....¹

This transformation can increase the size of the DBNs exponentially unless context-specific DBNs are used (Boutilier, Friedman, Goldszmidt, & Koller, 1996).

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... size.²

It is important to note that the existentially quantified goal formula for colored Blocksworld, when grounded, can be excessively long. This fact is a serious bottleneck for larger instances of this domain. Planners that avoid grounding should have a benefit here, but did not in the competition because our plan validator grounded the goal formula.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... graphs.³

The optimal scores do not necessarily apply to Planner Q, which is a conformant planner.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.