| description abstract | Worker safety and productivity and the factors that affect them, such as ergonomics, are essential aspects of construction projects. The application of ergonomics and the identification of the connections between workers and assigned tasks have led to a decrease in worker injuries and discomfort, beneficial effects on productivity, and a reduction in project costs. Nevertheless, workers in the construction area are often subjected to awkward body postures and repetitive motions that cause musculoskeletal disorders, in turn leading to delays in production. As a systematic and widely used procedure that generates a final document or form, physical demand analysis (PDA) assesses the health and safety of workers engaged in construction or manufacturing activities and proactively evaluates ergonomic risks. However, to gather the necessary information, traditional PDA methods require ergonomists to spend significant time observing and interviewing workers. To increase the speed and accuracy of PDA, this study focuses on developing a systematic PDA framework to automatically fill a posture-based PDA form and address the physiological aspects of task demands. In contrast to the traditional observation-based approach, the proposed framework uses a motion capture (MOCAP) system and a rule-based expert system to obtain joint angles and body segment positions in different work situations, convert the measurements to objective identification of activities and their frequencies, and then automatically populate the PDA forms. The framework is tested and validated in both laboratory and on-site environments by comparing the generated forms with PDA forms filled out by ergonomists. The results indicate that the MOCAP-/AI-based automated PDA framework successfully improves the performance of PDA in terms of accuracy, consistency, and time consumption. Ultimately, this framework can aid in the design of job tasks and work environments with the goal of promoting health, safety, and productivity in the workplace. | |
