Work Area: Neural Networks and Neuroscience
Keywords neural networks, machine vision, robotics, neuroscience, sensors
Start Date: 1 October 92 / Duration: 36 months / Status: starting
[ participants / contact ]
Abstract The project focuses on the analysis of natural and artificial sensors and is providing a better understanding of the information processing of sensory data occurring at the initial stages in vision, olfaction and hearing. Several biological mechanisms have been identified which are used to design electronic ear and develop efficient control strategies for robots.
The major aims of the project are twofold: to obtain a quantitative description of information processing occurring in early stages of vision, olfaction and hearing; and to propose new hardware architecture which will be useful for robotics. The specific aims of the project are to:
The project is interdisciplinary: it gathers expertise and knowledge from different fields, such as psychology, physics and information science, and intends to use both theoretical and experimental approaches. Modelling will explore analytical solutions (obtained by solving ordinary and/or partial differential equations) and computer simulations. Models will be compared with experimental results which already have been published, or with results specifically obtained in this project. The experimental analysis will use a variety of electrophysiological and optical techniques. The project also aims at making a comparison between artificial and natural sensors in order to see which features of biological sensory processing can be useful in designing new sensors or robots. In particular, the consortium will investigate an electronic ear. By comparing techniques inspired by the nervous system of the fly and traditional control techniques inspired by artificial intelligence, the team will also analyse the best strategies for controlling the navigation of a mobile vehicle.
During the first year of the project some relevant results have been obtained in the understanding of natural and artificial sensors:
The project is producing results which will be useful for industrial exploitations. The blueprint and the VLSI chip of the electronic ear is clearly relevant to automatic speech processing and many tasks in office automation. The development of visual routines for outdoor navigation and best strategies for controlling robots has generated interest in several European industries and agencies.
The results obtained in this project have been disseminated within the scientific community, by oral presentations in many international conferences and symposia. Several contacts with European companies have been established in order to strengthen interactions and collaborations.
Univeristà degli studi di Genova - I
Via Dodecaneso, 33
I - 16146 Genova
CNRS Marseille - F
CNR-ICB - I
University of Bristol - UK
University of Cambridge - UK
University of Loughborough - UK
Prof. V. Torre
tel +39/10 3536311
fax +39/10 314218
SSS - 6961, August 1994
please address enquiries to the ESPRIT Information Desk
html version of synopsis by Nick Cook