Optimizing Prosthetic Wrist Movement: A Model Predictive Control Approach
Francesco Schetter, Shifa Sulaiman, Shoby George, Paolino De Risi, Fanny Ficuciello
TL;DR
This work addresses naturalistic control of a tendon-driven soft wrist in a prosthetic hand by developing an inverse dynamic Model Predictive Control (MPC) framework grounded in Euler-Bernoulli/Lagrangian modeling. The controller operates with prediction horizon $p=10$ and control horizon $\nu=5$, using a KWIK active-set solver to minimize a multi-term cost $J(z_k)$ while enforcing physical constraints. Validation through Simulink simulations and real-time experiments shows excellent tracking accuracy in simulation ($\text{RMSE} \approx 2.1\times 10^{-3}$ rad, $t_s\approx1.2$ s) versus modestly degraded experimental performance ($\text{RMSE} \approx 6\times 10^{-2}$ rad, $t_s\approx1.35$ s$)$ due to material stiffness and hardware nonidealities. The approach demonstrates enhanced dexterity and responsiveness for prosthetic control, paving the way for predictive, user-intent-aware wrist articulation in soft robots.
Abstract
The integration of advanced control strategies into prosthetic hands is essential to improve their adaptability and performance. In this study, we present an implementation of a Model Predictive Control (MPC) strategy to regulate the motions of a soft continuum wrist section attached to a tendon-driven prosthetic hand with less computational effort. MPC plays a crucial role in enhancing the functionality and responsiveness of prosthetic hands. By leveraging predictive modeling, this approach enables precise movement adjustments while accounting for dynamic user interactions. This advanced control strategy allows for the anticipation of future movements and adjustments based on the current state of the prosthetic device and the intentions of the user. Kinematic and dynamic modelings are performed using Euler-Bernoulli beam and Lagrange methods respectively. Through simulation and experimental validations, we demonstrate the effectiveness of MPC in optimizing wrist articulation and user control. Our findings suggest that this technique significantly improves the prosthetic hand dexterity, making movements more natural and intuitive. This research contributes to the field of robotics and biomedical engineering by offering a promising direction for intelligent prosthetic systems.