Acknowledgements

The authors would like to thank Eugene Fink, Sven Koenig, Illah Nourbakhsh, Joseph O'Sullivan, Gary Pelton and the anonymous reviewers for feedback on this article. We would also like to thank the members of the Xavier and PRODIGY groups for feedback, comments and criticism on our research.

This research is sponsored in part by (1) the National Science Foundation under Grant No. IRI-9502548, (2) by the Defense Advanced Research Projects Agency (DARPA), and Rome Laboratory, Air Force Materiel Command, USAF, under agreement number F30602-95-1-0018, (3) the Natural Sciences and Engineering Council of Canada (NSERC), and (4) the Canadian Space Agency (CSA). The views and conclusions contained herein are those of the authors and should not be interpreted as necessarily representing the official policies or endorsements, either expressed or implied, of the NSF, DARPA, Rome Laboratory, the U.S. Government, NSERC or the CSA.

Biographies

Karen Zita Haigh is currently completing her Ph.D. in Computer Science at Carnegie Mellon University in Pittsburgh, Pennsylvania. Her undergraduate degree was completed in 1992 at the University of Ottawa in Ottawa, Ontario, Canada. Her thesis is a robot learning system that uses feedback from execution experience to improve efficiency of generated plans. It creates situation-dependent costs so that plans are tailored to particular situations, and is used on Xavier's route planner and in ROGUE. She also built analogical reasoning system to automatically generate high-quality routes in a city map. Her research interests include planning, machine learning, and robotics.

Manuela M. Veloso is a Finmeccanica Associate Professor in the Computer Science Department at Carnegie Mellon University. She received her Ph.D. in Computer Science from CMU in 1992. Dr. Veloso received the NSF Career Award and was the recipient of the Finmeccanica Chair in 1995. In 1997, she was awarded the Allen Newell Excellence in Research Award by the School of Computer Science at CMU. Dr. Veloso is the author of a monograph on ``Planning by Analogical Reasoning.'' She is co-editor of two books, ``Symbolic and Visual Learning'' and ``Topics of Case-based Reasoning.'' Dr. Veloso's research involves the integration of planning, execution and learning in dynamic environments, and in particular with multiple agents. She investigates memory-based machine learning techniques for the processing and reuse of problem experience.