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

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 net-based specification approach that gives the work formalism for effective modeling and implementation is elaborated. Finally, a computerized numeric control (CNC) milling operation virtual training prototype system is introduced to show the effectiveness of this approach.