Research map / Vision-language and multimodal / Robot action and navigation
Active Visual Information Gathering for Vision-Language Navigation
Hanqing Wang, Wenguan Wang, Tianmin Shu, Wei Liang, Jianbing Shen
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 changed | Where | Measured | Effect |
|---|---|---|---|
| add naive one-step exploration of all directions (ablation) | basic agent · R2R val unseen | SR: 0.52 -> 0.54 | Improved |
| add naive one-step exploration of all directions (ablation) | basic agent · R2R val unseen | TL: 10.7 -> 35.7 | Hurt |
| add learned exploration decision (explore only some directions) (ablation) | naive model · R2R val unseen | TL: 35.7 -> 13.7 | Improved |
| add learned exploration decision (explore only some directions) (ablation) | naive model · R2R val unseen | SR: 0.54 -> 0.55 | Improved |
| allow multi-step exploration (at most 4 steps) in the naive model (ablation) | naive model · R2R val unseen | SR: 0.54 -> 0.60 | Improved |
| allow multi-step exploration (at most 4 steps) in the naive model (ablation) | naive model · R2R val unseen | TL: 35.7 -> 58.8 | Hurt |
| raise maximum exploration length from 1 to 4 steps (ablation) | full model · R2R val unseen | SR: 0.55 -> 0.58 | Improved |
| raise maximum exploration length from 4 to 6 steps (ablation) | full model · R2R val unseen | SPL: 0.40 -> 0.38 | Mixed |
| add imitation loss on exploration module (ablation) | full model · R2R val unseen | SR: 0.53 -> 0.58 | Improved |
| add reinforcement loss on exploration module (ablation) | full model · R2R val unseen | SR: 0.55 -> 0.58 | Improved |
| add memory-based lazy action-taking strategy (ablation) | full model · R2R val unseen | TL: 44.5 -> 20.6 | Improved |
| add memory-based lazy action-taking strategy (ablation) | full model · R2R val unseen | SPL: 0.18 -> 0.40 | Improved |
Closest papers by meaning
- AwareVLN: Reasoning with Self-awareness for Vision-Language Navigation (CVPR 2026 · similarity 0.91)
- DecoVLN: Decoupling Observation, Reasoning, and Correction for Vision-and-Language Navigation (CVPR 2026 · similarity 0.91)
- Improving Vision-and-Language Navigation by Generating Future-View Image Semantics (CVPR 2023 · similarity 0.90)
- Active Test-time Vision-Language Navigation (NeurIPS 2025 · similarity 0.90)
- Vision-Language Navigation with Energy-Based Policy (NeurIPS 2024 · similarity 0.89)
- Towards Long-Horizon Vision-Language Navigation: Platform, Benchmark and Method (CVPR 2025 · similarity 0.89)
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.
