Simplification of the Hill Muscle Model Computation for Real-Time Walking Controllers with Large Time Steps

Van der Noot, Nicolas;Dzeladini, Florin;Ijspeert, Auke J.;Ronsse, Renaud
(2014) Dynamic Walking — Location: Zurich (10.June.2014)

Files

poster_dyn_walk_2014.pdf
  • Open Access
  • Adobe PDF
  • 1.66 MB
DW14Abstract_Hill_simpl.pdf
  • Open Access
  • Adobe PDF
  • 432 KB

Details

Authors
  • Van der Noot, NicolasUCLouvain
    Author
  • Dzeladini, FlorinEPFL
    Author
  • Ijspeert, Auke J.EPFL
    Author
  • Author
Abstract
Bio-inspired controllers are emerging as a promising way to implement dynamic walking. In this study, we implemented the one proposed by Geyer & Herr (2010), relying on reflex-controlled virtual Hill muscles. In this model, muscles’ state is determined by the length (lce) of their active, contractile element. However, its update rate is governed by a stiff and strongly non-linear state equation, thus requiring a small integration time step.
Affiliations

Citations

Van der Noot, N., Dzeladini, F., Ijspeert, A. J., & Ronsse, R. (2014). Simplification of the Hill Muscle Model Computation for Real-Time Walking Controllers with Large Time Steps. Dynamic Walking, Zurich. https://hdl.handle.net/2078.5/195392