Abstract

While in this paper, we present a model of fuzzy cerebellar cortex that puts together two sorts of learning: feedforward and predictive association founded on learning between granule cell ascending branch and parallel fiber inputs, and reinforcement learning with feedback error correction based on climbing fiber activity. To show the model's utility, we simulated the control of a robotic arm. Specification of the model is successfully used to learn how to control the timing release of the robotic arm in the course of the task. Biological control systems have always been studied as workable inspiration for construction of robotic controllers. The cerebellum can be involved in the production and learning of smooth, coordinated movements. The cerebellum is assumed to be the location of movement coordination within the body. It really is hypothesized that through the use of some form of learned internal model, the cerebellum is able to overcome inherent sensory latency and coordinate fast, accurate movement without needing complex mathematical algorithm.

Key words: Cerebellum, predictive, feedforward, cerebellar learning, motor control, CMAC neural     network, kinematics.