This report, prepared for the Computer Science Laboratory at SRI International, covers the development of a Learning component for PAGODA, a modular architecture for goal directed agents that explore an environment without need for constant commands from an operator. Two learning algorithms are presented and our algorithm implementations analyzed. The report highlights the changes made to PAGODA and the development challenges encountered. Additionally, several potential improvements to the PAGODA system are given.
Worcester Polytechnic Institute
Major Qualifying Project
All authors have granted to WPI a nonexclusive royalty-free license to distribute copies of the work, subject to other agreements. Copyright is held by the author or authors, with all rights reserved, unless otherwise noted.