Among various motion control strategies for quadruped robot control, Virtual Model Control (VMC) stands out for its intuitive design and its ability to achieve compliant interaction with the ground. However, a key challenge in applying VMC is the tuning of numerous control parameters. This paper presents a parameter tuning strategy for VMC, aiming to improve both control performance and implementation efficiency. The approach is demonstrated using the Unitree Go1 quadruped robot within the Webots simulator. A MATLAB-aided identification and offline tuning method is developed and applied, with height control selected as a representative task. Simulation results under different speed settings demonstrate that the proposed method improves overall control performance and stability. In addition, this method can be extended to optimize the virtual element parameters within other control frameworks that integrate the swing-phase VMC approach. Overall, this work presents a practical and lightweight alternative to computationally intensive optimization or learning-based methods, providing a reliable and effective solution for quadruped robots in simple task scenarios.

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

Parameter-Tuned Virtual Model Control for Quadruped Robot

  • Haiying Yuan,
  • Keng Goh,
  • Peter Andras,
  • Wenguang Luo

摘要

Among various motion control strategies for quadruped robot control, Virtual Model Control (VMC) stands out for its intuitive design and its ability to achieve compliant interaction with the ground. However, a key challenge in applying VMC is the tuning of numerous control parameters. This paper presents a parameter tuning strategy for VMC, aiming to improve both control performance and implementation efficiency. The approach is demonstrated using the Unitree Go1 quadruped robot within the Webots simulator. A MATLAB-aided identification and offline tuning method is developed and applied, with height control selected as a representative task. Simulation results under different speed settings demonstrate that the proposed method improves overall control performance and stability. In addition, this method can be extended to optimize the virtual element parameters within other control frameworks that integrate the swing-phase VMC approach. Overall, this work presents a practical and lightweight alternative to computationally intensive optimization or learning-based methods, providing a reliable and effective solution for quadruped robots in simple task scenarios.