Automatically generating abstractions for planning
Artificial Intelligence
Downward refinement and the efficiency of hierarchical problem solving
Artificial Intelligence
State-variable planning under structural restrictions: algorithms and complexity
Artificial Intelligence
Artificial Intelligence - Special issue on heuristic search in artificial intelligence
Sokoban: enhancing general single-agent search methods using domain knowledge
Artificial Intelligence - Special issue on heuristic search in artificial intelligence
Exhibiting Knowledge in Planning Problems to Minimize State Encoding Length
ECP '99 Proceedings of the 5th European Conference on Planning: Recent Advances in AI Planning
Decomposition of planning problems
AI Communications
Factored planning: how, when, and when not
AAAI'06 Proceedings of the 21st national conference on Artificial intelligence - Volume 1
Macro-FF: improving AI planning with automatically learned macro-operators
Journal of Artificial Intelligence Research
Temporal planning using subgoal partitioning and resolution in SGPlan
Journal of Artificial Intelligence Research
Planning with abstraction hierarchies can be exponentially less efficient
IJCAI'95 Proceedings of the 14th international joint conference on Artificial intelligence - Volume 2
WoLLIC '09 Proceedings of the 16th International Workshop on Logic, Language, Information and Computation
The computational complexity of avoiding spurious states in state space abstraction
Artificial Intelligence
Completeness-Preserving Pruning for Optimal Planning
Proceedings of the 2010 conference on ECAI 2010: 19th European Conference on Artificial Intelligence
Analyzing search topology without running any search: on the connection between causal graphs and h+
Journal of Artificial Intelligence Research
IJCAI'11 Proceedings of the Twenty-Second international joint conference on Artificial Intelligence - Volume Volume Three
Hi-index | 0.00 |
Although even propositional STRIPS planning is a hard problem in general, many instances of the problem, including many of those commonly used as benchmarks, are easy. In spite of this, they are often hard to solve for domain-independent planners, because the encoding of the problem into a general problem specification formalism such as STRIPS hides structure that needs to be exploited to solve problems easily. We investigate the use of automatic problem transformations to reduce this "accidental" problem complexity. The main tool is abstraction: we identify a new, weaker, condition under which abstraction is "safe", in the sense that any solution to the abstracted problem can be refined to a concrete solution (in polynomial time, for most cases) and also show how different kinds of problem reformulations can be applied to create greater opportunities for such safe abstraction.