Robustness of Utilizing Feedback in Embodied Visual Navigation

TL;DR

This paper introduces a training framework for embodied visual navigation agents that actively request feedback and remain robust when feedback is intermittently unavailable, improving overall navigation performance.

Contribution

It proposes a curriculum-based training method combining episodes with and without feedback to enhance agent robustness in object-goal navigation.

Findings

Improved navigation performance with mixed feedback training.

Agents maintain effectiveness even without feedback during deployment.

Robustness increases in scenarios with inconsistent feedback availability.

Abstract

This paper presents a framework for training an agent to actively request help in object-goal navigation tasks, with feedback indicating the location of the target object in its field of view. To make the agent more robust in scenarios where a teacher may not always be available, the proposed training curriculum includes a mix of episodes with and without feedback. The results show that this approach improves the agent's performance, even in the absence of feedback.