Virtual Reality (VR) training systems offer significant advantages over more traditional training methods and tools. First, in comparison to paper, audio, and video materials, virtual training is more attractive and engaging to users. Second, in contrast to playing physical scenarios, virtual training requires neither a physical infrastructure nor appliances. Hence, it allows users to avoid a risk posed by dangerous equipment and machines. It also liberates companies from acquiring expensive or hardly available devices, especially in industry 4.0 production environments, or suspending the work of necessary devices. Furthermore, virtual reality training can be accomplished to some extent without the assistance of instructors. Therefore, such training is typically easier to organize, cheaper, more efficient, and flexible than traditional training.
However, building useful VR training environments requires competencies in programming and 3D modeling, which is necessary to develop behavior-rich training scenes and objects, as well as domain knowledge, which is crucial to prepare practical and meaningful training scenarios in a given domain. Thus, this process typically involves IT specialists and domain experts, whose IT expertise is usually low. The need for collaboration between both groups makes the development of VR training environments complex, time-consuming, and expensive. Therefore, the availability of user-friendly tools for domain experts to design virtual training with domain knowledge becomes essential to reduce the required time and effort and consequently promote VR in training.
This project aims to develop a flexible VR training system for electrical operators working with high-voltage installations. We propose a new approach to creating and managing virtual reality training scenarios and scenes based on the semantic web. The approach consists of user-friendly semantic editors as well as semantic representations of VR training content built upon ontologies. The representation covers users, tasks, equipment, problems, and errors that may occur in training scenarios. The editors allow domain experts to design and manage scenarios and scenes using the representation. The design process is conducted in an intuitive visual way. The use of ontologies enables description of training scenarios and scenes in a domain-specific way. Moreover, the selection and combination of appropriate objects into training scenarios and scenes as well as the verification of modeling results are completed by well-established reasoning activities on the semantic web, such as instance checking, query answering, and consistency checking against the used ontologies.