User Modeling in a
Cognitive Architecture
 (Driving in the HUD)


             DongNyeok Jeong
Computational Cognitive
          Modeling
• Human behavior behind the human
  mind
• Artificial intelligence - Turing Test
• Computer - Brain
• Brain - Mind
• Computer - Mind ?
An Integrated Theory of
           the Mind
•   Integrated !

•   ACT-R

    •   perceptual-motor modules, goal module,
        declarative memory module (An Integrated
        Theory of the Mind, John R. Anderson et al.,
        2004)

•   Architecture, Representation, Ontology,
    Knowledge type, Inference type
Cognitive Architecture
•   A general framework for specifying
    computational behavioral models of
    human cognitive performance

• Abilities and motorstorage, recall, learning,
  perception,
              : memory
                         action

• Constraints : memory decay,and limited motor
  peripheral perception encoding,
                                  foveal versus

    performance
Driver using a HUD
• Understanding and processing meaning of
  icons
• Skip motor process
      ex) handling, gearing, foot down, eye-
      movement ..

• Multi-processing in cognition
• HUD attention in driving situation
Flowchart in ACT-R




 Find-Shift-Harvest Paradigm
Production Rules in ACT-R

•   Keep on eye on front widow   •   HUD direction Changed




•   Distinguish HUD direction
Targets in Visual Process
•   Level 1,2   •   Level 3




                              •   Visicon
Driver Modeling
•   Modeling Driver Behavior in a Cognitive
    Architecture (Dario D. Salvucci, Drexel University,
    Philadelphia, Pennsylvania, 2006)

    •   Steering profiles, lateral position profiles, gaze
        distributions

    •   lane keeping, curve negotiation, lane changing

    •   ETA framework (Byrne, 2001; Gray, 2000; Gray &
        Boehm-Davis, 2000)

•   Control, monitoring, decision making component
Driver Modeling
Visual Attention


• ACT-R’S EMMA (Eye Movements and
  Movement of Attention) module (An
  integrated model of eye movements and
  visual encoding, Salvucci, 2001)
• Based on Top-down process
Future ...

• 나 동일한 모델링을 하는가
    제한된 환경에서 주어진 실험과 얼마

•   실험 환경을 얼만큼 실제 환경과
    유사하게 모델링 하는가
•   타겟 모델링의 세밀함
• ACT-R’s EMMA source code
Future ... (cont.)

• 실제 코딩 능력보다 중요한 것은 인간
    두뇌 프로세스를 이해하는 것
•   학습과 기억, 관련된 인지 모델링
• The activation equation
• ACT-R 6.0 ~ integrated perceptual and
    Motor Modules (EPIC)

Labmeeting 0709

  • 1.
    User Modeling ina Cognitive Architecture (Driving in the HUD) DongNyeok Jeong
  • 2.
    Computational Cognitive Modeling • Human behavior behind the human mind • Artificial intelligence - Turing Test • Computer - Brain • Brain - Mind • Computer - Mind ?
  • 3.
    An Integrated Theoryof the Mind • Integrated ! • ACT-R • perceptual-motor modules, goal module, declarative memory module (An Integrated Theory of the Mind, John R. Anderson et al., 2004) • Architecture, Representation, Ontology, Knowledge type, Inference type
  • 4.
    Cognitive Architecture • A general framework for specifying computational behavioral models of human cognitive performance • Abilities and motorstorage, recall, learning, perception, : memory action • Constraints : memory decay,and limited motor peripheral perception encoding, foveal versus performance
  • 5.
    Driver using aHUD • Understanding and processing meaning of icons • Skip motor process ex) handling, gearing, foot down, eye- movement .. • Multi-processing in cognition • HUD attention in driving situation
  • 6.
    Flowchart in ACT-R Find-Shift-Harvest Paradigm
  • 7.
    Production Rules inACT-R • Keep on eye on front widow • HUD direction Changed • Distinguish HUD direction
  • 8.
    Targets in VisualProcess • Level 1,2 • Level 3 • Visicon
  • 9.
    Driver Modeling • Modeling Driver Behavior in a Cognitive Architecture (Dario D. Salvucci, Drexel University, Philadelphia, Pennsylvania, 2006) • Steering profiles, lateral position profiles, gaze distributions • lane keeping, curve negotiation, lane changing • ETA framework (Byrne, 2001; Gray, 2000; Gray & Boehm-Davis, 2000) • Control, monitoring, decision making component
  • 10.
  • 11.
    Visual Attention • ACT-R’SEMMA (Eye Movements and Movement of Attention) module (An integrated model of eye movements and visual encoding, Salvucci, 2001) • Based on Top-down process
  • 12.
    Future ... • 나동일한 모델링을 하는가 제한된 환경에서 주어진 실험과 얼마 • 실험 환경을 얼만큼 실제 환경과 유사하게 모델링 하는가 • 타겟 모델링의 세밀함 • ACT-R’s EMMA source code
  • 13.
    Future ... (cont.) •실제 코딩 능력보다 중요한 것은 인간 두뇌 프로세스를 이해하는 것 • 학습과 기억, 관련된 인지 모델링 • The activation equation • ACT-R 6.0 ~ integrated perceptual and Motor Modules (EPIC)

Editor's Notes