Decision Making in Partially Observable, Uncertain Worlds:
Exploring Insights from Multiple Communities
Barcelona, 18 July 2011
Thanks to all participants, invited speakers, and reviewers! Slides from the invited talks and the papers are available here.
Planning under uncertainty and partial observability is a key challenge for both single agent and multi-agent systems. While much progress has been made over the past decade in both areas, new breakthroughs are needed for scaling up to more realistic domains. In the related field of planning (classical, contingent, and conformant) modern solvers can handle much larger domains, under the simpler planning models. Moreover, scaling up of single and multi-agent partially observable problems has been done largely independently, as the two communities have little overlap.
In this workshop we seek to bring together members of these three communities that handle partial observability - conformant planning, POMDPs, and DEC-POMDPs - as well as people from other related communities, such as classical planning, MDPs, reinforcement learning, and distributed search, to discuss how each community may be able to leverage advances from the other. We will have invited talks by three experts in these areas: Hector Geffner, David Hsu, and Shlomo Zilberstein.
We hope you will join us in Barcelona!