An Adversarial Risk Analysis Model for an Autonomous Imperfect Decision Agent
2013 (English)In: Decision Making and Imperfection / [ed] Tatiana V. Guy, Miroslav Karny, David Wolpert, Springer Berlin/Heidelberg, 2013, 163-187 p.Chapter in book (Refereed)
Machines that perform intelligent tasks interacting with humans in aseamless manner are becoming a reality. A key element in their design is their abilityto make decisions based on a reasonable value system, and the perception of thesurrounding environment, including the incumbent persons. In this chapter, we providea model that supports the decision making process of an autonomous agent thatimperfectly perceives its environment and the actions performed by a person, whichwe shall designate user. The approach has a decision analytic flavour, but includesmodels forecasting the user’s behaviour and its impact over the surrounding environment.We describe the implementation of the model with an edutainment robotwith sensors that capture information about the world around it, which may serve asa cognitive personal assistant, may be used with kids for educational, recreationaland therapeutic purposes and with elderly people for companion purposes.
Place, publisher, year, edition, pages
Springer Berlin/Heidelberg, 2013. 163-187 p.
, Studies in Computational Intelligence, ISSN 1860-949X ; 474
Adversarial Risk Analysis (ARA), Game theory, Expected Utility, Maslow’s hierarchy of needs, robotics, multi-attribute utility function
Research subject Computer and Systems Sciences
IdentifiersURN: urn:nbn:se:su:diva-122136DOI: 10.1007/978-3-642-36406-8_6ISBN: 978-3-642-36405-1ISBN: 978-3-642-36406-8OAI: oai:DiVA.org:su-122136DiVA: diva2:865105