HAC: A unified view of reactive and deliberative activity

Marc S. Atkin, David L. Westbrook, Paul R Cohen

Research output: Chapter in Book/Report/Conference proceedingConference contribution

3 Citations (Scopus)

Abstract

The Hierarchical Agent Control Architecture (HAC) is a general toolkit for specifying an agent's behavior. By organizing the hierarchy around tasks to be accomplished, not the agents themselves, it is easy to incorporate multi-agent actions and planning into the architecture. In addition, HAC supports action abstraction, resource management, sensor integration, and is well suited to controlling large numbers of agents in dynamic environments. Unlike other agent architectures, HAC does not conceptually distinguish reactive from deliberative, or single-agent from multi-agent behaviors. There is no pre-determined number of cognitive "levels" in the hierarchy|all actions share the same form and are implemented with the same functions. GRASP is a multi-goal partial hierarchical planner that has been implemented using the HAC framework. GRASP illustrates two points: Firstly, that the same HAC mechanisms used to write reactive actions can be used to implement a cognitive activity such as planning; and secondly, that the problem of integrating reactive and deliberative behavior itself can be viewed as having to simultaneously achieve multiple goals. Throughout the paper, we show how HAC and GRASP were applied to an adversarial, real-time domain based on the game of "Capture the Flag".

Original languageEnglish (US)
Title of host publicationLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Pages92-107
Number of pages16
Volume2103 LNAI
StatePublished - 2001
Externally publishedYes
EventBalancing Reactivity and Social Deliberation in Multi-Agent Systems - 14th European Conference on Artificial Intelligence, ECAI 2000 - Berlin, Germany
Duration: Aug 20 2000Aug 25 2000

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume2103 LNAI
ISSN (Print)03029743
ISSN (Electronic)16113349

Other

OtherBalancing Reactivity and Social Deliberation in Multi-Agent Systems - 14th European Conference on Artificial Intelligence, ECAI 2000
CountryGermany
CityBerlin
Period8/20/008/25/00

Fingerprint

Planning
Architecture
Agent Architecture
Resource Management
Dynamic Environment
Game
Real-time
Partial
Sensor
Sensors
Hierarchy
Abstraction
Form
Framework

ASJC Scopus subject areas

  • Computer Science(all)
  • Theoretical Computer Science

Cite this

Atkin, M. S., Westbrook, D. L., & Cohen, P. R. (2001). HAC: A unified view of reactive and deliberative activity. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 2103 LNAI, pp. 92-107). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 2103 LNAI).

HAC : A unified view of reactive and deliberative activity. / Atkin, Marc S.; Westbrook, David L.; Cohen, Paul R.

Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol. 2103 LNAI 2001. p. 92-107 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 2103 LNAI).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Atkin, MS, Westbrook, DL & Cohen, PR 2001, HAC: A unified view of reactive and deliberative activity. in Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). vol. 2103 LNAI, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 2103 LNAI, pp. 92-107, Balancing Reactivity and Social Deliberation in Multi-Agent Systems - 14th European Conference on Artificial Intelligence, ECAI 2000, Berlin, Germany, 8/20/00.
Atkin MS, Westbrook DL, Cohen PR. HAC: A unified view of reactive and deliberative activity. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol. 2103 LNAI. 2001. p. 92-107. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
Atkin, Marc S. ; Westbrook, David L. ; Cohen, Paul R. / HAC : A unified view of reactive and deliberative activity. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol. 2103 LNAI 2001. pp. 92-107 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
@inproceedings{f6c804e0ab464504ab9c064d52b414fa,
title = "HAC: A unified view of reactive and deliberative activity",
abstract = "The Hierarchical Agent Control Architecture (HAC) is a general toolkit for specifying an agent's behavior. By organizing the hierarchy around tasks to be accomplished, not the agents themselves, it is easy to incorporate multi-agent actions and planning into the architecture. In addition, HAC supports action abstraction, resource management, sensor integration, and is well suited to controlling large numbers of agents in dynamic environments. Unlike other agent architectures, HAC does not conceptually distinguish reactive from deliberative, or single-agent from multi-agent behaviors. There is no pre-determined number of cognitive {"}levels{"} in the hierarchy|all actions share the same form and are implemented with the same functions. GRASP is a multi-goal partial hierarchical planner that has been implemented using the HAC framework. GRASP illustrates two points: Firstly, that the same HAC mechanisms used to write reactive actions can be used to implement a cognitive activity such as planning; and secondly, that the problem of integrating reactive and deliberative behavior itself can be viewed as having to simultaneously achieve multiple goals. Throughout the paper, we show how HAC and GRASP were applied to an adversarial, real-time domain based on the game of {"}Capture the Flag{"}.",
author = "Atkin, {Marc S.} and Westbrook, {David L.} and Cohen, {Paul R}",
year = "2001",
language = "English (US)",
isbn = "3540423273",
volume = "2103 LNAI",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
pages = "92--107",
booktitle = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

}

TY - GEN

T1 - HAC

T2 - A unified view of reactive and deliberative activity

AU - Atkin, Marc S.

AU - Westbrook, David L.

AU - Cohen, Paul R

PY - 2001

Y1 - 2001

N2 - The Hierarchical Agent Control Architecture (HAC) is a general toolkit for specifying an agent's behavior. By organizing the hierarchy around tasks to be accomplished, not the agents themselves, it is easy to incorporate multi-agent actions and planning into the architecture. In addition, HAC supports action abstraction, resource management, sensor integration, and is well suited to controlling large numbers of agents in dynamic environments. Unlike other agent architectures, HAC does not conceptually distinguish reactive from deliberative, or single-agent from multi-agent behaviors. There is no pre-determined number of cognitive "levels" in the hierarchy|all actions share the same form and are implemented with the same functions. GRASP is a multi-goal partial hierarchical planner that has been implemented using the HAC framework. GRASP illustrates two points: Firstly, that the same HAC mechanisms used to write reactive actions can be used to implement a cognitive activity such as planning; and secondly, that the problem of integrating reactive and deliberative behavior itself can be viewed as having to simultaneously achieve multiple goals. Throughout the paper, we show how HAC and GRASP were applied to an adversarial, real-time domain based on the game of "Capture the Flag".

AB - The Hierarchical Agent Control Architecture (HAC) is a general toolkit for specifying an agent's behavior. By organizing the hierarchy around tasks to be accomplished, not the agents themselves, it is easy to incorporate multi-agent actions and planning into the architecture. In addition, HAC supports action abstraction, resource management, sensor integration, and is well suited to controlling large numbers of agents in dynamic environments. Unlike other agent architectures, HAC does not conceptually distinguish reactive from deliberative, or single-agent from multi-agent behaviors. There is no pre-determined number of cognitive "levels" in the hierarchy|all actions share the same form and are implemented with the same functions. GRASP is a multi-goal partial hierarchical planner that has been implemented using the HAC framework. GRASP illustrates two points: Firstly, that the same HAC mechanisms used to write reactive actions can be used to implement a cognitive activity such as planning; and secondly, that the problem of integrating reactive and deliberative behavior itself can be viewed as having to simultaneously achieve multiple goals. Throughout the paper, we show how HAC and GRASP were applied to an adversarial, real-time domain based on the game of "Capture the Flag".

UR - http://www.scopus.com/inward/record.url?scp=84864848649&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84864848649&partnerID=8YFLogxK

M3 - Conference contribution

AN - SCOPUS:84864848649

SN - 3540423273

SN - 9783540423270

VL - 2103 LNAI

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 92

EP - 107

BT - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

ER -