Research map / Vision-language and multimodal / Robot action and navigation
Landmark-RxR: Solving Vision-and-Language Navigation with Fine-Grained Alignment Supervision
Keji He, Yan Huang, Qi Wu, Jianhua Yang, Dong An, Shuanglin Sima, Liang Wang
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 |
|---|---|---|---|
| replace nDTW fidelity reward with soft focal-oriented reward (ablation) | RL agent · en-RxR val unseen | SR (%): 29.3 -> 33.7 | Improved |
| replace nDTW fidelity reward with soft focal-oriented reward (ablation) | RL agent · Landmark-RxR val unseen | Loss Number: 5500 -> 5173 | Improved |
| replace nDTW fidelity reward with soft focal-oriented reward (ablation) | RL agent · en-RxR val unseen | nDTW: 43.2 -> 41.0 | Hurt |
| replace goal-oriented SR reward with hard focal-oriented reward (ablation) | RL agent · en-RxR val unseen | SR (%): 32.7 -> 34.9 | Improved |
| replace fidelity+SR reward with hard focal-oriented reward (ablation) | RL agent · en-RxR val unseen | SR (%): 33.6 -> 34.9 | Improved |
| replace fidelity+SR reward with soft focal-oriented reward (ablation) | RL agent · Landmark-RxR val unseen | SSA: 52.5 -> 53.3 | Improved |
| replace complete en-RxR pairs with Landmark-RxR sub pairs in training (ablation) | IL agent · R2R val unseen | SR (%): 33.1 -> 40.1 | Improved |
| add complete en-RxR pairs to R2R training (ablation) | IL agent · R2R val unseen | SR (%): 33.0 -> 33.1 | No clear effect |
| add sub pairs to complete pairs in training (ablation) | IL agent · en-RxR val unseen | SR (%): 20.7 -> 26.2 | Improved |
| train on joint sub, synthesized and complete pairs (ablation) | IL agent · en-RxR val unseen | SR (%): 26.2 -> 28.4 | Improved |
| train on synthesized pairs instead of complete pairs (ablation) | IL agent · en-RxR val unseen | SR (%): 20.7 -> 24.2 | Improved |
Closest papers by meaning
- Fine-Grained Alignment Supervision Matters in Vision-and-Language Navigation (TPAMI 2026 · similarity 0.97)
- Grounded Entity-Landmark Adaptive Pre-Training for Vision-and-Language Navigation (ICCV 2023 · similarity 0.90)
- LangMap: A Human-Verified Benchmark for Hierarchical Open-Vocabulary Goal Navigation (NeurIPS 2026 · similarity 0.87)
- Towards Long-Horizon Vision-Language Navigation: Platform, Benchmark and Method (CVPR 2025 · similarity 0.87)
- Learning Fine-Grained Alignment for Aerial Vision-Dialog Navigation (AAAI 2025 · similarity 0.87)
- Scene Map-based Prompt Tuning for Navigation Instruction Generation (CVPR 2025 · similarity 0.87)
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.
