2017년 11월 28일. Pfeifer 교수님이 OIST를 방문했다. 아는 사람은 알겠지만.. 맞다. 바로 그 유명한 책 “How the Body Shapes the Way We Think”의 저자! 무려 자신의 Wikipedia 페이지가 있는 인물이다! Respect!

Kenji Doya 교수님의 연구실을 방문하고 세미나를 한건데, 좋은 기회를 놓치지 않고 참여할 수 있었다. 그리고 세미나 후엔 내 연구를 간략히 소개해드리는 기회도 얻었다. 시간에 쫓겨서 자세히 소개는 못해드렸지만, 그래도 embodied cognition approach로 연구를 소개하니 잘 들어주신듯 하다.

아래는 Pfeifer 교수님의 세미나 중 메모한 내용을 정리..그냥 올린 것이다. 전반적으로 Soft Robotics의 중요성을 설파하고 가셨다. Soft Robotics라는 저널도 소개하고 가셨는데, 오늘 (2018년 1월 10일) 확인해보니, Impact Factor가 무려 8.649!!!! 로보틱스에서 이런 Impact Factor는 처음 본다.. 교수님의 세미나 내용 (키워드 위주)과 중간중간에 든 나의 의문점들을 나중에 까먹지 않기 위해서 일단 올려 본다.

“The next generation of smart machines” by Prof. Pfeifer

부제: Others beside deep NN

  • JIBO ~= mobile phone (functionality of an iPhone)
  • sensorimotor functionality
  • “Novelty wears off quickly”
  • Anthropomorphization
    • Attribute much more intelligence
  • “Full-featured humanoids not anywhere in sight”
  • Machine Learning: “Statistically impressive, individually unreliable”
  • Performance vs. Competence
    • Low Competence, High Performance: AlphaGo
    • Both High: Human
    • Both LOW: not interesting
    • High Competence, Low performance: Politicians? intellectuals? ;)
  • Recognize cat : but it doesn’t know what cat is -> low competence
  • “Not replacing humans, but enhancing human experience“
  • Some of the projects
    • RoboLounge Project: Experience the future
    • LEGO MINDSTORM
      • Same control: embedded in the different physical structure –> different behavior!
  • Structure, Materials, Environment!
  • Passive Dynamic Walker
  • Ecological Niche
  • Soft Robotics
    • Universal Coffee-balloon Gripper: Adaptation at periphery (Not centrally controlled)
    • Thimbles Experiment
  • Power of (soft) materials
    • Changeable materials / morphology
  • Reflex arc concept is psychology.. by John Dewey, 1896
  • Sensory-Motor Contingency - Noe
  • Visual - Action
  • Auditory - Action
  • Scaffolding
  • Eco - System: We should consider environment as well
    • Environment that robot & people live together
  • My Questions)
    • Robot body doesn’t change over time –> intelligence doesn’t change over time. Would it be weird if the cleaning robot becomes smarter?.
      • It supposed to be smart at the beginning.. But.. learning the structure of the house through experience of cleaning might make it look “getting smarter”
    • Human body change –> intelligence changes over time.
      • not necessarily increase. it might decrease at the later stages. e.g., dementia
      • Ok.. if the sensors are broken, motors have been weaken.. then, will people think its intelligence has decreased?
  • Soft Materials: The future of robotics is material!