Some Combinatorial and Geometric Problems in Maritime Safety

1
Some Combinatorial and Geometric Problems in
Maritime Safety

• Jayanta Majumder

Supervisor Prof. Dracos Vassalos
Dated 21st January, 2004
2
Objectives

• To introduce the theme.

• To introduce the problems addressed.

• To inform about the results obtained so far.

• To outline the future directions.

• To take your valuable comments into account

3
Combinatorial Problems

• Combinatorial problems involve study of the
  arrangement, grouping, ordering, or selection of
  discrete objects, usually finite in number.

• The configuration space for such problems can not
  be numerically parameterised.

• Challenge Combinatorial explosion.

4
Problems Addressed

• Modelling of shipboard environment for simulation
  of and automated reasoning about crisis.

• Automated reasoning for crisis management.

• Mathematical problems in evacuation/crowd flow
  simulation.

• Extraction of functional information from general
  arrangement drawings.

5
Shipboard Environment Modelling (for Crisis
Management and Situational Awareness)

• Shipboard environment is large, whereas
  individual entities of interest in a crisis
  (passenger, crew, sensors, doors, hatches, fire,
  fault etc.) are small and localized.

• The sum total of all localized spatial and
  temporal relationships among localized entities
  give rise to overall interrelationships
  dependencies, but implicitly.

• An environment model is required for overall
  situational awareness.

6
Automated Reasoning (for Crisis Management)

• Crisis response poses interesting combinatorial
  problems.

• Constrained routing problems How to route
  passengers to safety given a state of crisis,
  where several pieces of localized information
  about the crisis are available.

• Scheduling problems Having identified
  operational tasks and their deadlines, how should
  each task be scheduled with due cognizance of
  available operational resources.

7
Evacuation or Crowd Flow Modelling

• Evacuation simulation involves some interesting
  combinatorial and geometric problems like
  geometric obstacle avoidance and planning.

• Besides its current commercial value (at S_at_S
  Ltd., for example), it will serve as a predictive
  tool for crisis management software.

• A new and improved formulation evacuation
  simulation is being developed.

8
Syntactic Pattern Recognition

• Available general arrangement (GA) and
  construction drawings are dumb, i.e. they lack
  semantic information as required for reasoning
  tasks.

• Current approach to creating environment model
  instances is - to interpret the drawings manually
  and to hand-draft the model instance.

• Automated recognition saves a lot of monotonous
  work.

• Attributed-network query technology developed
  here, is useful in reasoning about crisis.

9
Results So Far

• A commercialised tool for evacuation modelling.

• A graphics recognition system that recognizes
  features from general arrangement drawings (Think
  of a graphic find/replace system for AutoCAD).

• A software toolkit (a pipeline) to convert
  scanned/CAD system prepared GAs into an shipboard
  environment model with adjacency topology encoded.

10
Results So Far

• A proof-of-concept crowd flow simulation program
  based on the new formulation.

• A tiny but generic input instance generator for
  Monte-Carlo simulation, currently applied in fire
  simulation.

• A STREP proposal for an IST project under EU-FP6
  (result awaited).

11
Future Directions

• Develop a full prototype crowd flow simulator
  based on the automatically inferred environment
  model, and transfer the technology for
  commercialisation.

• Develop automated reasoning modules that fuse
  sensor data and plan about prevention and
  mitigation of crisis.

• Develop onboard crisis management system
  (hardware- software integration).

12
Your Comments and Questions

• My Profuse Thanks for Your Interest and Patience.