Perceived realism in virtual reality experiments for human behaviour research

Virtual reality can be a useful research tool to investigate how people behave in different fire situations. For the results from such experiments to have good validity, i.e., to represent behaviours that can occur in a real environment, the VR environment must be perceived as a good substitute for the real environment. The project focuses on how to achieve such high-quality VR environments. Two different environments, a lecture hall and a corridor, were built up in two virtual versions each and compared with a method for measuring environmental perception, the Semantic Environmental Description (SED). The two VR environments in the lecture hall were designed in a detailed version and a slightly less detailed version, a more simplified version. In the VR environments in the corridor, the main question was how distances are perceived in VR compared to in a real corridor. A total of 159 subjects participated, and all completed a SED assessment in each environment and both in VR and in the real environment. The difference between the SED description for the lecture hall was small between the three environments (real and two VR environments). Distances in a corridor are judged to be most similar between the real corridor and a VR environment that is designed with correct proportions between length, width and height.