Balanced Standing on One Foot of Biped Robot Based on Three-Particle Model Predictive Control

Area of Science:

• Robotics
• Control Systems
• Mechatronics

Background:

• One-foot balancing in bipedal robots presents significant challenges due to a reduced support base and complex dynamics.
• Existing methods like nonlinear model predictive control (NMPC) are computationally intensive for real-time application.

Purpose of the Study:

• To develop a computationally efficient control strategy for real-time one-leg balancing in bipedal robots.
• To improve the robot's ability to withstand external disturbances during single-leg stance.

Main Methods:

• Proposed a novel three-particle model predictive control (TP-MPC) approach, simplifying the robot model to three inertialess particles.
• Integrated TP-MPC with hierarchical whole-body control (WBC) for dynamic balancing.
• TP-MPC generates swing leg trajectories, while WBC adjusts the center of mass.

Main Results:

• The linear nature of the three-particle model significantly reduces computational load, enabling real-time execution.
• Simulations demonstrated that the proposed TP-MPC combined with WBC can effectively resist larger external disturbances compared to WBC alone.

Conclusions:

• The TP-MPC approach offers a feasible and effective solution for real-time one-leg balancing in bipedal robots.
• This method enhances robustness to external disturbances, paving the way for more stable robotic locomotion.