Decision Theoretic Agent Design for Personal Rapid Transit Systems

—This paper details a learning decision-theoretic intelligent agent designed to solve the problem of guiding vehicles in the context of Personal Rapid Transit (PRT). The intelligent agents are designed using Bayesian Decision Networks. The agents are designed to utilize the known methods of machine learning with Bayesian Networks (BN): parameter learning and structure learning. In addition, a new method of machine learning with BNs, termed utility learning in this paper, is introduced. BN software for Matlab is used to realize the proposed agent. Additional software is written to simulate the PRT problem using various intelligent agents that utilize one or more learning methods.

• Type: Journal
• Year: 2007
• Where: ENGL
• Authors: Iheanyi C. Umez-Eronini, Ferat Sahin
• Keywords: Decision-theoretic Intelligent Agent, ENGL 2007, Intelligent Agents, Personal Rapid Transit