Estimation of body loads during industrial tasks, such as lifting and weight bearing, is central to
workplace ergonomics and the study of the safety and risk factors in work techniques. Evaluating those
loads requires data collection of body kinematics and the external forces prevailing during the task
under evaluation. Current practice calls for kinematic data to be gathered using optical motion capture
systems (OMC) and external forces, primarily ground reaction forces (GRFs), to be gathered using
force plates. However, this experimental methodology is confined to laboratory settings.
Modern motion capture systems, such as those based on Inertial Measurement Units (IMUs), pave the
way to more versatile motion analysis techniques not confined to labs. Inverse dynamics models have
been developed based on IMU kinematic data. In order to eliminate the need for force plates and to
make the experimental apparatus fully portable, those models estimate GRFs from measured
accelerations.
This study aimed to advance the state-of-the-art on IMU-based inverse dynamics analysis by
incorporating pressure insoles as the source of the vertical components of the GRFs, with a view to
improving the model fidelity while keeping the experimental apparatus portable. Specifically, it enabled
the development of a synchronized and automated inverse dynamics model, comprised of an inertial
motion capture suite and pressure insoles, that can estimate net joint forces and moments during manual
handling activities.
An experiment was designed to examine whether the GRFs measured by the pressure insole can detect
and differentiate among various sizes (and weights) of concrete masonry units (CMUs). The
instrumented pressure insoles were consistently able to identify three different CMU block weights (8
kg, 16kg, and 24 kg) during various gait patterns (along circular, square, and linear paths). On the other
hand, the results were inconclusive in distinguishing between one-handed and two-handed manual
handling of CMUs. An improved inverse dynamic model was introduced to calculate the joint loads
workers experience during material manual handling based only on measurements by IMU motion
capture suits and pressure insoles.
The outcome of this thesis was the development of a weight detection algorithm with a detection
accuracy of 89% across all three sizes of CMUS as well as an integrated inverse dynamic model
incorporating data collected by IMUs motion suits and pressure insoles.
