ErgoReport: A Holistic Posture Assessment Framework Based on Inertial Data and Deep Learning

Awkward postures are a significant contributor to work-related musculoskeletal disorders (WRMSDs), which represent great social and economic burdens. Various posture assessment tools assess WRMSD risk but fall short in providing an elucidating risk breakdown to expedite the typical time-consuming ergonomic assessments. Quantifying, automating, but also complementing posture risk assessment become crucial. Thus, we developed a framework for a holistic posture assessment, able to, through inertial data, quantify the ergonomic risk and also qualitatively identify the posture leading to it, using Deep Learning. This innovatively enabled the generation of a report in a graphical user interface (GUI), where the ergonomic score is intuitively associated with the postures adopted, empowering workers to learn which are the riskiest postures, and helping ergonomists and managers to redesign critical work tasks. The continuous posture assessment also considered the previous postures' impact on joint stress through a kinematic wear model. As use case, thirteen subjects replicated harvesting and bricklaying, work tasks of the two activity sectors most affected by WRMSDs, agriculture and construction, and a posture assessment was conducted. Three ergonomists evaluated this report, considering it very useful in improving ergonomic assessments' effectiveness, expeditiousness, and ease of use, with the information easily understandable and reachable.