Model predictive control of a humanoid robot

Andrei Herdt 1
1 BIPOP - Modelling, Simulation, Control and Optimization of Non-Smooth Dynamical Systems
Inria Grenoble - Rhône-Alpes, LJK - Laboratoire Jean Kuntzmann, INPG - Institut National Polytechnique de Grenoble
Abstract : The range of motions that humanoid robots are able to realize is strongly limited by inherent dynamical constraints so that any control law that does not consider these limitations, in one way or another, will fail to avoid falling. The Model Predictive Control (MPC) technique is capable of handling constraints on the state and the control explicitly, which makes it highly apt for the control of walking robots.We begin by unveiling the specific structure of these constraints, stressing especially the impor- tance of the supports on the ground. We give thereupon a sufficient condition for keeping balance and formulate an MPC law that complies with it. This formulation serves us then for the design of practicable controllers capable of more efficient and more robust control of humanoid robots.
Document type :
Theses
Complete list of metadatas

Cited literature [45 references]  Display  Hide  Download

https://pastel.archives-ouvertes.fr/pastel-01038025
Contributor : Abes Star <>
Submitted on : Wednesday, July 23, 2014 - 10:31:57 AM
Last modification on : Tuesday, February 5, 2019 - 1:38:02 PM
Long-term archiving on : Tuesday, April 11, 2017 - 4:55:51 PM

File

2012ENMP0113.pdf
Version validated by the jury (STAR)

Identifiers

  • HAL Id : pastel-01038025, version 1

Collections

Citation

Andrei Herdt. Model predictive control of a humanoid robot. Other. Ecole Nationale Supérieure des Mines de Paris, 2012. English. ⟨NNT : 2012ENMP0113⟩. ⟨pastel-01038025⟩

Share

Metrics

Record views

934

Files downloads

1262