Composite Event Patterns for Maritime Monitoring Manolis Pitsikalis - - PowerPoint PPT Presentation

Composite Event Patterns for Maritime Monitoring

Manolis Pitsikalis1,2, Ioannis Kontopoulos1,3, Alexander Artikis4,1, Elias Alevizos1, Paul Delaunay5, Jules-Edouard Pouessel5, Richard Dreo5, Cyril Ray5,6, Elena Camossi7, Anne-Laure Jousselme7, Melita Hadzagic7

1Institute of Informatics & Telecommunications, NCSR Demokritos, Athens, Greece 2Department of Informatics and Telecommunications, National & Kapodistrian

University of Athens, Greece

3Department of Informatics and Telematics, Harokopio University of Athens, Greece 4Department of Maritime Studies, University of Piraeus, Greece 5Naval Academy Research Institute, Brest, France 6Arts et Metiers ParisTech, France 7Centre for Maritime Research and Experimentation (CMRE), NATO, La Spezia, Italy

Composite Event Recognition

INPUT ◮ RECOGNITION ◮ OUTPUT

Event Recognition System CE Definitions Streams of SDEs . . . . . . . . . . . . Recognised CEs . . . . . . . . . . . .

October 31, 2018 SETN2018 Hellenic Conference on Artificial Intelligence 2

Maritime Monitoring

October 31, 2018 SETN2018 Hellenic Conference on Artificial Intelligence 3

Automatic Identification System

The Automatic Identification System is an extensively used autonomous tracking system that allows transmission of dynamic and static vessel information. Transmitted Data:

• Dynamic information: MMSI, timestamp, position, speed,

heading, course over ground, rate of turn

• Static & Voyage related information: IMO, name, type,

dimensions, destination, ETA, draught

October 31, 2018 SETN2018 Hellenic Conference on Artificial Intelligence 4

Tugging operation

• Two vessels, close in distance...
• one of them is a tug boat...
• and they are moving with same speed and direction at the

same time.

October 31, 2018 SETN2018 Hellenic Conference on Artificial Intelligence 5

Rendezvous

• Two vessels are close in distance...
• and they are stopped or sailing with low speed...
• for a period of time.

October 31, 2018 SETN2018 Hellenic Conference on Artificial Intelligence 6

Fishing

• A fishing vessel in a fishing area...
• sailing with trawling speed...
• and its course is “erratic”.

October 31, 2018 SETN2018 Hellenic Conference on Artificial Intelligence 7

Event Calculus

• A logic programming language for representing and

reasoning about events and their effects.

• Key components:

event (typically instantaneous). fluent: a property that may have different values at different points in time.

• Built-in representation of inertia:

F = V holds at a particular time-point if F = V has been initiated by an event at some earlier time-point, and not terminated by another event in the meantime.

October 31, 2018 SETN2018 Hellenic Conference on Artificial Intelligence 8

Run-Time Event Calculus : Predicates

Predicate Meaning happensAt(E, T) Event E occurs at time T initiatedAt(F = V , T) At time T a period of time for which F = V is initiated terminatedAt(F = V , T)

At time T a period of time for which

F = V is terminated holdsFor(F = V , I) I is the list of the maximal intervals for which F = V holds continuously holdsAt(F = V , T) The value of fluent F is V at time T union all([J1, . . . , Jn], I) I =(J1 ∪ . . . ∪ Jn) intersect all([J1, . . . , Jn], I) I =(J1 ∩ . . . ∩ Jn) relative complement all I = I ′ \ (J1 ∪ . . . ∪ Jn) (I ′, [J1, . . . , Jn], I)

October 31, 2018 SETN2018 Hellenic Conference on Artificial Intelligence 9

Run-Time Event Calculus: Windows

time Q178 Q182 Q181 Q180 Q179 Q177 Working Memory time Q178 Q182 Q181 Q180 Q179 Q177 Working Memory time Q178 Q182 Q181 Q180 Q179 Q177 Working Memory October 31, 2018 SETN2018 Hellenic Conference on Artificial Intelligence 10

Patterns: Rendezvous

holdsFor(rendezVous(Vessel1, Vessel2), I) ← holdsFor(proximity(Vessel1, Vessel2), Ip), holdsFor(lowSpeed(Vessel1), Il1), holdsFor(lowSpeed(Vessel2), Il2), holdsFor(stopped(Vessel1) = farFromPorts, Is1), holdsFor(stopped(Vessel2) = farFromPorts, Is2), holdsFor(tugging(Vessel1, Vessel2), Itug), I1 = Il1 ∪ Is1, I2 = Il2 ∪ Is2, I = (I1 ∩ I2 ∩ Ip) \ Itug, I > RVtime.

5 5 10 15 20 m

Critical points Vessel 1 AIS messages Vessel 2 AIS messages Vessel 1 trajectory Vessel 2 trajectory

Rendezvous between two fishing vessels.

October 31, 2018 SETN2018 Hellenic Conference on Artificial Intelligence 11

Patterns: Engaged in fishing

initiatedAt(trawlingMovement(Vessel), T) ← happensAt(change in heading(Vessel), T), holdsAt(withinArea(Vessel, fishing)), T). deadlineUE(trawlingMovement(Vessel), TrawlingMDtime). holdsFor(trawling(Vessel), I) ← Vessel : fishing, holdsFor(trawlSpeed(Vessel), It), holdsFor(withinArea(Vessel, fishing), Iw), holdsFor(trawlingMovement(Vessel), Itc), I = It ∩ Iw ∩ Itc, I > TrawlingDuration.

1 1 2 3 4 km

Critical points Raw AIS positions Trawling trajectory Fishing area

A fishing vessel during trawling activity.

October 31, 2018 SETN2018 Hellenic Conference on Artificial Intelligence 12

Experimental Setup: Dataset

Brest, France October 1st 2015 - 31st March 2016.

Data # Vessels 5K Position signals 18M Critical position signals 4.6M Spatio-temporal events 811K

October 31, 2018 SETN2018 Hellenic Conference on Artificial Intelligence 13

Experimental Setup: Streams

October 31, 2018 SETN2018 Hellenic Conference on Artificial Intelligence 14

Experimental Results: Accuracy

Event F1-Score Speed not compatible with area 1.000 Anchored 1.000 Moored 1.000 Aground 1.000 Loitering 1.000 Rendezvous 1.000 Under way 0.997 Movement ability affected 0.964 Trawling speed 0.961 Engaged in fishing 0.961 Speed not compatible with vessel type 0.937 Tugging 0.915 Dead in water, drifting 0.838

October 31, 2018 SETN2018 Hellenic Conference on Artificial Intelligence 15

Experimental Results: False Positives/Negatives

Critical Points Raw AIS messages TP TP FN TP FP October 31, 2018 SETN2018 Hellenic Conference on Artificial Intelligence 16

Average recognition time: all events

Step = 2 hours

2 4 8 16 2 4 6 Window size (hours) Average Recognition Time (sec) Enriched AIS Stream Critical Point Stream

Average recognition times.

October 31, 2018 SETN2018 Hellenic Conference on Artificial Intelligence 17

Average recognition time: per composite event (part A)

2 4 8 16 18 2 4 6 8 Window size (hours) Average Recognition Time (sec) underWay highSpeedNC travelSpeed

Enriched AIS stream

2 4 8 16 18 1 2 3 4 Window size (hours) Average Recognition Time (sec) underWay highSpeedNC travelSpeed

Critical point stream

October 31, 2018 SETN2018 Hellenic Conference on Artificial Intelligence 18

Average recognition time: per composite event (part B)

2 4 8 16 18 1 2 3 4 Window size (hours) Average Recognition Time (sec) atAnchor moored aground

Enriched AIS stream

2 4 8 16 18 1 2 3 4 Window size (hours) Average Recognition Time (sec) atAnchor moored aground

Critical point stream

October 31, 2018 SETN2018 Hellenic Conference on Artificial Intelligence 19

Average recognition time: per composite event (part C)

2 4 8 16 18 1 2 3 4 Window size (hours) Average Recognition Time (sec) trawling tugging inSAR

Enriched AIS stream

2 4 8 16 18 1 2 3 4 Window size (hours) Average Recognition Time (sec) trawling tugging inSAR

Critical point stream

October 31, 2018 SETN2018 Hellenic Conference on Artificial Intelligence 20

Average recognition time: per composite event (part D)

2 4 8 16 18 1 2 3 4 Window size (hours) Average Recognition Time (sec) rendezVous loitering vesselIST

Enriched AIS stream

2 4 8 16 18 1 2 3 4 Window size (hours) Average Recognition Time (sec) rendezVous loitering vesselIST

Critical point stream

October 31, 2018 SETN2018 Hellenic Conference on Artificial Intelligence 21

• Slides, complex event recognition software, datasets:

http://cer.iit.demokritos.gr

• Brest dataset:

https://zenodo.org/record/1167595

October 31, 2018 SETN2018 Hellenic Conference on Artificial Intelligence 22

Any questions?

October 31, 2018 SETN2018 Hellenic Conference on Artificial Intelligence 23