Exploiting Arch-like Foot Structure for Knee-Extended Walking in Bipedal Robots

Area of Science:

• Robotics
• Biomechanics
• Control Systems

Background:

• Human walking efficiency relies on knee extension, a challenging feat for bipedal robots due to pose singularities and high joint velocities.
• Traditional control methods struggle with efficient knee-extended locomotion in humanoid robots.

Purpose of the Study:

• To develop a pattern generation method for stable and energy-efficient knee-extended walking in bipedal robots.
• To simulate human arch motion using the inertial linear inverted pendulum model (ILIPM).
• To enhance robot stability through compliant control and advanced error correction techniques.

Main Methods:

• Utilized the inertial linear inverted pendulum model (ILIPM) for pattern generation.
• Designed a quadrilateral foot structure and compliant virtual leg control.
• Implemented a combined linear feedback and ankle joint strategy for divergent component of motion (DCM) error correction.

Main Results:

• Knee-extended walking with compliance control achieved the lowest energy consumption and minimized center of mass (COM) velocity oscillations.
• ILIPM-based walking demonstrated stable COM trajectory oscillations (amplitude ~0.015 m).
• ILIPM outperformed LIPM and Flywheel LIPM in maintaining COM posture angle and angular momentum, improving stability.

Conclusions:

• The proposed ILIPM-based pattern generation method with compliant control enables efficient and stable knee-extended walking in bipedal robots.
• Combined ankle joint and linear feedback control effectively corrects DCM errors, enhancing dynamic stability.
• This approach offers a promising solution for improving humanoid robot locomotion.