Analog CMOS circuits for bio-inspired locomotion controller - a comparison with Amari-Hopfield analog CPG circuit and Wilson-Cowan analog CPG circuit

摘要

In this report, we propose analog CMOS circuits that control rhythmic movements of locomotion in robotics. Animal locomotion, such as walking, running, swimming and flying is generated and controlled by the biological neural network, called the central pattern generator (CPG). In recent years, many researchers have applied CPG to movement control in robotics. However, most of these have been developed with digital processors that have several problems such as high power consumption. Hence, we designed two analog CMOS circuits that control rhythmic coordination of locomotion in robotics. In order to make use of the nonlinerity of analog circuit efficiently, each circuit is based on the Amari-Hopfield neuron model and Wilson-Cowan neuron model respectively. We confirmed operation of the circuits by using SPICE. As a result, it was shown that each circuit has the capability of generating various rhythmic patterns and the stability against disturbance.