Skip to Main content Skip to Navigation


Alexis Mifsud 1
1 LAAS-GEPETTO - Équipe Mouvement des Systèmes Anthropomorphes
LAAS - Laboratoire d'analyse et d'architecture des systèmes
Abstract : This thesis deals with the estimation and the stabilization of the state of the passive compliances present in the ankles of humanoid robot HRP-2. These compliances can be seen as a unique and observable degree of freedom, under some assumptions that are explicit. The estimator uses measurements from the inertial unit located in the torso of the robot and possibly force sensors located in its feet. An extended Kalman filter is used for state estimation. This filter uses a complete model of the dynamics of the robot, for which the internal dynamics of the robot, considered as perfectly known and controlled, has been decoupled from the dynamics of passive compliance of the robot. Local observability of the state has been shown by considering this model and measurements from the inertial unit alone. It has also been shown that the addition of the measurements of the force sensors in the feet of the robot makes it possible to complete the state with error measurements in the dynamic model of the robot. The estimator has been validated experimentally on the humanoid robot HRP-2. On this estimator was built a stabilizer of the state of the HRP-2 compliance. The commanded state is the position and speed of the robot's center of mass (indirect control of momentum), the orientation and angular velocity of its trunk (indirect control of kinetic momentum), as well as the orientation and angular velocity of compliance. The command quantities are the acceleration of the robot's center of mass and the angular acceleration of its trunk. A linear quadratic regulator (LQR) was used to compute state-return gains, based on a model called a "flexible reverse pendulum with a flywheel", which consists of a flexible, flexible, inverse clock. rotating mass around the center of mass of the robot represents the trunk of the robot. Tests were carried out on the dual support HRP-2 robot, using the previously described estimator with or without the force sensors.
Document type :
Complete list of metadata

Cited literature [61 references]  Display  Hide  Download
Contributor : Christine Fourcade <>
Submitted on : Friday, December 1, 2017 - 10:20:30 AM
Last modification on : Thursday, June 10, 2021 - 3:02:19 AM


Files produced by the author(s)


  • HAL Id : tel-01653163, version 1


Alexis Mifsud. ESTIMATION ET STABILISATION DE L'ÉTAT D'UN ROBOT HUMANOÏDE COMPLIANT. Automatique / Robotique. Institut national polytechnique de Toulouse (INPT), 2017. Français. ⟨tel-01653163⟩



Record views


Files downloads