Optimizing Ground Reaction Force Estimation in Gait Analysis Using an IMU Sensor and Kalman Filtering

Walking and running may seem simple, but mechanically they place significant stress on the body. Each heel strike generates high-frequency impulse loads, producing sudden and substantial contact forces that travel through the tibiofemoral joint. Understanding these ground reaction forces (GRFs) is vital for studying the development and progression of conditions such as osteoarthritis. Since direct in-vivo measurement of joint contact forces is not possible, computational modeling becomes indispensable. Developing a realistic multiphysics model of the tibiofemoral joint-encompassing bones, cartilage, and synovial fluid-requires accurate GRF data as external boundary conditions to capture joint loading behavior. However, existing methods for obtaining GRFs are largely confined to laboratory settings or demand expensive equipment, underscoring the need for affordable, wearable alternatives. In this work, we propose an IMU-based technique for estimating vertical GRF during gait. A single body-mounted accelerometer is used to capture vertical acceleration data, from which GRF is recovered using a discrete-time Kalman filter with Zero-Velocity Updates (ZUPT) for drift correction. Our simulation demonstrates that, even under strong sensor bias, the Kalman filter reliably reconstructs GRF by dynamically adjusting the estimated accelerometer bias. This allows for accurate force estimation across gait cycles, enabling the integration of wearable sensing into musculoskeletal modeling pipelines.