Neural networks IC for locomotion rhythm generator emulating living organism

Abstract

Living organisms has the excellent locomotion controller by using biological neural networks. Therefore, many researchers have been emulated the biological neural networks for applying similar control to small size robots. We are studying artificial neural networks for the purpose of controlling a small size robot without using software programs. Previously, we have constructed two types of insect type microrobot which could locomote by using their legs. The microrobot could locomote by using a link mechanism with a minimum number of small size actuators. In this paper, we will discuss a neural networks integrated circuit (IC) which can generate the locomotion rhythms for the microrobot. The neural networks IC consists of the pulse-type hardware neuron model which can generate oscillating electrical activity. In addition, we construct the locomotion rhythm generator using neural networks IC and peripheral circuits to generate the driving waveform for the actuator. Neural networks IC and peripheral circuits are mounted on the small size circuit board. As a result, our constructed locomotion rhythm generator can generate the driving waveform to locomote the microrobot without using software programs.