Multisensory Control of Movement

Description

Project Title:
Multisensory Control of Movement
Acronym:
MUCOM II
Number:
6615
Work Area:
Robotics (sensing and control)
Coordinator:
CNRS
Laboratoire de Physiologie neurosensorielle
Rue de l'Ecole de Médecine 15
F - 75006 PARIS
Coordinator Country:
F
Partners
Université Catholique de Louvain B
Universitätsspital Zürich CH
Université de Genève CH
Ruhr-Universität Bochum D
Université de Montpellier F
Université Pierre et Marie Curie F
Consiglio Nazionale delle Ricerche I
Katholieke Universiteit van Nijmegen NL
Associate Partner
Università di Roma I
Contact Point:
Dr. A. Berthoz
Telephone:
33/1-43296154
Fax:
33/1-43541653
E-Mail:
Keywords:
sensor fusion, sensorimotor coordination, movement perception, posture control, manipulators, 3-d movement control
Start Date:
1 July 92
Duration:
36 months
Status:
running
Abstract:
Goals to be addressed concern the problems of how the brain builds an internal representation of space and movement in order to allow action, orientation, and navigation, and the relation of these mechanisms with robotics of manipulators and mobile robots. The work builds on the results of MUCOM (3149).

AIMS

MUCOM aims to answer the following questions: (i) What are the reference frames in the control of manipulators and hand arm coordination? (ii) What are the mechanisms of the reduction of degrees of freedom and motor coordination in arm movement control? (iii) What are the spatio-temporal transformations involved in gaze control during orienting movements? (iv) What is the contribution of optic flow in self-motion and object-motion perception? (v) What is the contribution of inertial signals to self-motion perception, and what are the mechanisms for fusion of sensors and object motion detection during navigation?

APPROACH AND METHODS

- Acting. The reference frames involved in eye-head-hand coordination will be studied by recording of movements during pointing 3-D tasks. These behavioural data will be compared with neuronal mechanisms of 3-D arm reaching in the cortex and the modelling with neural networks. The relations between visual perception of target movement and movement control will be studied by psychophysical methods and modelling of possible algorithms for the kinematics control of arm trajectory.
- Orienting. This activity will combine neurophysiological studies and neural networks modelling for the study of multisensory integration in orienting movements. Eye, head, and arm movements will be measured in relation to the underlying activities in the superior colliculus and in other structures involved in orienting.
- Navigating. The visuomotor processes involved in 3-D motion perception will be studied by psychophysical methods, mathematical modelling and neurophysiological recordings. The role of inertial signals (provided by the vestibular system) in navigation will be studied by recording neurons in brain structures involved in spatial memory and by psychophysical experiments in humans. New methods for rapid detection of near objects in mobile robots will be investigated.

PROGRESS AND RESULTS

Acting: Mechanisms of arm movement control have been studied. It was shown that the brain uses a flexible egocentric frame of reference. A formal model of 3-D computation of arm reaching movements was developed using populations of neurons coding movement in body-centred reference frames combined with others using gaze-centred reference frames. Same principles used by the brain to reduce the number of degrees of freedom to be controlled have been discovered which may be useful for robot arm control. Finally, results have been obtained concerning multisensory coordination of arm movement in 3-D space: sensory-motor transformations necessary to establish a congruence between target visual information and kinesthetic information on hand position use an intermediate hybrid frame.
Orienting: The mechanisms of oriented eye-head movements have been studied at neuronal and behavioral level. A mechanism for slow correction of saccades has been described. Its neuronal mechanisms have been discovered by revealing the involvement of tecto-reticulo-spinal neurons. The relation between visual and acoustic specification of target contribution has been evaluated and a gaze position feedback discovered in the acoustic system. Superior colliculus neurons have also been shown to control both eye and hand movements during reaching. Models of the colliculus have been compared. These models stimulate dynamic feedback, dynamic memory, and multisensory information.
Navigating and fusion of sensors: The postural sway induced by moving visual patterns was studied and the importance of coupling mechanisms between vision and the active postural system was revealed. The threshold of motion detection are lowered during active movement of the observer which has implication in robot vision. Various aspects of the contribution of inertial cues to navigation have been studied combining modelling and experiments in humans and animals. Finally it has been shown that extraretinal signals contribute to depth measurement and the 3-D aspects of vestibular control of eye movements have been studied.

POTENTIAL

The work is expected to lead to new basic concepts and results concerning the algorithms and computational mechanisms of the brain; new and improved methods for the study of sensorimotor coordination and fusion of sensors; possible new methods for diagnostic tools usable in biomedical exploration; new concepts for the control of mobile robots and new methods and theories of brain studies derived from robotics: and principles of sensor fusion, which could be useful in telemanipulation or telepresence applications.

LATEST PUBLICATIONS

Sven Müßig, last update 07-nov-1995. Your feedback is welcome.