A Framework for Plan Library Evolution in BDI Agent Systems

The Belief-Desire-Intention (BDI) paradigm is a flexible framework for representing intelligent agents. Practical BDI agent systems rely on a static plan library to reduce the planning problem to the simpler problem of plan selection. However, fixed pre-defined plan libraries are unable to adapt to fast-changing environments pervaded by uncertainty. In this paper, we advance the state-of-the-art in BDI agent systems by proposing a plan library evolution architecture with mechanisms to incorporate new plans (plan expansion) and drop old/unsuitable plans (plan contraction) to adapt to changes in a realistic environment. The proposal follows a principled approach to define plan library expansion and contraction operators, motivated by postulates that clearly highlight the underlying assumptions, and quantified by decision-support measures of temporal information. In particular, we demonstrate the feasibility of the proposed contraction operator by presenting a multi-criteria argumentation based decision making to remove plans exemplified in a planetary vehicle scenario.