An agent-based approach to developing intelligent virtual reality-based training systems

Abstract

This paper presents an agent-based approach to developing intelligent Virtual Reality-based training systems. The architecture of an intelligent Virtual Reality-based training system is modeled as five collaborative agents: training task-planning agent, simulation agent, performance evaluation agent, on-line instruction agent, and intelligent interface agent. The behavioral and knowledge models of each agent type and the mechanism of inter-agent interaction and coordination are described. A Petri Nets-based specification approach that gives the work formalism for effective modeling and implementation is elaborated. Finally, Computerized Numeric Control (CNC) milling operations virtual training prototype system is introduced to show the effectiveness of this approach.