Control of Humanoid in Multiple Fixed and Moving Unilateral Contacts

TL;DR

This paper presents an improved online balance control method for humanoid robots in complex contact scenarios, enabling better task compliance without pre-computed balance targets.

Contribution

Enhances geometric balance methods for real-time integration into whole-body control, allowing flexible contact force distribution without pre-defined center-of-mass targets.

Findings

Successful implementation on HRP-4 humanoid robot

Improved computational performance for online balance control

Enhanced task compliance in complex contact scenarios

Abstract

Enforcing balance of multi-limbed robots in multiple non-coplanar unilateral contact settings is challenging when a subset of such contacts are also induced in motion tasks. The first contribution of this paper is in enhancing the computational performance of state-of-the-art geometric center-of-mass inclusion-based balance method to be integrated online as part of a task-space whole-body control framework. As a consequence, our second contribution lies in integrating such a balance region with relevant contact force distribution without pre-computing a target center-of-mass. This last feature is essential to leave the latter with freedom to better comply with other existing tasks that are not captured in classical twolevel approaches. We assess the performance of our proposed method through experiments using the HRP-4 humanoid robot.