Flexible Model Predictive Control for Bounded Gait Generation in Humanoid Robots

  • 0Institute of Automation, Chinese Academy of Sciences, Beijing 100089, China.

|

|

Summary

This summary is machine-generated.

Humanoid robots require bounded gaits for stability. A new Flexible Model Predictive Control (FMPC) framework uses an enhanced flexible model and constraints to ensure stable, bounded motion, validated in simulations and real robots.

Area Of Science

  • Robotics
  • Control Systems
  • Humanoid Locomotion

Background

  • Traditional Model Predictive Control (MPC) methods using Linear Inverted Pendulum (LIP) or Cart-Table (C-T) models are insufficient for robots with flexible joints.
  • Achieving stable bipedal locomotion in complex environments necessitates bounded gaits.

Purpose Of The Study

  • To propose a Flexible Model Predictive Control (FMPC) framework for humanoid robots.
  • To enable stable and bounded gait control by incorporating joint dynamics and advanced constraints.

Main Methods

  • Developed an enhanced flexible Cart-Table (C-T) model with an elastic layer and auxiliary center of mass (CoM).
  • Integrated Zero Moment Point (ZMP) velocity as a control variable.
  • Formulated a quadratic programming (QP) problem with CoM, bounded, and ZMP constraints for bounded CoM trajectories.

Main Results

  • The FMPC framework successfully generated bounded CoM/ZMP trajectories across diverse simulated conditions.
  • Simulations demonstrated the method's capacity to enhance gait control and stability for flexible humanoid robots.
  • Validation on CASBOT and Openloong robots confirmed the approach's effectiveness and robustness.

Conclusions

  • The proposed FMPC framework effectively addresses limitations of traditional MPC for flexible humanoid robots.
  • The method enhances stability and enables bounded gait control in various operational scenarios.
  • The FMPC approach shows significant potential for improving humanoid robot locomotion and real-world applicability.