Central pattern generators with biology observation for the locomotion control of hexapod robots

Abstract

This paper focuses on the control of hexapod locomotion based on a model of artificial central pattern generators (CPGs). CPG-based controllers are capable of producing coordinated patterns in open loops and rhythmic activities in certain joints. However, existing methods usually have too many parameters to set and lack unified expressions to achieve desired gaits for multiple legged robots. In this contribution, biology observation and CPG modeling are employed to build the controller. Coupled nonlinear oscillators serve as the elementary unit and coupling terms are discussed to get the typical hexapod locomotion patterns. At first, we collect the locomotion date of real cockroaches Blaptica Dubia and complete some data analysis with the approach of image processing. Then modified Hopf oscillators are applied to separately control the swing phase and stance phase, and analytical formulation of coupling terms is adopted to construct the control architecture. Next, we modulate the proposed controller on the basis of three-dimensional trace of cockroaches by kinematic analysis and curve fitting. Finally, we build and control two different hexapod robots (`hexabot' and `smarbot') to confirm the validation of the approach.