Ground reaction forces estimation using IMU-based kinematics and OpenSim Moco

The state-of-the-art method to measure GRF is based on force plates (FP) often integrated with motion capture (MoCap) in motion analysis laboratories. Despite the high accuracy, reliability, and repeatability of laboratory data in combination with musculoskeletal modeling it requires a controlled environment and highly skilled operators. In addition, it was observed that subjects may change their movement strategy when walking on a (limited environment – nonecological) treadmill instead of an open field overground. To overcome the need for a laboratory environment, several methods were developed to measure GRF using mobile devices such as wearable sensors. For instance, methods based on insoles (limitations, wear and tear, cons) that directly measure GRF or methods based on inertial measurement units (IMUs) that measure the motion of body segments and estimate GRF using musculoskeletal models (MSK) and/or machine learning methods1. Therefore, this thesis project aims to 1) develop a workflow to estimate the ground reaction force based on a novel foot contact sphere MSK model combined with kinematics: • OpenSim Moco – Full body model • Contact sphere properties: position, thickness, friction, etc • IK -> torques (as low as possible) • 2) evaluate its accuracy compared to the state-of-the-art method - MoCap and ML - during walking