Design of a domain-independent training system based on a smart pedagogical agent

Training in industry is one of the most critical and expensive tasks to be faced by the management. Furthermore, in some cases, it is dangerous or even impossible to directly train the operators on the real plants where security and safety problems may arise, making it very difficult to start training programs at low cost. For these reasons, the field of training in industry is rapidly developing using software or hardware solutions coming mainly from the following research areas: i) Human-Computer interaction, i.e., the use of complex and interactive human-machine interfaces, ii) plant simulators, i.e., software systems which are delivered with the plant itself to test and to learn complex tasks and processes, iii) Intelligent Training Systems, i.e., the availability of intelligent and personalized training systems where a virtual tutor guides users through a personalized learning path. In this paper we present the overall architecture of a system for industrial training, embedded into an Intelligent Tutoring System that can provide more effective and personalized training and learning in a context where working directly on the real plants can be difficult and very expensive. In particular we present a simulator for training operators in using power plants, based on a multimedia and interactive interface. This system is particularly suitable to be used for training in industrial electric and oil plants. Moreover the system allows operators for collaborative problem solving. Currently the system is under delivery to an Italian Electric industry.