VR System for Hazard Prediction of Unsafe Behaviors in Outbound Training

The construction industry has the most fatal industrial accidents in Japan, and many accidents are due to unsafe worker behavior. On-site hazard prediction training is effective in preventing unsafe behavior, but the need for completing construction projects within limited working hours and construction periods decreases the time available for such training. Therefore, it is important to conduct hazard prediction training against unsafe behavior before construction starts. Current pre-service training methods are limited to verbal or written reminders of safety and health management knowledge by safety managers, and it is difficult to provide workers with opportunities to predict hazards due to their own actions. In this study, we aim to enable safety managers to provide accurate hazard prediction training according to the hazard prediction ability of workers during outbound training, where time can more easily be secured. To that end, we develop a virtual reality system that enables risk prediction training for unsafe behavior in outbound training. We confirm that the system enables safety managers to provide appropriate guidance while grasping the comprehension level of construction site workers and improves the training’s sense of realism and understanding.