesize the effects of shorter measurement times.
MURI Kickoff/Progress 1-24-03 Scott and Mcclellan, Georgia Tech 19
US Government Test Facility
• VS2.2 AT mine 1 inch deep– 24 cm Diameter by
11.5 cm Height– Plastic
MURI Kickoff/Progress 1-24-03 Scott and Mcclellan, Georgia Tech 20
Surface Displacement over MineVersus Measurement Time
4s
1s
1/4 s
1/16 s
MURI Kickoff/Progress 1-24-03 Scott and Mcclellan, Georgia Tech 21
Images: VS2.2 AT Mine: 1 inch deep30 dB Scale: Versus Measurement Time
4s
1/4s
1s
1/16s
MURI Kickoff/Progress 1-24-03 Scott and Mcclellan, Georgia Tech 22
Possible Handheld Configuration
• Stationary seismic source• Hand scanned sensor• Audible presentation of
seismic waves
MURI Kickoff/Progress 1-24-03 Scott and Mcclellan, Georgia Tech 23
Audible Presentation• The above images will be difficult to generate
with a hand held mine detector.• An audible presentation of the signals are easy to
generate and require essentially no signal processing.– The signal sensed by the radar is directly played to the
operator. – The incident signal can be clearly heard by the operator.
This gives him confidence that the incident signal is present.
– The mine signal sounds hollow and is clearly distinguishable from the incident signal
MURI Kickoff/Progress 1-24-03 Scott and Mcclellan, Georgia Tech 24
5 10 15 20 25
-60
-40
-20
0
20
40
60
disp
lace
men
t, y=
0
time (ms)
TS-50 Mine, 3.0 cm Deep, SandboxVisual Presentation; Waterfall Graph Audible Presentation;
Sound File
MURI Kickoff/Progress 1-24-03 Scott and Mcclellan, Georgia Tech 25
Outline
• Objectives• Sensor Systems
– Buried Structures– Buried Landmines
• Material Parameter Measurements• Near and Far Term Goals
MURI Kickoff/Progress 1-24-03 Scott and Mcclellan, Georgia Tech 26
Material Parameters
• Soil has very inhomogeneous and complex mechanical and electromagnetic properties
• These inhomogeneities and complexities are generally the limiting factor for subsurface sensing systems
• Techniques for measuring these properties in situ will be investigated – In situ measurements are necessary because disturbing
the soil significantly changes its material properties
MURI Kickoff/Progress 1-24-03 Scott and Mcclellan, Georgia Tech 27
Material Parameters• Spectral Analysis of Surface Wave (SASW) techniques are
used by geophysicist and civil engineers to make in situ measurements of the mechanical properties.– However, they are generally interested in much deeper structures.– We have found that the complexities of the near surface cause
problems for these techniques.• Modifications to existing SASW techniques and new
techniques will be investigated.– How should the measurement
system be set up?– How to calculate wave velocities?– How should the data be inverted?– Raleigh or Love waves?
MURI Kickoff/Progress 1-24-03 Scott and Mcclellan, Georgia Tech 28
Typical Surface Sensor ArraysUsed in Experimental Model and
at Field Test Sites
Linear Array of 16 Triaxial Accelerometers
Linear Array of