Multi-template approach to artificial locomotion control

Abstract

In this paper a new approach, the so called Multi-Template Approach (MTA), to implement the Central Pattern Generator (CPG) for locomotion control in Cellular Neural Networks (CNNs) is presented. From a biological point of view an approximated chemical synapse is realized and implemented in a CNN structure. This allows one to extend the results previously obtained with a Reaction-Diffusion (RD) CNN for locomotion control in an insect-like prototype robot to a more general class of artificial CPGs in which the desired locomotion pattern and the switching among patterns is realized by means of a spatiotemporal algorithm implemented in the same CNN structure.