A human-like perceptual model for intelligent virtual agents = Modelo de percepción para agentes virtuales inteligentes

Herrero Martín, María del Pilar (2003). A human-like perceptual model for intelligent virtual agents = Modelo de percepción para agentes virtuales inteligentes. Thesis (Doctoral), Facultad de Informática (UPM).

Description

Title: A human-like perceptual model for intelligent virtual agents = Modelo de percepción para agentes virtuales inteligentes
Author/s:
  • Herrero Martín, María del Pilar
Contributor/s:
  • Antonio Jiménez, Angélica de
  • Greenhalgh, Chris
Item Type: Thesis (Doctoral)
Date: May 2003
Subjects:
Faculty: Facultad de Informática (UPM)
Department: Lenguajes y Sistemas Informáticos e Ingeniería del Software
Creative Commons Licenses: Recognition - No derivative works - Non commercial

Full text

[img]
Preview
PDF - Requires a PDF viewer, such as GSview, Xpdf or Adobe Acrobat Reader
Download (16MB) | Preview

Abstract

A priority in Intelligent Virtual Agents Systems (IVAS) and multi- Intelligent Virtual Agents Systems (mlVAS) is to introduce human-like sensitiva perception. Within the cognitive research área, there are some studies underwriting that human perception can be understood as a first level of an "awareness model" [Endsley 88], [Endsley 93]. FoUowing from this research, we have developed a human-like perceptual model based on one of the most successful awareness models in Computer Supported Cooperative Work (CSCW), called the Spatial Model of Interaction (SMI), which manages awareness in CoUaborative Virtual Environments (CVEs) through a set of key concepts. This perceptual model extends the key concepts of the SMI by introducing some human-like factors typical of human perception, which are reinterpreted with the aim of introducing them as the key concepts of an IVA human-like perceptual model. Agent perceptual capabilities should faithñilly resemble those of a human being. This is the first time that an awareness model designed for CSCW applications has been reinterpreted as an IVA perceptual model. We have focused on two senses -visión and hearing-, and some representative perceptual factors - like Sense Acuity or Sense Transition Región. Having formalised the perceptual model, we selected an appropriate agent architecture on which to use and test this perceptual model. We selected a vertical layered architecture, Nilsson's Triple Tower Model. The introduction of this perceptual model into the agent perceptual block has required the creation of a perceptual engine to maiiage not just the sensitive agent perception but also some other extemal and additional factors, such as attenuation. The agent's perception is composed of three modules: Sensitive Perception, Attenuation and Interna! Filtering. The sensitive perception module simulates the human process of perception by which organisms receive sensations from the environment, depending on some relevant sensorial concepts: Human Factors, Physical Factors, Object Factors and Adaptors. The implementation of our perceptual model has another important goal; our system introduces a semi-centralised approach to perception. This means that each agent is responsible for just part of its own perception, while the system is responsible for the rest. The final perceptual model allows an IVA to perceive its environment and surrounding objects in real-time with a human-like clarity of perception, gaining knowledge about its surroundings and experiencing any virtual simulated environment. The resulting perceptual model was designed to allow an IVA to be "aware" of the objects that the agent can perceive with a human-like clarity of perception Una prioridad en Sistemas de Agentes Virtuales Inteligentes (AVis) y sistemas multi-Agentes Virtuales Inteligentes (mAVIs), es introducir una percepción sensitiva similar a la que poseen los seres humanos. Algunos estudios dentro del área de investigación cognitiva introducen la percepción humana como el primer nivel de un modelo de "awareness^^^" [Endsley 88], [Endsley 93]. Basándonos en estos estudios, nosotros hemos desarrollado un modelo de percepción basado en uno de los más conocidos modelos de "awareness" desarrollados para trabajo colaborativo (CSCW). Este modelo es conocido "Modelo Espacial de Interacción" (MEI) y gobierna "awareness" en Entornos Virtuales Colaborativos (EVC). Este modelo de percepción extiende los conceptos fundamentales del MEI, introduciendo algunos factores propios de la percepción en humanos, reinterpreta estos conceptos clave con el fin de introducirlos en el sistema perceptivo de un AVI. De este modo, las capacidades perceptivas de un AVI serían similares a las de un humano. Esta es la primera vez que un miodelo de "awareness" diseñado para trabajo colaborativo (CSCW) es reinterpretado como un modelo de percepción para AVIs. Nos hemos centrado en los sentidos más utilizados en Entornos Virtuales (EV) por el momento — visión y oído - y en algxonos factores característicos del sistema perceptivo humano como son la Agudeza Sensitiva o la Región de Transición Sensorial. Una vez formalizado el modelo, hemos seleccionado una arquitectura de agentes para poder introducir, usar y validar nuestro modelo. De entre todas las posibles arquitecturas, hemos seleccionado el Modelo de Triple Torre de Nilsson. La introducción de este modelo de percepción dentro de la arquitectura de agente seleccionada ha requerido la creación de un "motor de percepción" que permita manejar no sólo la percepción sensitiva del agente sino también algunos factores externos y adicionales tales como la atenuación. Dentro del bloque de percepción del agente hemos querido distinguir tres módulos diferentes: Percepción Sensitiva, Atenuación y Filtros Internos. La percepción sensitiva simula el proceso de percepción humano por el cual los organismos sensoriales reciben sensaciones del entorno, dependiendo de algunos factores conceptos sensoriales relevantes: Factores Humanos, Factores del Objeto y Adaptadores. El modelo de percepción resultante permite que un AVI perciba, en tiempo real, el entorno que le rodea con una claridad de percepción similar a la de un ser humano mediante un. modelo semi-centratizado de percepción.

More information

Item ID: 4177
DC Identifier: http://oa.upm.es/4177/
OAI Identifier: oai:oa.upm.es:4177
Deposited by: Archivo Digital UPM
Deposited on: 09 Sep 2010 10:13
Last Modified: 20 Apr 2016 13:30
  • Logo InvestigaM (UPM)
  • Logo GEOUP4
  • Logo Open Access
  • Open Access
  • Logo Sherpa/Romeo
    Check whether the anglo-saxon journal in which you have published an article allows you to also publish it under open access.
  • Logo Dulcinea
    Check whether the spanish journal in which you have published an article allows you to also publish it under open access.
  • Logo de Recolecta
  • Logo del Observatorio I+D+i UPM
  • Logo de OpenCourseWare UPM