Programmable Bionic Control Circuit Based on Central Pattern Generator

Abstract

The central pattern generator (CPG) involves a group of neurons that produce rhythmic signals in a coordinated manner. Currently, CPG circuits capable of efficient online programming are rarely found in the literature. To address this issue, this article proposes a memristive control circuit based on CPG. First, an online amplification module is designed to adjust the positive and negative amplification coefficients. On the basis of this structure, a CPG unit circuit controlling a joint is proposed. According to the topology of CPG network model, a CPG network circuit composed of multiple units is devised. This network can coordinate multiple joints to produce a gait. In this article, the circuit is applied to generate the activity pattern of fish swimming. PSPICE simulation results demonstrate that four units can realize the basic swimming patterns of a robot fish. Through memristor programming, the circuit can achieve smooth online switching of robot fish swimming patterns. Moreover, hardware implementation proves the practicality of the circuit.