XCoM-Based Moving-Mass Control in a Backpack for Modulating Gait Stability During Load Carriage

Backpack load carriage can alter gait mechanics and dynamic stability, while existing assistive backpacks have primarily targeted load transmission and metabolic cost. This study investigated whether active vertical repositioning of a moving mass within a backpack could influence gait-stability-related outcomes. A frontal-plane double inverted pendulum model was combined with inertial measurement unit (IMU)-derived motion states to calculate a model-estimated mediolateral extrapolated center of mass (XCoM), and model predictive control (MPC) regulated a 5 kg moving mass in real time. Six participants walked on a treadmill at 1.2 m/s under a Normal condition, three fixed-mass conditions (Low, Mid, and High), and a Powered condition with XCoM-based moving-mass control. Short-term largest Lyapunov exponent (LLE), kinematic variability (KV), and trunk sway were evaluated as gait-stability-related outcomes. In the Powered condition, short-term LLE was 19.0%, 15.7%, and 24.4% lower than in the Low, Mid, and High conditions, respectively, with all six participants showing lower LLE in each comparison. However, no pairwise comparison of LLE, KV, or trunk sway remained statistically significant after Bonferroni correction. The model-estimated XCoM and filtered moving-mass velocity command were negatively correlated (r=−0.3310). These findings support the feasibility of state-dependent moving-mass control during load carriage and indicate the need for further evaluation of its effects on gait stability in a larger cohort.