WS 2: Models and Paradigms for Planning under Uncertainty: a Broad Perspective

Highlight: Invited Talk by Ronen I. Brafman on Planning under Uncertainty: Reductions, Replanning, Simplifications

Probabilistic planning research mainly focused on a subset of uncertainty models, e.g., probabilistic ones such as (partially observable) Markov Decision Processes, or set-based ones such as non-deterministic and conformant planning. However, many other frameworks for describing uncertainty have been studied in Artificial Intelligence in general (e.g., possibilities, fuzzy logics, imprecise probabilities, Dempster-Shafer’s belief function), and other communities confronted with real-world problems like risk management use these models in their decision-making systems. We believe a cross-pollination of research approaches and methodologies from these relevant, yet mostly distant, paradigms will be very beneficial for the ICAPS community, leading to new ideas for tackling realistic planning problems.

The full MPPU proceedings are now available.

Schedule

Objectives and Topics

Great strides have been made in automated AI planning under uncertainty in recent years, including symbolic and compact representations of planning problems and very efficient techniques for solving them. The effectiveness of these methods has been demonstrated in the past International Planning Competitions, and to some extent, in real-world applications such as navigation tasks, space operations, railway control, and rescue/evacuation tasks.

However, there are several remaining challenges for developing uncertainty models for the planning systems. When deployed in the real world, these systems often face a constantly changing environment, whose evolution is not deterministic. In addition to the environmental dynamics, planning systems must also deal with the partial knowledge about their surroundings, their models of the environment, and their goals in that environment. Addressing all these aspects successfully may require a range of modeling tools from precise and imprecise probabilities to fuzzy and possibilistic logic.

The aim of this workshop is to discuss various models and paradigms for planning under uncertainty in a broad sense, including but also going beyond the traditional probabilistic planning paradigms.

Relevant topics include but are not limited to:

• probabilistic or possibilistic (partially observable) Markov Decision Processes
• non-probabilistic uncertainty models for planning and algorithms
• conformant planning
• imprecise probability models and planning
• fuzzy and possibilistic logic
• planning/replanning with deterministic planners
• determinization-based approaches
• modeling imperfect actuators and/or sensors and controller synthesis
• belief-desire-intention (BDI) models, Dempster-Shafer theory
• default reasoning and belief revision models
• qualitative uncertainty models (e.g., Qualitative-Process (QP) theory, qualitative probability models)
• learning uncertainty models for planning

Accepted Papers

• A Relevance-Based Compilation Method for Conformant Probabilistic Planning.
  Ran Taig and Ronen I. Brafman.
• To Share or Not to Share? The Single Agent in a Team Decision Problem.
  Ofra Amir, Barbara Grosz and Roni Stern.
• Computing Contingent Plans via Fully Observable Non-Deterministic Planning.
  Christian Muise, Vaishak Belle and Sheila A. Mcilraith.
• Compiling Contingent Planning into Classical Planning: New Translations and Results.
  Hector Palacios, Alexandre Albore and Hector Geffner.
• Structured Possibilistic Planning using Decision Diagrams.
  Nicolas Drougard, Florent Teichteil-Königsbuch and Jean-Loup Farges.
• Monte-Carlo Tree Search: To MC or to DP?
  Zohar Feldman and Carmel Domshlak.
• A Contingent Planning-Based POMDP Replanner.
  Ronen Brafman, Alexander Gorohovski and Guy Shani.
• Diagnostic Problem Solving via Planning with Ontic and Epistemic Goals.
  Jorge Baier, Brent Mombourquette and Sheila Mcilraith.
• Factored Markov Decision Process with Imprecise Transition Probabilities.
  Karina V. Delgado, Leliane N. de Barros, Scott Sanner and Fabio Cozman.

Submission Procedure

Paper submission is in PDF only. Please format submissions in AAAI style . Refer to the author instructions on the AAAI web site for detailed formatting instructions and LaTeX style files. Final papers will be in the same format, and authors may submit long papers (8 pages plus up to one page of references) or short papers (4 pages plus up to one page of references).

Papers must be submitted by March 20th, 2014. All ICAPS deadlines refer to 23:59 in the UTC-12 time zone (i.e., if the deadline has not yet passed at some place in the world, you are on time.)

Paper submissions should be made through the workshop EasyChair web site.

Tutorial

We propose to organize a 1.5-hour tutorial on various models for planning under uncertainty. The covered topic will be:

• probabilistic planning (Andrey Kolobov)
• non-deterministic planning (Ugur Kuter)
• possibilistic planning (Florent Teichteil)
• dealing with partial observability (Guy Shani)

Presentation Guidelines

Every accepted paper will be presented both orally during the workshop and as a poster during the main conference. Note that the poster presentation is not compulsory, it is simply an additional opportunity (see for more information). During the workshop, oral presentations are 25 minutes total for long papers and 15 minutes total for short papers, including questions and answers.

Important Dates

• Paper submission: March 7th, 2014
• Notification of acceptance: March 20th, 2014
• Camera-ready paper submission: TBD
• Workshop date: June 23rd, 2014

Workshop Program Chairs

• Andrey Kolobov (MSR Redmond, USA), akolobov at microsoft dot com
• Ugur Kuter (SIFT, USA), ukuter at sift dot net
• Florent Teichteil-Königsbuch (ONERA, France), florent dot teichteil at onera dot fr

Program Committee (to be completed)

• Alexandre Albore (ONERA, France)
• Daniel Bryce (SIFT, USA)
• Juergen Dix (Clausthal University of Technology, UK)
• Malik Ghallab (LAAS, France)
• Andrey Kolobov (Microsoft Research, USA)
• Ugur Kuter (SIFT, USA)
• Steven Schockaert (Cardiff University, UK)
• Guy Shani (Ben-Gurion University of the Negev, Israel)
• Florent Teichteil-Königsbuch (ONERA, France)
• Paolo Traverso (IRST, Italy)