Demo 2: Hazard Avoidance

1
Demo 2 Hazard Avoidance

• Dr. Stathis Kasderidis

2
Packages Modeling

• From Jun-02 until Dec 03 use KCL package (KCLP).
• From Jan-03 till end we use both KCLs and QUB
  packages (QUBP).
• KCLs package is used mainly for prototyping and
  new algorithm development.
• QUBs package is used for algorithm testing,
  experiment design and statistical evidence
  gathering.

3
Use Cases

• 1. Traffic Jam Cooperative route planning.
• 2. Normal Driving Check for speeding, careless
  driving, boredom, etc.
• 3. Demanding Driving in Countryside Test
  movement of car / danger level against road
  conditions and circuit.
• 4. City Driving Test accident avoidance with
  pedestrians.
• 5. Highway Driving Test accident avoidance with
  approaching / overtaking cars.

4
Use Cases Cont.

• 6. Near-Accident Response Test the execution of
  alternative sequence of actions.
• 7. Fast response Test the reaction time of the
  system in decision making / alternative planning
  (not yet defined!)

5
Roadmap

• For end July-02 include pre-processing in KCLP.
• For end Sep-02 include level 3 artefacts
• Biometric Artefact
• Environmental State
• Car State
• Proximity Artefact
• Allow simple state representation, decision
  making and attention response for the above
  artefacts.

6
Roadmap Cont.

• End of Jan-03 include Hazard Avoidance level 4
  artefact.
• End of Jan-03 includes proper state
  representations and decision making, monitor
  response and attention control. Not included user
  adaptation.

7
Roadmap Cont. 2

• Use Cases of Normal Driving, Demanding Driving in
  Countryside and maybe part of Highway Driving
  will be used for system functional testing for
  the September DC Jumboree.

8
Questions

• How do we show emergent behaviour?
• What are the statistical measures (and their
  thresholds) that show improved performance for an
  (user) adapted system in comparison to a
  non-adapted one for this demo?

9
Biometric Artefact

• Input Variables
• Environmental Temperature
• Environmental Humidity
• Persons Temperature
• Persons Heart Rate
• Response Output Classes
• Normal
• Alert/Unknown
• Ill
• Dangerous
• Immediate

10
Environmental Status

• Input Variables
• Temperature
• Humidity
• Visibility
• Ice existence
• Fog existence
• Ambient light level (?)
• Wind (?)
• Time (?)

11
Environmental Status Cont.

• Output Variables
• Weather condition (Fog, Ice, Rain, Fair, Windy,
  Storm, etc?)
• Day / Night
• Time

12
Car State Artefact

• Input Variables
• Velocity vector
• Acceleration vector
• Heading vector
• Breaking Force
• Friction force
• Engine thrust force
• Road friction coefficient
• Steering wheel angle
• Existence of mechanical failures

13
Car State Artefact Cont.

• Response Output Classes
• Normal
• Alert/Unknown
• Damaged
• Dangerous
• Immediate

14
Proximity Artefact

• Input variables
• Radar sensor (provides range information for
  objects inside the radars horizon. Assumption
  Directed radar with 120 degrees forward coverage.
  Produces position velocity information for
  other objects).
• Position information (GPS). (Actually position
  vector of the cars centre and roads boundary
  points array).
• Output Variables
• Number of objects present (in the forward
  direction)
• Distance of each known object
• Velocity vector for each known object

15
Proximity Artefact Cont.

• Response Classes
• Approaching
• Overtaking
• Moving
• Stationary
• In Collision course
• Dangerous (In collision course and too close)
• Immediate