Sensordatafusion

1
Sensordatafusion

• Egils Sviestins
• SaabTech Systems

2
Fusion levels (JDL model)
Level 1 Objects
Level 2 Situations
Level 3 Intentions
Sources
Level 4 Process
3
Terminologi
Objekt
Situationer
Avsikter
Sensordata- fusion
Sensor- data
Informations- fusion
Andra data
Styrning Optimering
4
Modeller

• Mätningar/information räcker inte
• Modeller krävs!
• Matematiska
• exempel
• Idéer om verkligheten/mentala modeller
• Begränsat av naturlagar, ekonomiska lagar,
  mänsklig förmåga etc.
• Mätningar/information snävar in möjligheterna

1
2
3
5
Från verkligheten...
Rån stöld e.d. som utförs under hot om våld
6
Context
7
Data processing Improvement or Destruction?
Raw information
Sensor
User
Meaningful information
8
Synkanalen (hypotetiskt!)
9
Hörselkanalen (hypotetiskt!)
10
Early fusion...
... or late?
WSC
11
Seeing (hypothetical)
WSC
12
Artskilda sensorer
13
Tidig fusion - för och emot

• Mindre risk för tvetydigheter
• Osäkerheter kan lättare beskrivas statistiskt -
  Bayes teori kan användas
• Mindre robust m a p systematiska fel
• Svårt hantera artskilda källor

14
Inte så enkelt...
15
Fusionsprincip i hjärnan?
16
The Radar Data Processing Chain
Extractor
Receiver
Tracker
A12
A07
Raw video
Plots (R,az)
Tracks (,x,y,vx,vy,...)
WSC
17
Steps in Tracking
18
The Tracking Cycle
WSC
19
Filtering techniques

• Linear regression (least squares batch
  processing) (hardly used in this context)
• (70s) Alpha-Beta
• (80s) Adaptive Kalman
• (90s) Interactive Multiple Model (IMM)
• (2000s ?) Non-linear filtering?

20
Linear regression
x
How to handle maneuvering targets???
t
21
Alpha-Beta filtering
a and b are tuning constants between 0 and 1
Prediction step
Updating step
ab0 Measurement has no effect ab1 History
has no effect
22
Kalman filtering
Current state uncertainties Measurement
uncertainties New state uncertainties
Like a-b-filter, but Automatically optimizes a
and b Best weighting between history and
measurement Output includes estimated accuracy
23
Probability densities
.
x
Update
Prediction
Measurement
x
24
IMM States
25
IMM structure
26
Bayes teori
27
Associering

• M målspår, N plottar hur koppla samman?
• OBS! Falska/saknade plottar, falska/saknade
  målspår
• Närmaste granne?
• Närmaste granne i statistiskt avstånd?
• Global optimering statistiskt avstånd(minimera
  )?
• Söka globalt mest sannolika koppling?Hur man än
  gör kan det bli fel. Motiverar multihypotes

28
Measurement-to-track association

• Clusters with M measurements and N tracks
• Form hypotheses like
• Calculate probabilities for each hypothesis, e.g.

29
LPQ association Plot Track clusters
30
Bayesian track initiation
Given a tentative track. Two hypotheses H0
Track is false H1 Track is genuine Cnp(H1)
Credibility at scan n Obtained measurement z.
Spurious plot density ps.
31
Initiation by Credibility

• Required Fast initiation and low false track
  rate
• Sequential hypothesis testing
• Credibility C likelihood that a potential track
  is genuine

cred
32
Andra sensorer

• Bildalstrande
• TV
• FLIR (Forward Looking Infrared)
• Millimetervågsradar
• SAR (Synthetic Aperture Radar)
• Icke bildalstrande
• Störbäringsavtagare
• Signalspaning
• IRST (Infrared Search Track)
• Akustiska/Hydroakustiska sensorer
• GPS

33
Decentralized Multi-Radar Tracking
34
Centralized Multi-Radar Tracking
35
Filling coverage gaps
Two radars Coverage gap
Red single radar track lost and reinitiated
Decentralized MRT may give confusing picture
Centralized MRT performs well
36
Disadvantages of centralized multi-radar tracking

• More sensitive to bias errors
• Bias compensation required
• Difficult to distribute CPU load on several
  processors
• But not impossible
• Existing data links often do not supply plot
  level data
• Sometimes requires hybrid solutions
• Sensors sometimes include extensive processing
• Sometimes requires hybrid solutions

37
Strobes only
150 km
38
Crossings
39
Reasons for Multi-Sensor Tracking

• Radars can be jammed
• Protective need to keep radars silent
• Radars dont always give best target detection
• May support target identification

40
Target Type Identification

• Based on
• Direct observations
• ESM / IRST measurements
• Kinematics
• Each track carries a vector with probabilities of
  possible target types.
• Requires a library of target type characteristics

41
MST scenario
42
Example
Lockheed F16
MiG-29
Mirage 2000
Lockheed U2
MiG-25
3 3 3 1 1 3 3 3 3 1
3 3 3 2 2 3 3 3 3 2
3 3 3 4 5 3 3 3 4 5
3 3 3 4 5 6 3 3 3 4
3 3 3 4 5 6 7 6 6 7
43
Kinematic typingOffline Create Target Type
Database

• Max altitude
• Min/Max speed as function of altitude
• Max climb rate as function of altitude
• Max distance from base
• Max linear/turn acceleration as function of
  altitude

44
Step 1 - Collect flight data

• Max altitude
• Min/max velocity as function of altitude
• Max climb rate
• Max distance from base
• ltMax linear/turn acceleration as function of
  altitudegt
• Utilise meteorological data if available

45
Step 2 - Update Probability Vector
CollectedFlight Data
NewProbabilityVector p(F16),...
PreviousProbabilityVector
Bayes Rule
p(F16),...
Target TypeDatabase
46
Avrundning

• Sensordatafusion - uppgifter om enskilda objekt
  baserat (mest) på sensordata
• Bygger oftast på matematiska modeller
  ochBayesiansk hypotesprövning
• Många svåra områden återstår
• Sensorer som ger knepiga data
• Svårtolkade scenarier (t ex mark och undervatten)
• Gemensam lägesbild (distribuerad fusion)
• Fusion av starkt artskilda sensorer
• Integration med infofusion