Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

A Learning-based Model Reference Adaptive Controller Implemented on a Prosthetic Hand Wrist

Shifa Sulaiman, Mohammad Gohari, Francesco Schetter, Fanny Ficuciello

TL;DR

This work addresses the challenge of achieving accurate, real-time control of a tendon-driven soft wrist in a prosthetic hand. It proposes a learning-based Model Reference Adaptive Controller (NN-MRAC) built on a Timoshenko beam model to estimate tendon forces from deflection errors and guide online adaptation toward a reference model with transfer function $T(s)=\frac{-4}{s^2+3s+5}$. The approach combines a neural network for online force estimation with MRAC by generating training data from the MRAC scheme, and it is validated through both simulation (RMSE $=6.14\times10^{-4}$ m, settling time $3.2$ s) and real-time experiments (RMSE $=5.66\times10^{-3}$ m, steady-state error $8.05\times10^{-3}$ m, settling time $1.58$ s). Results demonstrate improved motion accuracy and responsiveness with reduced computational burden, highlighting the potential of adaptive intelligent control for wearable soft robotics. The study also notes practical limitations due to hardware stiffness variations and suggests directions for more robust, sensor-fed control in future work.

Abstract

The functionality and natural motion of prosthetic hands remain limited by the challenges in controlling compliant wrist mechanisms. Current control strategies often lack adaptability and incur high computational costs, which impedes real-time deployment in assistive robotics. To address this gap, this study presents a computationally efficient Neural Network (NN)-based Model Reference Adaptive Controller (MRAC) for a tendon-driven soft continuum wrist integrated with a prosthetic hand. The dynamic modeling of the wrist is formulated using Timoshenko beam theory, capturing both shear and bending deformations. The proposed NN-MRAC estimates the required tendon forces from deflection errors and minimizes deviation from a reference model through online adaptation. Simulation results demonstrate improved precision with a root mean square error (RMSE) of $6.14 \times 10^{-4}$ m and a settling time of $3.2$s. Experimental validations confirm real-time applicability, with an average RMSE of $5.66 \times 10^{-3}$ m, steady-state error of $8.05 \times 10^{-3}$ m, and settling time of $1.58$ s. These results highlight the potential of the controller to enhance motion accuracy and responsiveness in soft prosthetic systems, thereby advancing the integration of adaptive intelligent control in wearable assistive devices.

A Learning-based Model Reference Adaptive Controller Implemented on a Prosthetic Hand Wrist

TL;DR

This work addresses the challenge of achieving accurate, real-time control of a tendon-driven soft wrist in a prosthetic hand. It proposes a learning-based Model Reference Adaptive Controller (NN-MRAC) built on a Timoshenko beam model to estimate tendon forces from deflection errors and guide online adaptation toward a reference model with transfer function . The approach combines a neural network for online force estimation with MRAC by generating training data from the MRAC scheme, and it is validated through both simulation (RMSE m, settling time s) and real-time experiments (RMSE m, steady-state error m, settling time s). Results demonstrate improved motion accuracy and responsiveness with reduced computational burden, highlighting the potential of adaptive intelligent control for wearable soft robotics. The study also notes practical limitations due to hardware stiffness variations and suggests directions for more robust, sensor-fed control in future work.

Abstract

The functionality and natural motion of prosthetic hands remain limited by the challenges in controlling compliant wrist mechanisms. Current control strategies often lack adaptability and incur high computational costs, which impedes real-time deployment in assistive robotics. To address this gap, this study presents a computationally efficient Neural Network (NN)-based Model Reference Adaptive Controller (MRAC) for a tendon-driven soft continuum wrist integrated with a prosthetic hand. The dynamic modeling of the wrist is formulated using Timoshenko beam theory, capturing both shear and bending deformations. The proposed NN-MRAC estimates the required tendon forces from deflection errors and minimizes deviation from a reference model through online adaptation. Simulation results demonstrate improved precision with a root mean square error (RMSE) of m and a settling time of s. Experimental validations confirm real-time applicability, with an average RMSE of m, steady-state error of m, and settling time of s. These results highlight the potential of the controller to enhance motion accuracy and responsiveness in soft prosthetic systems, thereby advancing the integration of adaptive intelligent control in wearable assistive devices.
Paper Structure (7 sections, 11 equations, 8 figures, 2 tables)

This paper contains 7 sections, 11 equations, 8 figures, 2 tables.

Table of Contents

  1. INTRODUCTION
  2. Mathematical model of a soft wrist section
  3. Model Reference Adaptive Controller developed for the soft wrist section
  4. Results and Discussions
  5. Simulation study
  6. Experimental validation
  7. Conclusion

Figures (8)

  • Figure 1: Soft wrist section (a) Conceptual design of wrist section attached to hand (b) Dimension of disc (c) Bending structure of wrist section
  • Figure 2: (a)Control scheme(b)ANN architecture
  • Figure 3: Training and validation graphs (a)Regression (b)Performance
  • Figure 4: Motion of wrist (a) Radial-1 (b) Radial-2 (c) Ulnar-1 (d) Ulnar-2 (e) Flexion-1 (f) Flexion-2 (g) Extension-1 (h) Extension-2
  • Figure 5: (a)Comparison of response with reference signal (b)Error in deflection during simulation
  • ...and 3 more figures