Cyber-physical postural training system for construction workers

Risks of work-related musculoskeletal injuries can be reduced by adequately training construction workers on performing work in safe postures. Traditional training approaches provide limited support for performing work tasks and receiving real-time feedback. This paper describes a cyber-physical postural training environment where workers can practice to perform work with reduced ergonomic risks. The proposed system uses wearable sensors, Vive trackers, machine learning and virtual reality to track body kinematics, and engagement with physical construction resources, and provides feedback via an interactive user interface. The postural training system was developed for training workers engaged in wood frame construction. User study of the effectiveness of the feedback and user interface was conducted. Findings showed that the user interface was perceived as convenient with limited interference with the workspace. The feedback was understandable in learning risks associated with participant's postures. Further research will involve conducting formative workload evaluation of the user interface and developing a reinforcement learning model for adapting the feedback based on the state of learning.