Postural Stability Of Biomechanical Model Using Passivity Based Nonlinear Model Predictive Control

Abstract

Maintaining postural stability is crucial in human biomechanics, particularly for activities like standing and walking. This thesis explores a novel control strategy combining Nonlinear Model Predictive Control (NMPC) with Passivity-Based Control (PBC) to enhance the stability of a single-link biomechanical model representing the human ankle joint. The model, resembling an inverted pendulum, captures the dynamics of balance, including gravitational forces and inertia. Traditional control methods, such as ProportionalIntegral-Derivative (PID) and H∞, often struggle with the nonlinearities inherent in human biomechanics, which can significantly impact stability and pose challenges in managing postural sway. In this work, PBC is employed to impose passive behavior on the system, enhancing its natural stability. The passivity property ensures energy dissipation and stability, even in the presence of nonlinearities. NMPC is integrated with PBC to optimize control performance, solving an online optimization problem at each sampling instant, thereby ensuring effective and physiologically safe control actions. Simulation results demonstrate that the combined NMPC-PBC strategy effectively stabilizes the ankle joint model, quickly returning the system to equilibrium after disturbances. This study offers a robust control approach with significant implications for rehabilitation, fall prevention, and assistive technologies, advancing our understanding of human postural stability