Evolutionary Deployment of Central Pattern Generators for Legged Robots Using Nengo
摘要
In this paper, we present an evolutionary design methodology of Central Pattern Generator (CPG)-based locomotion systems for hexapod and quadrupedal robots. The CPGs are built as Spiking Neural Networks, whose synaptic connections and weights are directly configured by an evolutionary algorithm in order that CPGs generate rhythmic and periodical signals to carry out robotic locomotion. The CPGs are fully designed and implemented using Nengo simulator. There were obtained CPGs for different locomotion patterns for both, hexapod and quadruped robots. The obtained CPGs achieve behaviours reported in the state of the art with smaller architectures.