Inefficiency in real-time visualization and user interaction in traditional accident causation models (ACMs) necessitates the development of a dynamic ACM that can foster real-time hazard identification, accident prevention and interactive safety training. A virtual reality-based accident causation model (VR-ACM) may serve such a purpose. In this study, we performed a comprehensive literature review on different ACMs and safety training practices. The limitations of the existing models and practices are identified. A VR-ACM model is proposed comprising three modules: VR-based modelling and simulation, accident causation and safety training. Several research issues for VR-ACM are highlighted. An experimental study with 22 crane operators is presented, showing the applicability of the proposed model. The proposed VR-ACM serves as a medium for analysis of potential underlying causes of accidents, the three-dimensional perspective of visual analysis, real-time user interactions and real-time judgement and decision-making.
Keywords: accident causation; hazard identification; safety training; virtual prototype; virtual reality.