The altruistic robot: Do what i want, not just what i say

Forskningsoutput: Kapitel i bok/rapport/Conference proceedingKonferenspaper i proceeding

Abstract

As autonomous robots expand their application beyond research labs and production lines, they must work in more flexible and less well defined environments. To escape the requirement for exhaustive instruction and stipulated preference ordering, a robot’s operation must involve choices between alternative actions, guided by goals. We describe a robot that learns these goals from humans by considering the timeliness and context of instructions and rewards as evidence of the contours and gradients of an unknown human utility function. In turn, this underlies a choice-theory based rational preference relationship. We examine how the timing of requests, and contexts in which they arise, can lead to actions that pre-empt requests using methods we term contemporaneous entropy learning and context sensitive learning. We provide experiments on these two methods to demonstrate their usefulness in guiding a robot’s actions.

Detaljer

Författare
  • Richard Billingsley
  • John Billingsley
  • Peter Gärdenfors
  • Pavlos Peppas
  • Henri Prade
  • David Skillicorn
  • Mary Anne Williams
Externa organisationer
  • University of Technology Sydney
Forskningsområden

Ämnesklassifikation (UKÄ) – OBLIGATORISK

  • Människa-datorinteraktion (interaktionsdesign)
Originalspråkengelska
Titel på värdpublikationScalable Uncertainty Management - 11th International Conference, SUM 2017. Proceedings
RedaktörerSerafin Moral, Daniel Sanchez, Nicolas Marin, Olivier Pivert
FörlagSpringer
Sidor149-162
ISBN (elektroniskt)978-3-319-67582-4
ISBN (tryckt)9783319675817
StatusPublished - 2017 jan 1
PublikationskategoriForskning
Peer review utfördJa
Externt publiceradJa
Evenemang11th International Conference on Scalable Uncertainty Management, SUM 2017 - Granada, Spanien
Varaktighet: 2017 okt 42017 okt 6

Publikationsserier

NamnLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volym10564 LNAI
ISSN (tryckt)0302-9743
ISSN (elektroniskt)1611-3349

Konferens

Konferens11th International Conference on Scalable Uncertainty Management, SUM 2017
LandSpanien
OrtGranada
Period2017/10/042017/10/06