Institut des Systèmes Intelligents
et de Robotique

Partenariats

UPMC

CNRS

INSERM

Tremplin CARNOT Interfaces

Labex SMART

Rechercher

Titre de la thèse : Control of an anthropomorphic arm-hand robot for grasping and dexterous manipulation

 

Mots clés : max-torque, force control, natural gesture, tenodesis, gravitation-based manipulation 

 

Cette soutenance aura lieu le

 

Jeudi 26 septembre 2013 à 10h00

UPMC – Campus Jussieu, 4 Place Jussieu - 75005 Paris

Tour 55, 2ème étage, couloir 55-65, salle 211.

 

Jury :

 M. Philippe          POIGNET             

M. Jean-Pierre      GAZEAU            

M. Guillaume        MOREL              

M. Franck            QUAINE              

M. Kaspar            ALTHOEFER      

Mme Véronique    PERDEREAU   

 

Résumé :

This thesis deals with the control of an anthropomorphic arm-hand robot by focusing on two aspects: the control of the fingertip force and the coordination between the arm and the hand.

The force control of a robotic finger remains difficult despite the advances in current state-of-art. This is due to the small size of the finger, its low communication bandwidth, the lack of precision of the position sensors and the significant backlash in the actuation systems. A new approach controlling the fingertip force by adjusting the joint torque saturation parameter shows better results. Not limited to pure force control, this control method is proved to also have good performance when applying to indirect and hybrid position/force control. 

Usually ignored in literature while considering dexterous manipulation, the position and movement of the arm play a very important role. Many in-hand manipulation tasks cannot be realized without a proper movement of the arm. One typical example is the rotation of the manipulated object relative to the palm without moving the fingers thanks to inertial and gravitational effects. Besides, arm movement is also an important factor contributing to the natural appearance of the grasping gestures. In this thesis, the movement of the grasped object under gravitational effect was analyzed and a grasping strategy was elaborated. In addition to this, some mechanical constraints (tenodesis effect in particular) contributing to the human natural gestures were deciphered and such natural gestures were reproduced on an anthropomorphic arm-hand robot in redundant grasping situations.