On Embodied Visual Navigation in Real Environments Through Habitat

TL;DR

This paper introduces a Habitat-based tool that uses real-world images and noise models to improve the training and evaluation of visual navigation policies, enhancing their real-world generalization without physical deployment.

Contribution

It presents a novel approach combining real-world images with simulated noise in Habitat to better train and assess navigation policies for real environments.

Findings

Policies trained with real images and noise models generalize better to real-world scenarios.

Sensor and actuator noise improve the robustness of navigation policies.

The tool enables effective training and evaluation without real-world navigation episodes.

Abstract

Visual navigation models based on deep learning can learn effective policies when trained on large amounts of visual observations through reinforcement learning. Unfortunately, collecting the required experience in the real world requires the deployment of a robotic platform, which is expensive and time-consuming. To deal with this limitation, several simulation platforms have been proposed in order to train visual navigation policies on virtual environments efficiently. Despite the advantages they offer, simulators present a limited realism in terms of appearance and physical dynamics, leading to navigation policies that do not generalize in the real world. In this paper, we propose a tool based on the Habitat simulator which exploits real world images of the environment, together with sensor and actuator noise models, to produce more realistic navigation episodes. We perform a range of experiments to assess the ability of such policies to generalize using virtual and real-world images, as well as observations transformed with unsupervised domain adaptation approaches. We also assess the impact of sensor and actuation noise on the navigation performance and investigate whether it allows to learn more robust navigation policies. We show that our tool can effectively help to train and evaluate navigation policies on real-world observations without running navigation pisodes in the real world.