Postulates for revising BDI structures

Abstract

The process of rationally revising beliefs in the light of new information is a topic of great importance and long-standing interest in artificial intelligence. Moreover, significant progress has been made in understanding the philosophical, logical, and computational foundations of belief revision. However, very little research has been reported with respect to the revision of other mental states, most notably propositional attitudes such as desires and intentions. In this paper, we present a first attempt to formulate a general framework for understanding the revision of mental states. We develop an abstract belief-desire-intention model of agents, and introduce a notion of rationality for this model. We then present a series of formal postulates characterizing the processes of adding beliefs, desires, and intentions, updating costs and values, and removing beliefs, desires, and intentions. We also investigate the computational complexity of several problems involving the abstract model and comment on algorithms for revision.

Refinement of Intentions

“…One possible relational postulate for revision is that the end intentions in the revised database should be a subset of the end intentions of the original database. There is currently little work on linking intention revision with instrumentality, with the exception of [11,19]. However, these contributions are still preliminary as many issues are not yet solved, such as the frame problem or the relation between basic and non-basic actions.…”

Plan Selection for Probabilistic BDI Agents

“…Here for convenience we call each ϕ i ← P i a ranked item. [18] proposes two revision methods to make the intention base consistent: i) dropping some intentions; ii) modifying some intentions (i.e., removing some ranked items in the modified intentions) 5 ; to make the intention base consistent. Now we reduce the uncertainty of plans into preferences on intentions.…”

“…In [4], the main focus is to measure the cost of a plan; [6] studies cooperative plan selection through trust among multi-agents; Other works like [20], [19] apply machine learning to derive the probability for the condition of a plan and use it for plan selection. In the meantime, intention revision, which is closely related to plan selection w.r.t the revision/selection methods, is also investigated by recent works in BDI systems, e.g., [22], [21], [7], [5], [18]. While these works studies on plan selection or intention revision, they seldom consider uncertainty that can emerge in agent's beliefs (see [13], [9], [12]).…”

A Higher-Order Agent Model with Contextual Planning Management for Ambient Systems

“…The on the fly revision of plans throughout the evolution of the agent is in fact a recent research topic in MAS approaches, e.g. [19]. In [8], the authors propose to revise the agent plans from the modifications of the agents goals, in particular from the fact that some plans corresponding to intentions can be conflictual.…”