Title: Improving plan execution robustness through capability aware maintenance of plans by BDI agents

Authors: Alan White; Austin Tate; Michael Rovatsos

Addresses: Artificial Intelligence Applications Institute, Centre for Intelligent Systems and their Applications, School of Informatics, University of Edinburgh, UK ' Artificial Intelligence Applications Institute, Centre for Intelligent Systems and their Applications, School of Informatics, University of Edinburgh, UK ' Artificial Intelligence Applications Institute, Centre for Intelligent Systems and their Applications, School of Informatics, University of Edinburgh, UK

Abstract: In a realistic environment, intentions of belief-desire-intention (BDI) agents may be threatened by exogenous change. Subsequent activity failure may incur debilitative consequences that hinder both recovery and subsequent goal achievement. Capability aware, maintaining plans (CAMP-BDI) embodies BDI agents with capability knowledge, allowing anticipation of threats to activity success and stimulating the proactive, preventative modification of intended plans. We describe resultant agent-level algorithms and supporting architecture, including extension to provide decentralised, distributed maintenance through structured messaging. Our results show superior goal achievement to a reactive equivalent in a stochastic environment, increasing with the likelihood of debilitative failure effects. We suggest CAMP-BDI offers a valuable approach towards robustness, particularly in tandem with reactive recovery methods.

Keywords: multi-agent teamwork; belief-desire-intention; BDI; planning; capability; robustness.

DOI: 10.1504/IJAOSE.2017.087679

International Journal of Agent-Oriented Software Engineering, 2017 Vol.5 No.4, pp.306 - 335

Accepted: 16 Jun 2017
Published online: 30 Oct 2017 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article