by Michael Beetz and Maren Bennewitz
Abstract:
Scheduling the tasks of an autonomous robot office courier and carrying out the scheduled tasks reliably and efficiently pose challenging problems for autonomous robot control. The controller has to accomplish longterm efficiency rather than optimize problem-solving episodes. It also has to exploit opportunities and avoid problems flexibly because often the robot is forced to generate schedules based on partial information. We propose to implement the controller for scheduled activity by employing concurrent reactive plans that reschedule the course of action whenever necessary and while performing their actions. The plans are represented modularly and transparently to allow for easy transformation. Scheduling and schedule repair methods are implemented as plan transformation rules.
Reference:
Michael Beetz and Maren Bennewitz, "Planning, Scheduling, and Plan Execution for Autonomous Robot Office Couriers", In Proceedings of the workshop "Integrating Planning, Scheduling and Execution in Dynamic and Uncertain Environments" at the Fourth International Conference on AI in Planning Systems (AIPS), AAAI Press, vol. Workshop Notes 98-02, 1998.
Bibtex Entry:
@InProceedings{Bee98Pla,
author = {Michael Beetz and Maren Bennewitz},
title = {Planning, Scheduling, and Plan Execution for Autonomous Robot Office Couriers},
booktitle = {Proceedings of the workshop ``Integrating Planning, Scheduling and Execution in Dynamic and Uncertain Environments'' at the Fourth International Conference on AI in Planning Systems (AIPS)},
editor = {R. Bergmann and A. Kott},
volume = {Workshop Notes 98-02},
year = {1998},
publisher = {AAAI Press},
bib2html_pubtype = {Refereed Conference Paper},
bib2html_rescat = {Plan-based Robot Control},
bib2html_groups = {IAS},
bib2html_funding = {ignore},
bib2html_keywords = {Robot, Planning},
abstract = {Scheduling the tasks of an autonomous robot office courier and carrying out the scheduled tasks
reliably and efficiently pose challenging problems for autonomous robot control. The controller has
to accomplish longterm efficiency rather than optimize problem-solving episodes. It also has to
exploit opportunities and avoid problems flexibly because often the robot is forced to generate
schedules based on partial information. We propose to implement the controller for scheduled
activity by employing concurrent reactive plans that reschedule the course of action whenever
necessary and while performing their actions. The plans are represented modularly and transparently
to allow for easy transformation. Scheduling and schedule repair methods are implemented as plan
transformation rules.}
}