ME-WARD: A multimodal ergonomic analysis tool for musculoskeletal risk assessment from inertial and video data in working places
This study addresses the limitations of traditional ergonomics assessments, which rely on specialized equipment and lack compatibility with multimodal motion data, hindering deployment in resource-constrained settings. To overcome these challenges, the authors propose a multimodal upper-limb musculoskeletal risk assessment system based on the Rapid Upper Limb Assessment (RULA) method, integrating inertial measurement units (IMUs) and monocular video-based human pose estimation. This approach enables, for the first time, universal support for heterogeneous motion capture data—including low-cost video streams. The system combines deep learning–based pose estimation, IMU-derived joint angle computation, RULA scoring, and multimodal fusion. Evaluated in a conveyor-belt assembly task, it demonstrates high agreement with IMU-based ground truth RULA scores and robust performance using only monocular video input, offering both high reliability and scalability.