Gait generation and transition for a five-link biped robot by Central Pattern Generator

In recent decades, the locomotion control has been a main research field in robotics. In locomotion control problems, understanding the relationship between neuroscience and robotics is very important. In this work, a bio-inspired method for the gait generation and transition of a biped robot is implemented. This method uses a Central Pattern Generator (CPG), which is a neural network responsible for generating rhythmic output signals without sensory information. Humans and animals have the ability to produce various gaits by CPG according to the environmental conditions. The 4-cell CPG model proposed by Golubitsky et al, and symmetry in this model are used for walk and run gaits generation. In this work, Transitions from walk to skip and from run to skip are achieved by changing coupling weights of walk and run. The Morris-Lecar oscillator is used as the oscillator of each cell. The numerical simulations of periodic signals corresponding to the five-link bipedal walk and run gaits and transitions from walk to skip and run to skip are shown in the 4-cell CPG model. The periodic signals resulting of this model can be used as joint angles of hip and knee joints of a five-link biped robot. The results obtained in this work show the feasibility of using CPG-based methods to generate five-link biped locomotion and transitions between different gaits.