Reduce planning for quantized Boolean formulas

Question

Why don't we reduce the planning task in AI for the TQBF Version SAT in practical solutions.

Many planning problems are in practice "compiled" or reduced to the SAT problem, which, in turn, is solved by SAT Solvers. The problem is that since planning is PSPACE Complete and SAT is NP Complete, an exponential number of literals may be required.

Why then do practical planners use this approach? Why don't we solve the TQBF SAT problem and then “compile” Planning to TQBF, which should only take polynomial time?

Answer 1

This is already done.

Typically, TQBF is used to model consistent planning, but there are encodings of tasks with purely propositional logic (formulas of polynomial size) TQBF.

The main disadvantage is that although we have a much smaller formula, this is not easy to solve. A TQBF solution is not as close as researching a SAT solution, and planning like a TQBF is still slightly behind in performance.

Here is one post that describes such a conversion (mine):

http://users.cecs.anu.edu.au/~ssanner/ICAPS_2010_DC/Abstracts/cashmore.pdf

Answer 2

SAT . ( , SAT ).

, SAT, , SAT-. , , , , .

, SAT. , , (SAT- LISP, ).

TQBF?

, , ( ). , , , , , TQBF- , - , , .

TQBF-solver. , - TQBF- ( ). , , SAT ( , , Deolalikar ).

, . , .