Unsupervised Contact Learning for Humanoid Estimation and Control

TL;DR

This paper introduces an unsupervised fuzzy clustering method for estimating contact states in humanoid robots using only proprioceptive sensors, improving base state estimation over traditional force-based methods.

Contribution

It presents a novel unsupervised contact probability estimator that leverages fuzzy clustering of proprioceptive data for enhanced humanoid robot state estimation.

Findings

Contact probability estimation improves state accuracy.

Method performs well on rough, low-friction terrain.

Outperforms force-based contact detection methods.

Abstract

This work presents a method for contact state estimation using fuzzy clustering to learn contact probability for full, six-dimensional humanoid contacts. The data required for training is solely from proprioceptive sensors - endeffector contact wrench sensors and inertial measurement units (IMUs) - and the method is completely unsupervised. The resulting cluster means are used to efficiently compute the probability of contact in each of the six endeffector degrees of freedom (DoFs) independently. This clustering-based contact probability estimator is validated in a kinematics-based base state estimator in a simulation environment with realistic added sensor noise for locomotion over rough, low-friction terrain on which the robot is subject to foot slip and rotation. The proposed base state estimator which utilizes these six DoF contact probability estimates is shown to perform considerably better than that which determines kinematic contact constraints purely based on measured normal force.