A Robust Disturbance Rejection Whole-Body Control Framework for Bipedal Robots Using a Momentum-Based Observer

Area of Science:

• Robotics
• Control Systems
• Mechanical Engineering

Background:

• Bipedal robots require robust control strategies to maintain stability during locomotion.
• External disturbances pose a significant challenge to bipedal robot performance and safety.
• Current methods often rely on foot-mounted sensors for disturbance estimation, limiting their applicability.

Purpose of the Study:

• To develop a comprehensive planner and controller scheme for bipedal robots.
• To enhance robustness against external disturbances without relying on foot-mounted sensors.
• To improve the accuracy of ground reaction force estimation and disturbance compensation.

Main Methods:

• High-level planning using Model Predictive Control (MPC) optimized foothold location and step duration based on the divergent component of motion (DCM).
• Low-level control employing a momentum-based observer to estimate external forces on stance and swing legs.
• Integration of full-body dynamics and estimated disturbances into Weighted Whole-Body Control (WBC) for accurate ground reaction force calculation.

Main Results:

• Demonstrated enhanced gait robustness against external disturbances.
• Successfully estimated external forces without foot-mounted sensors.
• Validated the controller's effectiveness in mitigating disturbance effects through simulations and experiments on the BRUCE robot.

Conclusions:

• The proposed planner and controller scheme significantly improves bipedal robot robustness against external disturbances.
• The method offers a sensor-independent approach to disturbance estimation and compensation.
• Experimental validation confirms the framework's effectiveness in diverse challenging scenarios.