Multimodal Attention Networks for Low-Level Vision-and-Language Navigation

TL;DR

This paper introduces PTA, a fully-attentive Transformer-based architecture for low-level vision-and-language navigation that effectively integrates multiple modalities and long-term dependencies, improving navigation performance.

Contribution

The paper presents PTA, the first Transformer-like model for low-level VLN that combines natural language, images, and actions with early and late fusion strategies.

Findings

PTA achieves promising results on R2R dataset.

PTA performs well on the R4R benchmark.

The model effectively handles multi-modality and long-term dependencies.

Abstract

Vision-and-Language Navigation (VLN) is a challenging task in which an agent needs to follow a language-specified path to reach a target destination. The goal gets even harder as the actions available to the agent get simpler and move towards low-level, atomic interactions with the environment. This setting takes the name of low-level VLN. In this paper, we strive for the creation of an agent able to tackle three key issues: multi-modality, long-term dependencies, and adaptability towards different locomotive settings. To that end, we devise "Perceive, Transform, and Act" (PTA): a fully-attentive VLN architecture that leaves the recurrent approach behind and the first Transformer-like architecture incorporating three different modalities - natural language, images, and low-level actions for the agent control. In particular, we adopt an early fusion strategy to merge lingual and visual information efficiently in our encoder. We then propose to refine the decoding phase with a late fusion extension between the agent's history of actions and the perceptual modalities. We experimentally validate our model on two datasets: PTA achieves promising results in low-level VLN on R2R and achieves good performance in the recently proposed R4R benchmark. Our code is publicly available at https://github.com/aimagelab/perceive-transform-and-act.