A framework for adaptive and empathic response system in an ambient intelligent space

College

College of Computer Studies

Department/Unit

Software Technology

Document Type

Archival Material/Manuscript

Publication Date

2009

Abstract

Behavior prediction is very important in any empathic ambient intelligent environment because the system has to learn first what the user is about to do before it can support the user. To fully support the activity of a user at a given moment, the system should not only consider the time of day or the history of the user's activity but rather, it should also consider the affect of the user. The proposed system considers affect because affect makes a person divert from normal routine. Moreover, most systems are trained in the beginning to learn the support that can be given to the user. The learned responses then are used in responding to the actions of the user. While previous works do not consider the affect and behavior of the user after the system provides the response, the proposed system will consider these to allow the system to improve over time. Since the system is to be deployed in an actual living environment, it is important for the system to learn from the user. This paper presents the framework to achieve a system that considers user's affect, behavior and context in predicting the behavior of the user to be supported. Since there are varying definitions of context, the paper shall also present the definition of context to be considered by the system. The system will be using gait, facial expression to identify human affect, previous behavior and context will be tracked using sensors that are installed in the living space. Issues surrounding a system that provides self-improving and adaptive empathic responses will be discussed. This paper presents previous works on ambient intelligent spaces where the system supports the behavior of the user by only considering behavior. This paper will discuss methods that are considered, methods for evaluating the system and data collection approaches. This paper discusses past and current works on context aware homes and the respective behavior prediction algorithms used. The system presents a framework for a system that issues adaptive empathic responses in an ambient intelligent environment by using context and past behavior of the user.

html

Disciplines

Computer Sciences | Physical Sciences and Mathematics | Software Engineering

Keywords

Ambient intelligence; Human-computer interaction

Upload File

wf_no

This document is currently not available here.

Share

COinS