Decoupling feature extraction from policy learning: assessing benefits of state representation learning in goal based robotics

TL;DR

This paper evaluates various state representation learning methods for goal-based robotics, proposing a new unsupervised model that improves sample efficiency and robustness compared to end-to-end reinforcement learning.

Contribution

It introduces a novel unsupervised stacking model that combines multiple representation learning approaches, enhancing efficiency and interpretability in robotic control tasks.

Findings

The proposed model encodes relevant features effectively.

It outperforms or matches end-to-end learning in sample efficiency.

The method is robust to hyper-parameter variations.

Abstract

Scaling end-to-end reinforcement learning to control real robots from vision presents a series of challenges, in particular in terms of sample efficiency. Against end-to-end learning, state representation learning can help learn a compact, efficient and relevant representation of states that speeds up policy learning, reducing the number of samples needed, and that is easier to interpret. We evaluate several state representation learning methods on goal based robotics tasks and propose a new unsupervised model that stacks representations and combines strengths of several of these approaches. This method encodes all the relevant features, performs on par or better than end-to-end learning with better sample efficiency, and is robust to hyper-parameters change.