Estimating 3D Ground Reaction Forces During Daily Activities: A Reduced Sensor Setup and Virtual Pivot Point Approach.

Abstract

Ground reaction forces (GRF) during daily activities are critical for assessing joint loading, particularly in individuals with osteoarthritis (OA). Traditional GRF measurements rely on force plates, which restrict their use to laboratory environments. This study presents a novel method for estimating 3D GRF using a minimal sensor setup comprising three inertial measurement units (IMUs) and pressure insoles (PI), and exploiting the biomechanical concept of Virtual Pivot Point (VPP) to distribute the total GRF between the feet. Data were collected during various activities of daily living (ADL), including walking tasks, stair ascent/descent, and sit-to-stand movements. The proposed system demonstrates high accuracy, achieving relative root mean squared errors (rRMSE) below 15% and correlation coefficients exceeding 0.7 for all tasks, except sit-to-stand movements during Timed Up and Go test (TUG). This approach significantly reduces the sensor burden while maintaining performance comparable to more extensive setups. By combining the estimated 3D GRF with kinematics, joint loading can be estimated, enabling a portable setup for monitoring healthy subjects during ADL in real-world settings. The open-source MATLAB code and dataset are available in the 4TU Research Data repository [1].