Plan Execution: A Reality Check
Bridging the gap between plans and their execution.
A workshop held in conjunction with
The 15th International Conference on Planning and Scheduling
ICAPS 2005
Monterey, California, USA.
Planning for realistic domains presents a varied set of challenges. Key among
those challenges is understanding and representing the execution time behavior
of the generated plans. Recent experiences in designing and deploying planning
systems provide significant insight into the execution of plans generated by
automated planners. This experience strongly suggests the presence of a gap
between how plan execution is treated in the plan generation process, and what
happens when the resulting plan is actually executed.
For automated planning systems to be successful in the real world, it is
essential that the nature of this gap be understood and some techniques for
bridging it be developed. Some of the relevant issues include the following
questions:
1. Is there a fundamental problem in our understanding of
what planning is versus what execution is?
2. What is the range of possible semantics for execution?
Are current domain modeling languages adequate to the
task of representing execution?
3. How do planning systems fit architecturally with
executives and hardware controllers? What are the
strengths and weaknesses of current implementations,and
where is more work needed?
4. How do planners cope with the mismatch between their
representation of the domain and reality? (in terms of
time latency, inaccuracies in modeling etc..) Is there a
fundamental difference between how this prediction
uncertainty is handled in control, and how it should be
handled for planning?
5. What tools and practices may be adopted to bridge the gap?
What can we learn from case studies, deployment
experiences and other associated areas (such as hybrid
controller design, real-time controls etc).
This workshop will bring together researchers who are working on answering
these questions to discuss these issues as well as real systems that are under
development. Those wishing to attend are encouraged to submit either a long
paper of up to 8 pages in AAAI format, or a short position paper of up to 3
pages. Please email pdf or ps versions of the papers to "sailesh at
email.arc.nasa.gov"
Important Dates:
Submissions due: March 21, 2005.
Notifications of acceptance: April 11, 2005
Camera-ready copy due: April 18, 2005.
Organizing Committee:
Michael Beetz, Technical University Munich.
Gautam Biswas, Vanderbilt University.
Mark Boddy, Adventium Labs.
Felix Ingrand, LAAS.
Nicola Muscettola, NASA Ames.
Issa A. D. Nesnas, JPL.
Sailesh Ramakrishnan, QSS Group Inc, NASA Ames.
_______________________________________________
Haskell mailing list
[email protected]
http://www.haskell.org/mailman/listinfo/haskell
