ADAPT 2009 Workshop: Overview

Agent Design: Advancing from Practice to Theory
AAMAS 2009 -- Budapest, Hungary

Overview

Highlights

Schedule

Submission

Schedule - See you in Budapest!

MOTIVATION

 AAMAS is the premier international conference for both theoretical and applied agent-based research. However, in today's agents community, theory and practice seem to be walking separate paths with no clear feedback between the two.

The ADAPT workshop hopes to bridge this gap by fostering an exchange where the lessons learned from taking theory to practice are used to understand the limitations of the theoretical models and the lessons learned directly from practice and used to discover the issues that theoretical models need to address.

Over the last several years, members of the agents community have been involved with real-world systems in robotics, decision-support agents, personal-assistant agents in military, space and commercial domains. Many of the lessons learned from these endeavors involve the surprises, difficulties and flaws. These lessons are often not shared because they bring about questions for which an answer is not apparent. Nevertheless, it is important for our community to know and discuss these challenges if we are to produce the ideas and technology that are transformative and ultimately, utilized by practitioners.

Some of these challenges may include:

   * Giving users a language to compactly communicate their problems; problem description
   * Translating a problem description into a chosen formalism; modeling
   * The lack of information to populate a model
   * The effects of bounded memory and processing on algorithms
   * Real-world communication channels
   * Communicating with a human user, e.g., giving instructions, receiving updates
   * Getting real information from the environment; sensing
   * User error or misinterpretation of instructions

An important aspect of this workshop is not only to discuss these challenges, but also to develop directions for existing and new theoretical models so that algorithms and approaches that result can actually function in practice.