Improving Robot Localisation by Ignoring Visual Distraction

TL;DR

This paper introduces Neural Blindness, a method for neural networks to ignore specific distractor classes, enhancing robot localisation by focusing on relevant visual information and reducing distraction.

Contribution

The paper presents Neural Blindness, a novel approach enabling neural networks to completely ignore designated distractor classes, improving localisation accuracy.

Findings

Neural Blindness effectively suppresses distractor classes in neural representations.

Ignoring distractors improves localisation performance in cluttered environments.

The method demonstrates potential for more robust robot perception systems.

Abstract

Attention is an important component of modern deep learning. However, less emphasis has been put on its inverse: ignoring distraction. Our daily lives require us to explicitly avoid giving attention to salient visual features that confound the task we are trying to accomplish. This visual prioritisation allows us to concentrate on important tasks while ignoring visual distractors. In this work, we introduce Neural Blindness, which gives an agent the ability to completely ignore objects or classes that are deemed distractors. More explicitly, we aim to render a neural network completely incapable of representing specific chosen classes in its latent space. In a very real sense, this makes the network "blind" to certain classes, allowing and agent to focus on what is important for a given task, and demonstrates how this can be used to improve localisation.