Developing the Cerebellar Chip as a General Control Module for Autonomous Systems

摘要

Biological systems have evolved robust, adaptive control strategies to deal with a wide range of control tasks in time varying systems and environments. The cerebellum is the brain structure particularly associated with the control of skilled movements, the advantageous properties of the cerebellum can be exploited for robotic control applications. In this contribution we present a bioinspired cerebellar control algorithm. We extend the existing cerebellar inspired adaptive filter control algorithm, previously applied to plants of specific order, to the control of general n~(th) order plants. This is done by augmenting the existing cerebellar algorithm with a reference model, a technique used in model reference adaptive control. This augmented cerebellar controller is applied successfully to the simulated control of a general plant, and to the real time control of a dielectric electroactive polymer actuator. This augmented biomimetic control strategy has promise for the control of human-centred robots operating in unstructured environments.