Participants

Although twenty teams registered for the competition initially, seven teams from four continents ultimately competed. They produced ten different planners, which were evaluated on various subsets of the problem domains. The groups and their planners were:

• Group C. UMass

  Participants: Zhengzhu Feng (University of Massachusetts) and Eric Hansen (Mississippi State University).

  Description: Symbolic heuristic search.

• Group E. Dresden (“FLUCAP”, formerly “FCPlanner”)

  Participants: Eldar Karabaev and Olga Skvortsova (both of Dresden University of Technology).

  Description: First-order heuristic search.

• Group G. ANU (“NMRDPP”)

  Participants: Charles Gretton, David Price and Sylvie Thiébaux (all of The Australian National University).

  Descriptions: G1: Planner primarily designed for domains with non-Markovian rewards, and G2: NMRDPP augmented with control knowledge.

• Group J. Purdue

  Participants: SungWook Yoon, Alan Fern and Robert Givan (all of Purdue University).

  Descriptions: J1: Human-written policy in Classy's policy language (“Purdue-Humans”), J2: Offline policy iteration by reduction to classification, automatically acquiring a domain-specific policy (“Classy”), and J3: Deterministic replanner using FF (“FF-rePlan”).

• Group P. Simón Bolívar (“mGPT”)

  Participants: Blai Bonet (Universidad Simón Bolívar) and Héctor Geffner (Universitat Pompeu Fabra).

  Description: Labeled RTDP with lower bounds extracted from the problem description.

• Group Q. Michigan Tech (“Probapop”)

  Participants: Nilufer Onder, Garrett C. Whelan and Li Li (all of Michigan Technological University).

  Description: POP-style planner (no sensing).

• Group R. CERT

  Participants: Florent Teichteil-Königsbuch and Patrick Fabiani (both of CERT).

  Description: Probabilistic reachability heuristic and DBNs.