Skip to Main content Skip to Navigation

Approche neuro-robotique pour le contrôle des systèmes anthropomorphiques

Minh Tuan Tran 1
1 LAAS-GEPETTO - Équipe Mouvement des Systèmes Anthropomorphes
LAAS - Laboratoire d'analyse et d'architecture des systèmes
Abstract : This thesis presents a neuro-robotics approach to the control of reaching motion in anthropomorphic systems such as humanoid robots. The objective of this study is twofold. First, it presents the state of the art of existing control models in movement neuroscience and describes a set of motor control principles which can be used for the control of humanoid robots. Second, it proposes the usage of formalisms in robotics for the modeling of the necessary sensorimotor transformation process for the execution of voluntary movement. Particularly, it shows that visual-based formalism with kinematic and dynamic models of articulated chains, which play an essential role in the modeling of the motion control problem in robotics, can provide some key elements to answer open questions in neuroscience. In the first aspect of the work, we developed a control method, based on a movement optimization model proposed in neuroscience, then applied it to the control of reaching movements of the robot HRP2. The movements produced by this method seem very similar to the movements observed in humans and present major features of human movements : quasistraight hand trajectory with bell-shaped velocity. We also developed another control method inspired from the theory of motor primitives in neuroscience. This method allows to simplify the complexity of the control problem by producing rapidly realistic movements of the robot from a set of reference movements. These results show that the theories on human motor control can be used successfully to elaborate control methods of reaching movements on humanoid robots. In the second aspect of the work, we developed a eye-hand coordination model to test and compare movements generated from the eye-centered and the body-centered reference frame. This model, which is based on biologically-inspired controllers for hand and eye, in closedloop with sensory information, allows to simultaneously control the movement of the hand toward a moving target an d the movement of the eye toward the target. By comparing the trajectories generated by this model in the eye-centered and body-centered frame, we show that the movements made in eye-centered frame are more robust to perception errors. While the issue of identification of the reference frame used by the brain for encoding the movement is the subject of a controversial debate in neuroscience, this result provides computational arguments supporting an eye-centered coding of visually guided reaching movements.
Document type :
Complete list of metadatas

Cited literature [115 references]  Display  Hide  Download
Contributor : Arlette Evrard <>
Submitted on : Wednesday, January 13, 2010 - 3:05:37 PM
Last modification on : Friday, January 10, 2020 - 9:10:09 PM
Long-term archiving on: : Thursday, October 18, 2012 - 12:25:23 PM


  • HAL Id : tel-00446797, version 1


Minh Tuan Tran. Approche neuro-robotique pour le contrôle des systèmes anthropomorphiques. Automatique / Robotique. Université Paul Sabatier - Toulouse III, 2009. Français. ⟨tel-00446797⟩



Record views


Files downloads