Free Gait Transition and Stable Motion Generation Using CPG-based Locomotion Control for Hexapod Robots

In this paper, we address the problem of gait transition for legged robots, focusing on the challenges of versatile phase transitions and stable motion generation. To tackle the phase transition problem, we begin by analyzing the undesirable phase-lock phenomenon and identifying the phase lock conditions for conventional diffusively coupled nonlinear oscillators. We then propose a novel diffusive Central Pattern Generator (CPG) model and theoretically prove its ability to handle phase transitions across various gait patterns. For the stability problem during transitions from one gait to another, we introduce a motion generation framework that provides an interface for stability criteria, enabling smooth trajectory adjustments based on the robot's stability. We evaluate the proposed method using both simulations and a physical hexapod robot. Experimental results demonstrate that our approach can achieve transitions across a wide range of gait patterns while ensuring stable and smooth motion during the transitions.