A Robust Architecture for Multiple-Agent Reinforcement Learning

Hi,

I'm looking for a soft copy of the following document:

"A Robust Architecture for Multiple-Agent Reinforcement Learning"

J. Andrew Bagnell

This report is submitted in partial fulfillment for the
requirements for Highest Honors

Fall Term 1998

Abstract

This thesis addresses the design and implementation of a robust
architecture for improving the performance of physically realized
behavior-based mobile robots through parallel reinforcement
learning (RL). Custom hardware to support radio-communication
links between agents, custom software to implement isochronous
packet-switched communication, and the parallel learning scheme
are all described. Finally this report presents the results of
the application of this architecture, integrated with the best-
studied reinforcement learning algorithm, Q-learning.

I hope someone can help.

Thanks,

Uri

kartoun@xxxxxxxxx
.