Research map / Vision-language and multimodal / Robot action and navigation
Towards Learning a Generic Agent for Vision-and-Language Navigation via Pre-Training
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 changed | Where | Measured | Effect |
|---|---|---|---|
| pre-train a vision-language navigation encoder and transfer it to R2R | Prevalent · R2R test unseen, single instruction | SPL: 47 -> 51 | Helped |
| pre-train a vision-language navigation encoder and transfer it to R2R | Prevalent · R2R test unseen, multiple instructions | SPL: 53 -> 56 | Helped |
| transfer the R2R-pre-trained encoder to CVDN | Prevalent · CVDN test unseen, Mixed | Goal Progress: 2.35 -> 2.44 | Helped |
| transfer the R2R-pre-trained encoder to HANNA | Prevalent · HANNA test unseen, UNSEEN-ALL | SPL: 25.50 -> 28.72 | Helped |
| add action prediction loss to MLM pre-training (ablation) | Prevalent · CVDN All input, Mixed | Goal Progress: 3.14 -> 3.15 | Helped |
| replace image-attended MLM with BERT MLM pre-training (ablation) | Prevalent · CVDN All input, Mixed | Goal Progress: 3.15 -> 2.64 | Hurt |
| replace VLN pre-training with off-the-shelf BERT fine-tuning (ablation) | Prevalent · CVDN All input, Mixed | Goal Progress: 3.15 -> 2.28 | Hurt |
| fine-tune cross-attention layers after feature-based training (ablation) | Prevalent · R2R test unseen | SPL: 0.49 -> 0.51 | Helped |
| add action prediction loss to MLM pre-training (ablation) | Prevalent · HANNA test UNSEEN-ALL | SPL: 24.27 -> 28.72 | Helped |
| replace VLN pre-training with BERT feature-based representations (ablation) | Prevalent · HANNA test UNSEEN-ALL | SPL: 28.72 -> 11.50 | Hurt |
| replace VLN pre-training with BERT fine-tuning (ablation) | Prevalent · HANNA test UNSEEN-ALL | SPL: 28.72 -> 12.66 | Hurt |
Closest papers by meaning
- Improving Vision-and-Language Navigation by Generating Future-View Image Semantics (CVPR 2023 · similarity 0.93)
- Curriculum Learning for Vision-and-Language Navigation (NeurIPS 2021 · similarity 0.92)
- A New Path: Scaling Vision-and-Language Navigation With Synthetic Instructions and Imitation Learning (CVPR 2023 · similarity 0.92)
- ADAPT: Vision-Language Navigation With Modality-Aligned Action Prompts (CVPR 2022 · similarity 0.91)
- NavGPT: Explicit Reasoning in Vision-and-Language Navigation with Large Language Models (AAAI 2024 · similarity 0.91)
- General Scene Adaptation for Vision-and-Language Navigation (ICLR 2025 · similarity 0.90)
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.