The Probablistic Planning Track of the 2004 International Planning Competition.

Organizers

Michael L. Littman and Håkan L. S. Younes.

Students: John Asmuth, Paul Batchis, David Weissman.

Documentation and Links

NEW (Oct. 2007): MDPSim 2.1: The PPDDL Plan Evaluation Simulator.
Talk slides from the results presentation at ICAPS in Vancouver.
Data Poster from ICAPS.
The PPDDL evaluation simulator is now available in gzipped, tar format: mdpsim-1.3.tar.gz.
Domain generators used for the competition: Blocksworld, Boxworld, Fileworld.
Competition domains (observable versions)
More example domains
Summary of runs by team/problem. Columns: 1: ID; 2: group; 3: problem; 4: attempts; 5: fail; 6: success; 7: time average (?); 8: value
Competition log files (compressed tar)
The Winners!
Online Proceedings.
IPC 2004 Probabilistic Planning Track: FAQ 1.01 (November 1, 2003)
IPC 2004 Probabilistic Planning Track: FAQ 0.5 (September 13, 2003)
IPC 2004 Probabilistic Planning Track: FAQ 0.1 (May 7, 2003)

Software

To install the plan evaluator (under Unix), save the tar file (for example: use wget mdpsim-1.3.tar.gz or save from browser). Uncompress (for example: gunzip mdpsim-1.3.tar.gz). Extract everything (for example: tar xvf mdpsim-1.3.tar). Look at the file mdpsim/README for more instructions.

Participants

Zhengzhu Feng
UMass (with Eric Hansen)
Symbolic heuristic search
Olga Skvortsova
Dresden University of Technology (with Eldar Karabaev)
First-order value iteration in fluent calculus
Pascal Poupart
University of Toronto
Greedy linear value-approximation
Charles Gretton
Australian National University
Re-engineered version of NMRDPP
Robert L Givan
Purdue (with Alan Fern, Sungwook Yoon)
Policy iteration with policy language bias (learn from sample problems)
Blai Bonet
Universidad Simon Bolivar
GPT: value function; simulation; plan search
Nilufer Onder
Michigan Technological University (with Li and Garrett Whelan)
POP-style planner (possibly goal oriented, no sensing only)
Florent Teichteil-Konigsbuch
ONERA-DCSD (with Patrick Fabiani)
Explicit state enumeration and DBNs, value functions
This material is based upon work supported by the National Science Foundation under Grant No. 0315909. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation.