Evaluating coverage based intention selection

The Belief Desire Intention (BDI) agent paradigm provides a powerful basis for developing complex systems based on autonomous intelligent agents. These agents have, at any point in time, a set of intentions, the various tasks the agent is working on which represent the agent's multiple focus of attention. Despite its importance for intelligent behaviour, the problem of selecting which intention to progress at any point in time has received almost no attention and has been left to the programmer to resolve in an application-dependent manner. In this paper we implement and evaluate a previous proposal for domain-independent intention selection using the notion of plan ``coverage,'' as well as a slight variation which we predicted to perform better. We compare these with the commonly used intention selection mechanisms of First-In-First-OutFIFO and Round RobinRR. We show that the coverage-based technique performs better under all circumstances, but particularly with low coverage and volatile environments. Interestingly, we found that a simple one-step look-ahead applicability check is responsible for the largest part of the improvement. This is important in that this can readily be applied to FIFO and RR, giving an extremely simple and effective mechanism to be added to existing BDI frameworks.