An Acoustic / Radar System for Automated Detection, …€¦ · Task 14.6 Task 14.7 External Data...

1

An Acoustic / Radar System for AutomatedDetection, Localization, and Classification of

Birds in the Vicinity of Airfields

Dr. Sebastian M. Pascarelle & Dr. BruceStewart (AAC)

T. Adam Kelly & Andreas Smith (DeTect)Dr. Robert Maher (MSU)

2

Outline

• Introduction

• Acoustic Sensor

• Acoustic Field Test Results

• Parabolic Dish Microphone Results

• Acoustic Classification Techniques

3

Introduction

• Hybrid Birdstrike Monitoring System:– Acoustic array

– Radar

– Parabolic dish microphone

• Data fusion

• Acoustic classification

4

Introduction

• Phase 2 STTR– Sponsor: Air Force Office of Scientific Research

Dr. Willard Larkin

– Team:AAC –Project management, system integration, acoustic array,

acoustic signal processing and classification

DeTect, Inc. –Bird Detection Radar, signal processing, radar dataanalysis, PCBCIA field test, bird strike experts

MSU –University partner, parabolic dish microphone, acousticclassification, atmospheric compensation model

5

Acoustic Sensor

Sparsely Populated Volumetric Array (SPVA)

• 18 hydrophones embedded in polyurethane• Provides 12.5 dB gain• Covers very large frequency range• 4π steradian coverage• Real-time angle of arrival without

beamforming• Fractional degree angle accuracy• Fiber optic telemetry

6

SPVA

• Proven sensor and signal processing technology currentlydeployed in the Navy fleet

• Outperforms legacy systems• Multiple sensors can give target range

7

Air Acoustic Array

• 18 microphonesmounted on rods

• Covers 0.2-20 kHzfrequency range

• Sound absorber tomitigate reflections

8

Complete Acoustic Sensor System

Digital recorder for offline signalprocessing (production system will do

real-time signal processing)

Air SPVA sensorand pre-amplifiers

9

SPVA Real-Time Displays

10

Data Fusion

DFECData

FusionExternal

Communication

DFFSData

Structures

SPVAD/C/L

CRHD/C/L

RadarD/C/L

ESMD/C/L

IRD/C/L

Shipboard Organic Sensors

UAV Sensors

VideoD/C/L

IRD/C/L

DFNCData Fusion

Control

DFENData Fusion

Engine

DFDPData Fusion

Display

DFDBData Fusion

Database

DFRGData FusionRegistration

DataFusion

IntelligentController

Task 1.1

Task 1.3

Task 1.4 Task 1.5

Task 1.6

Task 1.2

Task 1.7

Data Fusion Function

SystemNetwork

DFECData

FusionExternal

Communication

DFFSData

Structures

SPVAD/C/L

CRHD/C/L

RadarD/C/L

ESMD/C/L

IRD/C/L

Shipboard Organic Sensors

UAV Sensors

VideoD/C/L

IRD/C/L

DFNCData Fusion

Control

DFENData Fusion

Engine

DFDPData Fusion

Display

DFDBData Fusion

Database

DFRGData FusionRegistration

DataFusion

IntelligentController

Task 1.1

Task 1.3

Task 1.4 Task 1.5

Task 1.6

Task 1.2

Task 1.7

Data Fusion Function

SystemNetwork

Task 14.1

Task 14.2

Task 14.3

Task 14.4 Task 14.5

Task 14.6

Task 14.7

DFECData

FusionExternal

Communication

DFFSData

Structures

SPVAD/C/L

CRHD/C/L

RadarD/C/L

ESMD/C/L

IRD/C/L

Shipboard Organic Sensors

UAV Sensors

VideoD/C/L

IRD/C/L

DFNCData Fusion

Control

DFENData Fusion

Engine

DFDPData Fusion

Display

DFDBData Fusion

Database

DFRGData FusionRegistration

DataFusion

IntelligentController

Task 1.1

Task 1.3

Task 1.4 Task 1.5

Task 1.6

Task 1.2

Task 1.7

Data Fusion Function

SystemNetwork

DFECData

FusionExternal

Communication

DFFSData

Structures

SPVAD/C/L

CRHD/C/L

RadarD/C/L

ESMD/C/L

IRD/C/L

UAV Sensors

VideoD/C/L

IRD/C/L

DFNCData Fusion

Control

DFENData Fusion

Engine

DFDPData Fusion

Display

DFDBData Fusion

Database

DFRGData FusionRegistration

DataFusion

IntelligentController

Task 1.3

Task 1.4 Task 1.5

Task 1.6

Task 1.2

Task 1.7

Data Fusion Function

SystemNetwork

Task 15.2

Task 15.3

Task 15.4 Task 15.5

Task 15.6

Task 15.7

Task 15.1

11

Atmospheric Compensation

• Wind speed• Temperature• Humidity

12

Field Test: PCBCIA

• Located in proximity to airport runway,trees, and water

• Test at beginning of Nov. to catch Fallmigration

Test Location

13

Field Test: PCBCIA

14

Aircraft Detection & Tracking

• Track of small aircraft passing nearly directlyoverhead proves system capability

• Angle accuracy is < 10 deg

15

Aircraft Detection & Tracking:Radar Confirmation

16

Ø In a typical 30 minute time interval, at least30 episodes of flight calls were detected.

Ø Calls from a single bird were frequent,every 1 to 3 seconds.

Ø Each episode consisted of many calls,typically 4 to 30.

Ø Presumed local, not migratory flight

Ø Calls were mostly at higher frequencies,indicating small, low-threat birds.

Morning Flight Calls

17

Morning Flight Calls

Human interpreters will easily recognize tracksdue to frequent and numerous calls.

18

Acoustic detection,localization, and tracking

Tracking software maintains integrity of overlapping tracks

19

• Test setup has a singleSPVA, so range is notknown.

• (Relative) ranges areestimated fromamplitudes of calls.

• There is one freeparameter, to be fixedfrom radar data.

Morning Flight Calls

20

Morning Flight Calls:Radar Confirmation

• Radar confirmationshows acousticdetections rangedup to 600 ft

21

Evening Flight Calls

Ø During a typical 6-minute interval, 5episodes of flight calls were detected.

Ø Interval between calls in one episode istypically 5 to 10 seconds.

Ø Each episode has very few calls, typically1 to 4.

Ø Presumed migratory flight

Ø Calls were mostly at higher frequencies,indicating small, low-threat birds.

22

Evening Flight Calls

Acoustic detections ranged up to 1500 feet

23

Multi-Source DetectionSystem can detect and track multiple targets simultaneously

24

Bat Detection

• Results of a bat detection event consisting of 8 calls• Calls are at the upper end of the detection band

-1 -0.5 0 0.5 1

-0.4

-0.2

0

0.2

0.4

0.6

0.8

1

1.2

12.5 sec

17.5

18.7

21.922.123.1

27.0

27.2

Azimuth bearings with detection times,indicating an erratic flight path

conventionalspectrogram

complexreassigned

spectrogram

25

Parabolic Dish Microphone

• Classification requires high S/N data• Lots of competing background noise at airfield• Need directional, high-gain microphone• Large electronically steered arrays expensive• Solution:

– Commercially available dish microphone– Mounted on two-axis servo– Mechanically steered by:

• radar track data• acoustic bearing data

– Provides signal isolation and gain

26

Parabolic Dish Performance

• Plot shows parabolic dish performanceimprovement over simple microphones

• As much as 25 dB gain in laboratory tests

55

60

65

70

75

80

85

0 2000 4000 6000 8000 10000

Frequency (Hz)

Ampl

itude

(dB

)

DPAPanasonicDish

27

Parabolic Dish Performance

Two sparrow calls: the first is heard faintlyby the air array, strongly with the dish.

air arrayelement 9

parabolicdish

microphone

10 kHz

8 kHz

6 kHz

4 kHz

10 kHz

8 kHz

6 kHz

4 kHz

28

Parabolic Dish Performance

Parabolic dish provides 19 dB gain for bird calls

29

Parabolic Dish Steering

Microcontroller circuit directs the dish tobearings from air array signal processing

Microcontroller

Dual Full-Bridge Driver

24V DC

Dual Full-Bridge Driver

RS-232Serial Cable

ElevationStepper

AzimuthStepper

Microcontroller(eval board)

StepperControlBoards

30

Classification Software

• Frequency Track Analysis– AAC

– MSU

• Cortical Processing Theory– AAC & UMD

• Composite Classifier– MSU

31

Frequency Track Analysis

Blue Jay Call

Spectrogram Frequency Tracks

•Compute spectrogram and smooth

•Find local maxima at each time and connect (peak tracks)

•Remove short and weak tracks

•Compute features (min, max, and mean frequency, length, slope,… )

•Compare features statistically with training set

32

Frequency track classifier results

• MSU: 12 species and 16 synthesized sounds,99% success with 12 db SNR added noise

• AAC: 4 species trained, 10 species tested with0 false positives (except blue jay)

Blue jay matching tracks Herring gull syllable recognition

33

Center for AuditoryandAcoustic Research

Response field in threedimensions (rate, scale, time)visualized using isosurfaces

4.5 5 5.5 6 6.5 7-6

-4

-2

0

2

4

6Principal Components of Rate-Scale Matrices

prin

cipa

l com

pone

nt 2

house wrenchipping sparrow

In-flight chip calls from house wren (top) and sparrow (bottom) are clearlydistinguished and classified using rate-scale representation.

freq

uenc

yfr

eque

ncy

time

time

scal

esc

ale

rate

rate

wren

sparrow

Prof. S. Shamma

Cortical Processing Theory

34

Bird call recognition from rate-scale matrix

35

MSU Compound Classifier

• Auditory cortex ripple-based• Frequency track analysis• Other candidate methods• Weighted comparison of all will give optimal

result

Classifier N

Classifier 2

Classifier 1

• Environmental status• High-level analysis• a priori inputs

Signal PropertiesDetermination

Input Signal

Classification

Reliability RatingEstimate

WeightedComparison

36

Conclusions

• AAC’s highly successful underwater acoustic array sensorhas been transitioned to an air sensor

• Field testing has proven its capability to detect a variety ofacoustic sources at significant distances

• Testing alongside radar has shown that the two systemsare highly complementary

• Parabolic dish provides significant gain over array• Combined system of array, radar, and dish is a robust

solution to monitoring bird activity at airfields• System can be used to detect, track, and classify other

activity as well:– vehicles, watercraft, aircraft, people, bats– Potential Homeland Security applications –perimeter security,

border security