The content discusses a trajectory optimization framework focusing on agile maneuvers for quadruped robots, specifically the ANYmal C. The framework leverages full-centroidal dynamics and implicit inverse kinematics to optimize foothold locations and contact forces directly, enabling high-acceleration maneuvers with low computational overhead. Real-world experiments demonstrate successful execution of high-acceleration motions like linear and rotational jumps, surpassing hardware limitations. The method requires minimal user references and enhances convergence rates compared to state-of-the-art approaches.
Vers une autre langue
à partir du contenu source
arxiv.org
Questions plus approfondies