A flexible plan step execution model for BDI agents

Authors:
Jacqueline Jarvis;Dennis Jarvis;Ralph Rönnquist;Lakhmi Jain
Affiliations:
(Correspd. E-mail: d.jarvis@cqu.edu.au) Centre for Intelligent and Networked Systems, CQUniversity, Rockhampton, Queensland 4702, Australia;Centre for Intelligent and Networked Systems, CQUniversity, Rockhampton, Queensland 4702, Australia;Intendico Pty. Ltd. Suite 40, 85 Grattan St. Carlton Victoria 3053, Australia;KES Centre, School of Electrical and Information Engineering, University of South Australia, Levels Campus, South Australia 5095, Australia
Venue:
Multiagent and Grid Systems - Innovations in intelligent agent technology
Year:
2008

Citing 7
Cited 0

Soar Papers: Research on Integrated Intelligence

Soar Papers: Research on Integrated Intelligence
A Formal Specification of dMARS

ATAL '97 Proceedings of the 4th International Workshop on Intelligent Agents IV, Agent Theories, Architectures, and Languages
Experiences with the design and implementation of an agent-based autonomous UAV controller

Proceedings of the fourth international joint conference on Autonomous agents and multiagent systems
A Flexible BDI Architecture Supporting Extensibility

IAT '05 Proceedings of the IEEE/WIC/ACM International Conference on Intelligent Agent Technology
Comparative analysis of frameworks for knowledge-intensive intelligent agents

AI Magazine - Special issue on achieving human-level AI through integrated systems and research
A team-based holonic approach to robotic assembly cell control

Journal of Network and Computer Applications - Special issue: Innovations in agent collaboration
Using agent teams to model enterprise behaviour

Multiagent and Grid Systems - Innovations in intelligent agent technology

Quantified Score

Hi-index	0.00

Visualization

Abstract

For the past 20 years, BDI (Belief, Desire, Intention) frameworks such as PRS [1], dMARS [2] and JACK [3] have provided, together with Soar [4], the two environments of choice for the development of knowledge rich, industrial strength intelligent agent applications [5]. However, we have observed that while the BDI model of plan execution works well for the tactical reasoning component of such applications, operational reasoning often requires a richer execution model. In this paper, we present an alternative, but complementary model for plan step execution by BDI agents. In the BDI model, plan steps either succeed or fail; if a plan step fails, then the plan fails and reconsideration of the current goal may occur. We have found that this approach is problematic when used for applications where resource contention is a regular occurrence, such as in manufacturing execution [6]. In these situations, it is necessary to review progress after each step, regardless of the step outcome. Our alternative model for plan step execution allows for the explicit modelling of the plan step lifecycle and the utilisation of infrastructure to manage the progression of that lifecycle. The model is realised using the JACK™ Intelligent Agents (JACK) product suite [3] and its feasibility is demonstrated through the development of an execution system for a robotic assembly cell.