Development of hardware neural networks generating driving waveform for electrostatic actuator

Abstract

The authors are studying to control the locomotion of the microrobot system using hardware neural networks (HNN). In previous research, a waveform generator was used to drive the electrostatic actuators of the microrobot. Once the driving circuit is constructed using HNN, the controlling circuit and the driving circuit can be integrated into a single chip. In this paper, the authors will propose the driving circuit using HNN. The HNN consists of two self-oscillating cell body models, six separately-excited cell body models, four excitatory-synaptic models, and six inhibitory-synaptic models. The single self-oscillating cell body model outputs the electrical oscillated square waveform as 3 MHz of frequency. The proposed HNN generates a long delay without using large capacitors. As a result, the proposed HNN can generate the driving waveform of electrostatic actuators with variable frequency. The frequency of the driving waveform could vary from 50 to 100 Hz. Also, the proposed HNN connected to the Central Pattern Generator (CPG) model. The CPG model with proposed HNN outputs the driving waveform of the electrostatic actuator which can perform the tripod gait pattern of the microrobot.