Automatic SAT-compilation of planning problems
Michael D. Ernst, Todd D. Millstein, Daniel S. Weld · 1997
Recent work by Kautz et al. provides tantalizing evidence that large, classical planning problems may be efficiently solved by translating them into propositional satisfiability problems, using stochastic search techniques, and translating the resulting truth assignments backinto plans for the original problems. We explore the space of such transformations, providing a simple framework that generates eight major encodings (generated by selecting one of four action representations and one of two frame axioms) and a number of subsidiary ones.