Research map / Vision-language and multimodal / Multimodal LLMs
Let's Reward Step-by-Step: Step-Aware Contrastive Alignment for Vision-Language Navigation in Continuous Environments
Haoyuan Li, Rui Liu, Hehe Fan, Yi Yang
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 Soft Score ranking from PGSA auditor to GRPO (ablation) | SACA · RxR-CE Val-Unseen | SR: 49.9 -> 51.8 | Improved |
| add All-Failure Rescue (ablation) | SACA · RxR-CE Val-Unseen | SR: 51.8 -> 56.4 | Improved |
| add All-Failure Rescue (ablation) | SACA · R2R-CE Val-Unseen | SR: 55.4 -> 58.2 | Improved |
| add Repair Resampling (ablation) | SACA · R2R-CE Val-Unseen | SR: 58.2 -> 60.3 | Improved |
| add Repair Resampling (ablation) | SACA · RxR-CE Val-Unseen | SR: 56.4 -> 60.3 | Improved |
| add outcome-only GRPO over the SFT baseline (ablation) | SFT baseline · RxR-CE Val-Unseen | SR: 48.6 -> 49.9 | Improved |
| add Consistency Align loss on valid prefix (ablation) | SACA · R2R-CE Val-Unseen | SPL: 51.6 -> 55.1 | Improved |
| add Contrastive Correct loss at divergence point (ablation) | SACA · RxR-CE Val-Unseen | SR: 56.4 -> 60.3 | Improved |
| add Margin-Based Rescue (ablation) | SACA · R2R-CE Val-Unseen | SR: 58.5 -> 60.3 | Improved |
| add Negative-Only Scaling (ablation) | SACA · R2R-CE Val-Unseen | SR: 59.1 -> 60.3 | Improved |
| increase group size from 4 to 8 (ablation) | SACA · RxR-CE Val-Unseen | SR: 55.2 -> 60.3 | Improved |
| increase group size from 8 to 16 (ablation) | SACA · R2R-CE Val-Unseen | SR: 60.3 -> 60.1 | No clear effect |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- ETPNav — ETPNav: Evolving Topological Planning for Vision-Language Navigation in Continuous Environments (TPAMI 2024) · better on 2 of 2 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Contrastive Instruction-Trajectory Learning for Vision-Language Navigation (AAAI 2022 · similarity 0.92)
- M$^3$E: Continual Vision-and-Language Navigation via Mixture of Macro and Micro Experts (ICLR 2026 · similarity 0.90)
- Progress-Think: Semantic Progress Reasoning for Vision-Language Navigation (CVPR 2026 · similarity 0.90)
- Bridging the Gap Between Learning in Discrete and Continuous Environments for Vision-and-Language Navigation (CVPR 2022 · similarity 0.89)
- Ground Slow, Move Fast: A Dual-System Foundation Model for Generalizable Vision-Language Navigation (ICLR 2026 · 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.
