Expected Value of Communication for Planning in Ad Hoc Teamwork

A desirable goal for autonomous agents is to be able coordinate on the fly with previously unknown teammates. Known as “ad hoc teamwork”, enabling such a capability has been receiving increasing attention in research community. One of central challenges ad teamwork quickly recognizing current plans other and planning accordingly. In this paper, we focus scenario which teammates can communicate one another, but only at cost. Thus, they must carefully balance plan recognition based observations vs. that communication. This paper proposes new metric evaluating how similar are two policies teammate may following - Expected Divergence Point (EDP). We then present novel algorithm teamwork, determining query ask demonstrate effectiveness range increasingly general communication problems.