Please use this identifier to cite or link to this item:
Title: Towards smart assistants in two-party collaboration
Keywords: human machine collaboration; planning under uncertainty; intention tracking; markov decision process; sampling-based algorithms
Issue Date: 21-Dec-2012
Citation: NGUYEN DINH TRUONG HUY (2012-12-21). Towards smart assistants in two-party collaboration. ScholarBank@NUS Repository.
Abstract: In this work, we build a framework for smart assistants in two-party collaboration, which depicts the scenario where the AI planner takes the role of a subordinate helping the lead agent achieve a set of common objectives, or subgoals. In the framework, each subgoal is modeled as a Markov Decision Process (MDP), the solutions of which dictate how the team should behave to achieve each subgoal. By assuming that the lead agent is optimal at the subgoal level, we can estimate his expected behavior as part of the optimal subgoal solution. This allows an efficient way to track the lead's intention using Bayesian inference, based on observations of his action history. We then use utility-theoretic maximization to select actions for the assistant. The approach is shown to yield near-human level assistance while incurring a running time that scales quadratically to the number of subgoals.
Appears in Collections:Ph.D Theses (Open)

Show full item record
Files in This Item:
File Description SizeFormatAccess SettingsVersion 
NguyenDTH.pdf2.82 MBAdobe PDF



Page view(s)

checked on May 23, 2019


checked on May 23, 2019

Google ScholarTM


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.