Backtrack-free search for real-time constraint satisfaction

Abstract

A constraint satisfaction problem (CSP) model can be preprocessed to ensure that any choices made will lead to solutions, without the need to backtrack. This can be especially useful in a real-time process control or online interactive context. The conventional machinery for ensuring backtrack-free search, however, adds additional constraints, which may require an impractical amount of space. A new approach is presented here that achieves a backtrack-free representation by removing values. This may limit the choice of solutions, but we are guaranteed not to eliminate them all. We show that in an interactive context our proposal allows the system designer and the user to collaboratively establish the tradeoff in space complexity, solution loss, and backtracks. © Springer-Verlag Berlin Heidelberg 2004.