AI Paper Atlas

Research map / Vision-language and multimodal / Robot action and navigation

Beyond the Nav-Graph: Vision-and-Language Navigation in Continuous Environments

Jacob Krantz, Erik Wijmans, Arjun Majumdar, Dhruv Batra, Stefan Lee

ECCV 2020 · Vision-language and multimodal › Robot action and navigation · approach: Pretraining and data augmentation for VLN · 334 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
use cross-modal attention instead of Seq2Seq (teacher forcing)Seq2Seq baseline · VLN-CE val-unseenSPL: 0.18 -> 0.22Improved
use cross-modal attention instead of Seq2Seq (with all three augmentations)Seq2Seq PM+DA*+Aug · VLN-CE val-unseenSPL: 0.22 -> 0.30Improved
add DAgger training (ablation)Seq2Seq baseline · VLN-CE val-unseenSPL: 0.18 -> 0.23Improved
add DAgger training (ablation)Cross-Modal · VLN-CE val-unseenSPL: 0.22 -> 0.25Improved
add progress monitor loss (ablation)Seq2Seq baseline · VLN-CE val-unseenSPL: 0.18 -> 0.15Hurt
add synthetic data augmentation (ablation)Seq2Seq baseline · VLN-CE val-unseenSPL: 0.18 -> 0.17Hurt
add progress monitor loss (ablation)Cross-Modal · VLN-CE val-unseenSPL: 0.22 -> 0.19Hurt
add synthetic data augmentation (ablation)Cross-Modal · VLN-CE val-unseenSPL: 0.22 -> 0.19Hurt
combine progress monitor, data augmentation and DAgger fine-tuningCross-Modal · VLN-CE val-unseenSPL: 0.22 -> 0.30Improved
add progress monitor to augmented training (ablation)Cross-Modal + Aug · VLN-CE val-unseenSPL: 0.19 -> 0.22Improved
add data augmentation to DAgger fine-tuning (ablation)Cross-Modal + DA · VLN-CE val-unseenSPL: 0.25 -> 0.26Improved
add DAgger fine-tuning to PM+Aug pretraining (ablation)Cross-Modal PM+Aug · VLN-CE val-unseenSPL: 0.22 -> 0.30Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy