On the implementation of central pattern generators for periodic rhythmic locomotion

This paper presents the feasibility study of the efficient digital hardware implementation of a neural model to generate locomotion patterns of periodic rhythmic movements inspired by biological neural networks found in animal nervous system called Central Pattern Generators (CPGs). The proposed implementation contains a dedicated digital module that mimics the functionality and organization of the fundamental Amari-Hopfield CPG. This module is attached to an embedded processor running the uclinux operating system. The present paper deals only with the implementation of the basic CPG component and how to embed it under a System on a Chip (SoC) approach in order to be controlled by external commands in a high level transparent way for application development. The system is implemented on a Field Programmable Gate Array (FPGA) device providing a compact, flexible and expandable solution for generating periodic rhythmic patterns in robot control applications. According to experimental results, the architecture can be used as a basis for a biomimetic intelligent embedded control platform for articulated autonomous robots.