Planning and acting in partially observable stochastic domains
Artificial Intelligence
Partially observable Markov decision processes for spoken dialog systems
Computer Speech and Language
| Hi-index | 0.00 |
A Situated Conversational Agent (SCA) is an agent that engages in dialog about the context within which it is embedded. Situated dialog is characterized by its deep connection to the embedding context, and the precise cross-timing of linguistic and non-linguistic actions. This paper describes initial research into the construction of an SCA that engages in dialog about collaborative physical tasks, in which agents engage in dialog with the joint goal of manipulating the physical context in some manner. Constructing an SCA that can interact naturally in such tasks requires an agent with the ability to interleave planning, action, and observation while operating in a partially observable environment. Consequently, I propose to model an SCA as a Partially Observable Markov Decision Process (POMDP).