Exploring Inertial Sensor Fusion Methods for Direct Ergonomic Assessments

The industrial site, particularly assembly lines, encompass repetitive labour processes which are considered an ergonomic risk factor for the onset of musculoskeletal disorders. Direct assessments methods promote faster ergonomic feedback, supporting the development of sustainable working conditions. This work presents an upper-body motion tracker framework using inertial sensors to provide direct measurements for ergonomics research. An experimental assessment performed by 14 subjects was completed in order to evaluate the joint angle reconstruction of the proposed method while using the measures of an optical motion capture system as reference. This study investigated the results of three distinct complementary sensor fusion techniques, namely the quaternion-based complementary filter, the Mahony filter and the Madgwick filter. Furthermore, foreseeing the possibility of magnetic disturbance in industrial environments, a comparison was conducted between methods that use magnetic data, i.e. 9-axis, and other inertial-based approaches that do not require magnetic information, i.e. 6-axis. A quantitative analysis was performed using two metrics, the cumulative distribution function and the root-mean-square error, hence, providing an evaluation for the different sensor fusion approaches. The overall results suggest that the 9-axis Madgwick filter although noisier presents a more accurate angular reconstruction.