Scholarlink

<p>This paper proposes a motion control framework that integrates terrain complexity analysis, aiming to address the motion control challenges of quadruped robots in complex terrains. First, a visual perception system constructs a terrain elevation map, and a novel Terrain Complexity Index (TCI) is proposed to quantify terrain features in real time. The TCI enables adaptive velocity reference generation and optimal foothold selection, which are incorporated as key inputs into the motion optimization framework. Second, a Nonlinear Model Predictive Control (NMPC) approach jointly optimizes the terrain-driven velocity reference and foothold strategy, solving the time-domain optimal control problem under dynamic and kinematic constraints to compute optimal foot-end contact forces. Experimental results demonstrate that this method effectively minimizes center-of-mass oscillations during locomotion, significantly enhancing motion stability. Finally, a Whole-Body Control (WBC) system calculates joint torques, positions, and velocities based on the reaction forces output by the NMPC and the real-time velocity reference, forming a complete motion control chain. Experiments on the Cyberdog platform demonstrate that this method can synergistically improve velocity adaptability, motion stability, and system robustness in complex terrains.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Motion control of quadruped robots in complex terrain

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年11月5日

查看全文
  • Hongbo Kang,
  • Zhuang Wang,
  • Kang Wang,
  • Chunjie Yang

摘要

This paper proposes a motion control framework that integrates terrain complexity analysis, aiming to address the motion control challenges of quadruped robots in complex terrains. First, a visual perception system constructs a terrain elevation map, and a novel Terrain Complexity Index (TCI) is proposed to quantify terrain features in real time. The TCI enables adaptive velocity reference generation and optimal foothold selection, which are incorporated as key inputs into the motion optimization framework. Second, a Nonlinear Model Predictive Control (NMPC) approach jointly optimizes the terrain-driven velocity reference and foothold strategy, solving the time-domain optimal control problem under dynamic and kinematic constraints to compute optimal foot-end contact forces. Experimental results demonstrate that this method effectively minimizes center-of-mass oscillations during locomotion, significantly enhancing motion stability. Finally, a Whole-Body Control (WBC) system calculates joint torques, positions, and velocities based on the reaction forces output by the NMPC and the real-time velocity reference, forming a complete motion control chain. Experiments on the Cyberdog platform demonstrate that this method can synergistically improve velocity adaptability, motion stability, and system robustness in complex terrains.