Science China Information Sciences

, Volume 54, Issue 3, pp 653–663

Neighboring optimal control for periodic tasks for systems with discontinuous dynamics

  • ChengGang Liu
  • Christopher G. Atkeson
  • JianBo Su
Research Papers

DOI: 10.1007/s11432-011-4185-z

Cite this article as:
Liu, C., Atkeson, C.G. & Su, J. Sci. China Inf. Sci. (2011) 54: 653. doi:10.1007/s11432-011-4185-z

Abstract

We propose a trajectory-based optimal control method for periodic tasks for systems with discontinuous dynamics. A general method, dynamic programming, suffers from the problem of dimensionality. We use local models of the optimal control law to construct a local controller. We combine a parametric trajectory optimization method and differential dynamic programming (DDP) to find the optimal periodic trajectory in a periodic task. By formulating the optimal control problem with an infinite time horizon, DDP generates time-invariant local models of the optimal control law. For DDP, the value function at dynamics discontinuities is approximated by a second order Taylor series. The utility of the proposed method is evaluated using simulated walking control of a five-link biped robot. The results show lower torques and more robustness from the proposed controller than a PD servo controller.

Keywords

trajectory optimization differential dynamic programming optimal control biped walking control 

Copyright information

© Science China Press and Springer-Verlag Berlin Heidelberg 2011

Authors and Affiliations

  • ChengGang Liu
    • 1
  • Christopher G. Atkeson
    • 2
  • JianBo Su
    • 1
  1. 1.Department of AutomationShanghai Jiao Tong UniversityShanghaiChina
  2. 2.Robotics InstituteCarnegie Mellon UniversityPittsburghUSA

Personalised recommendations