Active Improvement of Control Policies with Bayesian Gaussian Mixture Model

TL;DR

This paper presents an active learning framework that enhances control policy generalization in robot learning from demonstration by leveraging Bayesian Gaussian mixture models and information-density optimization.

Contribution

It introduces a novel active learning method based on BGMMs and quadratic Rényi entropy to improve robot control policies from demonstrations.

Findings

Effective in cluttered environments for reaching tasks

Improves generalization with fewer demonstrations

Validated on both toy example and real robot

Abstract

Learning from demonstration (LfD) is an intuitive framework allowing non-expert users to easily (re-)program robots. However, the quality and quantity of demonstrations have a great influence on the generalization performances of LfD approaches. In this paper, we introduce a novel active learning framework in order to improve the generalization capabilities of control policies. The proposed approach is based on the epistemic uncertainties of Bayesian Gaussian mixture models (BGMMs). We determine the new query point location by optimizing a closed-form information-density cost based on the quadratic R\'enyi entropy. Furthermore, to better represent uncertain regions and to avoid local optima problem, we propose to approximate the active learning cost with a Gaussian mixture model (GMM). We demonstrate our active learning framework in the context of a reaching task in a cluttered environment with an illustrative toy example and a real experiment with a Panda robot.