Research map / Vision-language and multimodal / Robot action and navigation
Constraint-Aware Zero-Shot Vision-Language Navigation in Continuous Environments
Kehan Chen, Dong Un An, Yan Huang, Rongtao Xu, Yifei Su, Yonggen Ling, Ian D. Reid, 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 |
|---|---|---|---|
| add direction, object and location constraints (ablation) | CA-Nav · R2R-CE val-unseen | SR: 20.0 -> 25.3 | Improved |
| add object constraint alone to no-constraint baseline (ablation) | CA-Nav · R2R-CE val-unseen | SR: 20.0 -> 23.8 | Improved |
| add direction constraint to object+location constraints (ablation) | CA-Nav · R2R-CE val-unseen | SR: 24.0 -> 25.3 | Improved |
| add direction constraint to location constraint only (ablation) | CA-Nav · R2R-CE val-unseen | SPL: 10.4 -> 7.2 | Hurt |
| add trajectory mask and historical decay for value map update (ablation) | CA-Nav · R2R-CE val-unseen | SR: 22.3 -> 25.3 | Improved |
| add trajectory mask on top of historical decay (ablation) | CA-Nav · R2R-CE val-unseen | SR: 24.6 -> 25.3 | Improved |
| use superpixel-based instead of FBE-based waypoint selection (ablation) | CA-Nav · R2R-CE val-unseen | SR: 21.9 -> 25.3 | Improved |
| use superpixel-based instead of pixel-based waypoint selection (ablation) | CA-Nav · R2R-CE val-unseen | SR: 22.9 -> 25.3 | Improved |
| add min/max constraint switching thresholds (10/25 vs 0/0) (ablation) | CA-Nav · R2R-CE val-unseen | SR: 22.3 -> 25.3 | Improved |
| vary other switching thresholds (5/25, 15/25, 10/15, 10/35) (ablation) | CA-Nav · R2R-CE val-unseen | SR: 24.1, 24.8, 25.0, 24.4 | No clear effect |
| increase superpixel size from 48x48 default to 100x100 (ablation) | CA-Nav · R2R-CE val-unseen | SR: 25.3 -> 21.4 | Hurt |
| decrease superpixel size to 25x25 (ablation) | CA-Nav · R2R-CE val-unseen | SR: 24.9 -> 23.7 | Hurt |
Closest papers by meaning
- Hi-Nav: Hierarchical Framework for Continuous Vision-Language Navigation via Map Guidance and Waypoint Reasoning (ECCV 2026 · similarity 0.89)
- STRIDER: Navigation via Instruction-Aligned Structural Decision Space Optimization (NeurIPS 2025 · similarity 0.89)
- Hierarchical Semantic-Augmented Navigation: Optimal Transport and Graph-Driven Reasoning for Vision-Language Navigation (NeurIPS 2025 · similarity 0.87)
- VoroNav: Voronoi-based Zero-shot Object Navigation with Large Language Model (ICML 2024 · similarity 0.87)
- Towards Long-Horizon Vision-Language Navigation: Platform, Benchmark and Method (CVPR 2025 · similarity 0.87)
- LangMap: A Human-Verified Benchmark for Hierarchical Open-Vocabulary Goal Navigation (NeurIPS 2026 · 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.
