Iterative Learning-Based Centre-Of-Mass Impedance Control for Articulated-Soft Humanoid Robots
Yibin Wang, Lin Zhou, Sacha Morris, Shan Luo, Emmanouil Spyrakos-Papastavridis
AI summary
Problem
Articulated-soft humanoid robots struggle with balance and disturbance rejection due to flexible joints, high degrees of freedom, and highly nonlinear dynamics, making existing rigid-body or model-heavy controllers ineffective for safe physical interaction.
Approach
The authors develop a contact-force-based iterative learning control framework that iteratively updates a gross force compensation term using zero-moment point errors from previous trials, integrated with a center-of-mass impedance controller and stabilized via power-shaping signals.
Key results
- ZMP-error-driven iterative learning scheme for Cartesian force compensation
- Theoretical stability guarantee via DCM error convergence and power-shaping control
- Successful sim-to-real transfer on the compliant BRUCE humanoid
- Superior external impact rejection and recovery stability compared to baseline controllers
Why it matters
Enables safer, more robust physical interaction and balance recovery for next-generation compliant humanoid robots in unstructured human environments.
Abstract
Achieving safe and robust interaction in articulated-soft humanoid robots (ASRs) remains a major challenge due to their compliant joints, high degree of freedom, and highly nonlinear coupled dynamics, which makes them especially sensitive to external disturbances. This paper presents a novel contact-force-based iterative learning center-of-mass (CoM) impedance control framework (CF-IL-CIC) specifically designed to enhance disturbance robustness in floating-base ASRs. The key idea is to iteratively derive a time-series gross force compensation term from zero moment point (ZMP) tracking errors of previous trials, using a proportional-derivative (PD)-type update rule in simulation. This compensation is integrated with a contact-force-based CoM impedance controller to improve push recovery without requiring precise dynamic models or heavy online optimization. The approach is accompanied by mathematical proof of divergent component of motion (DCM) error convergence, ensuring theoretical stability guarantees. The proposed method is validated through both dynamic simulations and real-robot experiments on the compliant humanoid BRUCE, demonstrating significant improvements in external impact rejection and recovery stability compared to baseline controllers.