Research map / Vision-language and multimodal / Robot action and navigation
GridMM: Grid Memory Map for Vision-and-Language Navigation
Zihan Wang, Xiangyang Li, Jiahao Yang, Yeqi Liu, Shuqiang Jiang
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 any map (topological, semantic, object, grid) over no map (ablation) | GridMM base model · R2R-CE val unseen | SR: 45.19 -> 47.02 | Improved |
| use grid features instead of top-down semantic map features (ablation) | GridMM · R2R-CE val unseen | SR: 46.36 -> 49.05 | Improved |
| use CLIP grid features instead of detected object features (ablation) | GridMM · R2R-CE val unseen | SR: 47.61 -> 49.05 | Improved |
| use egocentric relative coordinates instead of absolute coordinates (ablation) | GridMM · R2R-CE val unseen | SR: 48.72 -> 49.05 | Improved |
| add navigation trajectory memory (ablation) | GridMM · R2R-CE val unseen | SR: 47.42 -> 49.05 | Improved |
| use instruction relevance aggregation instead of average pooling (ablation) | GridMM · R2R-CE val unseen | SR: 48.63 -> 49.05 | Improved |
| increase map scale from 14x14 to 20x20 (ablation) | GridMM · R2R-CE val unseen | SR: 49.05 -> 49.86 | Mixed |
| decrease map scale from 14x14 to 8x8 (ablation) | GridMM · R2R-CE val unseen | SR: 49.05 -> 47.07 | Hurt |
| GridMM versus DUET on REVERIE val unseen | GridMM · REVERIE val unseen | SR: 46.98 -> 51.37 | Improved |
| GridMM versus BEVBert on REVERIE val unseen | GridMM · REVERIE val unseen | SR: 51.78 vs 51.37 | No clear effect |
| GridMM versus DUET on SOON test unseen | GridMM · SOON test unseen | SPL: 21.42 -> 21.25 | No clear effect |
Later papers that compared with this method
5 later accepted papers report GridMM as a baseline in their results tables; 4 of them report a better number on most of the results they share with it.
- Run, Ruminate, and Regulate: A Dual-process Thinking System for Vision-and-Language Navigation (AAAI 2026 · better on 4 of 4 shared results)
- PanoWorld: Towards Spatial Supersensing in 360◦ Panorama World (NeurIPS 2026 · better on 1 of 1 shared result)
- ESceme: Vision-and-Language Navigation with Episodic Scene Memory (IJCV 2024 · better on 1 of 3 shared results)
- Vision-and-Language Navigation via Causal Learning (CVPR 2024 · better on 2 of 2 shared results)
- PRET: Planning with Directed Fidelity Trajectory for Vision and Language Navigation (ECCV 2024 · better on 1 of 1 shared result)
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.
- DUET — DUET: Cross-Modal Semantic Grounding for Contrastive Zero-Shot Learning (AAAI 2023) · better on 9 of 10 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
- Structured Scene Memory for Vision-Language Navigation (CVPR 2021 · similarity 0.91)
- Towards Long-Horizon Vision-Language Navigation: Platform, Benchmark and Method (CVPR 2025 · similarity 0.89)
- Scene Map-based Prompt Tuning for Navigation Instruction Generation (CVPR 2025 · similarity 0.89)
- MemoNav: Working Memory Model for Visual Navigation (CVPR 2024 · similarity 0.89)
- LangMap: A Human-Verified Benchmark for Hierarchical Open-Vocabulary Goal Navigation (NeurIPS 2026 · similarity 0.88)
- Hierarchical Semantic-Augmented Navigation: Optimal Transport and Graph-Driven Reasoning for Vision-Language Navigation (NeurIPS 2025 · similarity 0.88)
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.
