Generals Charge

This is the charge I for my generals examination, established on approximately 13 June 1999. The charge is effectively the question or challenge you respond to in your generals examination. In general, it defines the sub-discipline of interest, a body of literature to work from, and an approach to take in synthesizing these. In response to my charge, I wrote my generals paper and gave my generals presentation (for the examination itself).
Paper [4] presents CPlan, a powerful planning engine based on compilation to constraint satisfaction; however, the system relies on properties and heuristics encoded by hand. The other three papers address problem reformulation using abstraction, simplification, and static analysis. Do these papers suggest a way to automate the CPlan framework? Does the work on reformulation suggest more effective or general encodings for STRIPS planning problems? Previous STRIPS compilers generate a separate encoding for each plan length considered, increasing the number of steps until a plan is found; do reformulation techniques suggest an alternative to this method? What other domain knowledge might these techniques be able to extract?

Primary Papers

  1. Christensen, Jens. "A Hierarchical Planner that Generates its Own Hierarchies". In Proceedings of the Eighth National Conference on Artificial Intelligence, pages 1004-1009, Boston, MA, 1990.
  2. Crawford, James and Ginsberg, Matthew and Luks, Eugene and Roy, Amitabha. "Symmetry-Breaking Predicates for Search Problems". In Proceedings of the Fifth International Conference on Principles of Knowledge Representation and Reasoning. Morgan Kaufman, pages 148-159, San Francisco, CA, 1996.
  3. Fox, M. and Long, D. "The Automatic Inference of State Invariants in TIM". In Journal of Artificial Intelligence Research, 1998, Volume 9, pages 367-421.
  4. Van Beek, Peter and Chen, Xinguang. "CPlan: A Constraint Programming Approach to Planning". To appear AAAI-99.

Steve Wolfman
15 August 2003