Temporal Difference (TD) Based Critic-Actor Adaptive Control for a Fine Hand Motion Rehabilitation Robot

  • Xianwei Huang
  • Fazel Naghdy
  • Haiping Du
  • Golshah Naghdy
  • Catherine Todd
Chapter

Abstract

Robot assisted post-stroke rehabilitation training is an effective approach in delivering the highly intensive repetitive training, aiming to retrain the neural pathways in the brain thus to restore and improve the affected mobility skills. The adaptive control of robotic devices, especially assist-as-needed control providing exact assistive force intensity along the intended motion trajectory for fine motion, can be a complex but effective method. A temporal difference based critic-actor reinforcement learning control method is explored in this study. The effectiveness of the method is verified through Matlab simulation and implemented on a hand rehabilitation robotic device. Results suggest that the control system can fulfil the control task with high performance and reliability, thus holding the promise of improving the fine hand motion rehabilitation training efficiency.

Keywords

Stroke recovery Rehabilitation robot Adaptive control Reinforcement learning Neural network 

References

  1. 1.
    Sale, P., V. Lombardi, and M. Franceschini. 2012. Hand robotics rehabilitation: Feasibility and preliminary results of a robotic treatment in patients with Hemiparesis. Stroke Research and Treatment 2012: 1–5.CrossRefGoogle Scholar
  2. 2.
    Hwang, C.H., J.W. Seong, and D.S. Son. 2012. Individual finger synchronized robot-assisted hand rehabilitation in subacute to chronic stroke: A prospective randomized clinical trial of efficacy. Clinical Rehabilitation 26: 696–704.CrossRefGoogle Scholar
  3. 3.
    Kaelbling, L.P., M.L. Littman, and A.W. Moore. 1996. Reinforcement learning: A survey. Journal of Artificial Intelligence Research 4: 237–285.Google Scholar
  4. 4.
    Lewis, F.L., D. Vrabie, and K.G. Vamvoudakis. 2012. Reinforcement learning and feedback control: Using natural decision methods to design optimal adaptive controllers. Control Systems, IEEE 32: 76–105.MathSciNetCrossRefGoogle Scholar
  5. 5.
    Papahristou, N., and I. Refanidis. 2011. Training neural networks to play backgammon variants using reinforcement learning. In Applications of evolutionary computation, vol. 6624, 113–122, eds. Di Chio C., S. Cagnoni, C. Cotta, M. Ebner, A. Ekárt, and A. Esparcia-Alcázar, et al. Heidelberg: Springer.Google Scholar

Copyright information

© Springer International Publishing AG, part of Springer Nature 2018

Authors and Affiliations

  • Xianwei Huang
    • 1
  • Fazel Naghdy
    • 1
  • Haiping Du
    • 1
  • Golshah Naghdy
    • 1
  • Catherine Todd
    • 1
  1. 1.University of WollongongWollongongAustralia

Personalised recommendations