Mimicking Gait Pattern Generators

Abstract

To replicate the behavior of (large scale) biologically inspired networks through neuromorphic circuits the question arises on how to model the synaptic interconnections between neurons. Biological representations often use unidirectional couplings between neurons which waive problems on electrical models, as electrical networks in general are not free from feedback. We propose a simple solution based on a four port circulator to overcome this issue. The method involves attaching appropriate resistances to this circulator, depending on the desired coupling. The wave digital representation of this approach reveals that our modeling enables unidirectional and free from feedback interconnections between neurons. We use the proposed method to investigate multiple gait pattern generators of a four-legged animal.