Reactive Planning with Weak Plan Instances

Reactive planning serves for fast but intelligent reasoning of intelligent artificial agents. In this paper we show some weak spots of such kind of planning together with our approach how to solve them. To be specific, we present our preliminary work on another BDI based agent that shows more flexibility in its reasoning than todaypsilas typical BDI agents. Their principles lie on late variables binding and restriction of such bindings during agents´ run. We describe here basic principles of such reasoning and provide a case study that proves that this approach is able to solve situations where classical BDI systems fail.