An empirical evaluation of reasoning about resource conflicts

It is important that intelligent agents are able to pursue multiple goals in parallel, in a rational manner. This work experimentally evaluates mechanisms presented previously which allow agents to detect and deal with situations where multiple goals conflict over limited resources. We describe X-JACK, our extension to JACK, a state of the art agent development toolkit. X-JACK incorporates an explicit structure for goals and the reasoning to detect and classify resource conflicts. We compare X-JACK to JACK experimentally, under a range of situations designed to stress test the conflict reasoning algorithms, as well as situations designed to be more similar to real applications. We find that the cost of the additional reasoning is small, even with large numbers of conflicts to reason about. The benefit however is noticeable, and is statistically significant, even when the amount of conflict and parallelism is relatively small.