CPG-based online trajectory generation for quadruped rovers

Legged rovers are often considered as viable solutions for traversing unknown terrain. A 2D sagittal plane rover model, based on a domestic cat, is considered in this paper, and an online model-free gait planning framework is implemented using Central Pattern Generators. The framework is used to generate joint trajectories for any arbitrarily varying speed profile, and regulate locomotion transition and speed modulation, both continuously and endogenously. For a continuously changing speed profile, the gait transition is continuous as well. For a discontinuously changing speed profile, a hopping motion is observed, because the rover's thrust speed cannot increase discontinuously.