Control of a robotic manipulating arm by a neural network simulation of the human cerebral and cerebellar cortical processes

Abstract

The authors propose a system commanding a robotic manipulating arm under visual control, based on brain modeling. In this model, the movement command is learned by a network which links two subsystems together: a cerebral subsystem which can learn a goal, and a second subsystem responsible for quantitative adjustments and coordination. Those two subsystems are complementary because each subsystem is necessary for fast learning and participates in the final overall task performance. The neural network model described is based on known architectural and functional properties of the cerebral and cerebellar cortices. In each cortical structure, it is possible to define a basic crystalline unit, consisting of several neuronal types, which recurs throughout the structure. Computer simulation suggests how cellular mechanisms in these two structures may be responsible for two different types of adaptive process and how their mutual interactions can produce automatic and refined motor sequences.<>