1999 Fiscal Year Final Research Report Summary
Bilateral control and visual perception in man and robot
| Project/Area Number |
09044189
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (B).
|
| Allocation Type | Single-year Grants |
|---|
| Section | 一般 |
|---|
| Research Field |
Intelligent informatics
|
|---|
| Research Institution | Kansai University |
|---|
Principal Investigator |
COOK N.D. Kansai University, Faculty of Informatics, Professor, 総合情報学部, 教授 (00268319)
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
HAYASHI Takefumi Kansai University, Faculty of Informatics, Associate Professor, 総合情報学部, 助教授 (90268326)
INUI Toshio Kyoto University, Graduate School of Informatics, Professor, 大学院・情報学研究科, 教授 (30107015)
|
|---|
| Project Period (FY) |
1997 – 1999
|
| Keywords | Intelligent Robot / Bilateral Control / Human Vision / Neural Network / Cognitive Psychology / Neurological Psychology / Visual Cognition |
|---|
| Research Abstract |
The purpose of this research is to clarify the functions of the left and right cerebral hemispheres in object recognition and in motor control, and try to apply those functions, in a bilateral robot. We have developed a bilateral simulation models and a bilateral robots.
In a simulation of certain basic properties of human consciousness, we have devise a mechanism for the bilaterally-symmetrical activation of bilaterally-asymmetrical neural networks. The simulation is designed such that two semi-autonomous "cerebral hemispheres" process sensory information and collaborate across a "corpus callosum" to produce motor output in an object recognition task. The activation of appropriate verbal labels is achieved in two steps. The first is bilateral, Mexican hat-like activation of labeled maps ; the topology of excitation and inhibition being a function of the relative location of learned labels on the maps. The second step is an inhibitory ipsilateral activation. These step produce asymmetrical activation of the maps, corresponding to a nuclear focus of conscious attention sad a "fringe" of peripheral awareness. The decision for object-naming depends on the total activation of the bilateral net.
In light of these results, we plan to use bilateral neural network in navigation tasks and other robot tasks.
|
