Research map / Multi-agent systems and games / Human-AI and goal-conditioned
Seeing the Un-Scene: Learning Amodal Semantic Maps for Room Navigation
Medhini Narasimhan, Erik Wijmans, Xinlei Chen, Trevor Darrell, Dhruv Batra, Devi Parikh, Amanpreet Singh
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 point prediction and point navigation to map-based room policy (ablation) | Map Generation + Room Navigation Policy · RoomNav validation | RoomNav SPL: 0.16 -> 0.31 | Improved |
| add map generation to point prediction + PointNav (ablation) | Point prediction + PointNav · RoomNav validation | RoomNav SPL: 0.17 -> 0.31 | Improved |
| add map generation to point prediction + PointNav with fine-tuning (ablation) | Point prediction + PointNav + Fine-tune · RoomNav validation | RoomNav SPL: 0.21 -> 0.35 | Improved |
| add fine-tuning of point navigation policy on predicted points (ablation) | Map Generation + Point Prediction + PointNav · RoomNav validation | RoomNav SPL: 0.31 -> 0.35 | Improved |
| add fine-tuning of point navigation policy on predicted points (ablation) | Point prediction + PointNav · RoomNav validation | RoomNav SPL: 0.17 -> 0.21 | Improved |
| add semantic maps and point navigation instead of vanilla RL policy (ablation) | Vanilla Room Navigation Policy · RoomNav validation | RoomNav SPL: 0.10 -> 0.35 | Improved |
| add recurrent memory (LSTM) to map generator (ablation) | CNN mapper · Map generation | mIoU: 25.66 -> 32.92 | Improved |
| feed previous semantic maps as input to LSTM mapper (ablation) | LSTM(no maps) · Map generation | mIoU: 32.92 -> 41.45 | Improved |
| feed previous semantic maps as input to LSTM mapper (ablation) | LSTM(no maps) · Map generation | Average pixel accuracy: 33.59 -> 43.39 | Improved |
| replace predicted maps with ground truth maps | Map Generation + Point Prediction + PointNav · RoomNav validation | RoomNav SPL: 0.31 -> 0.61 | Improved |
| replace predicted maps with ground truth maps in point predictor | Point prediction MLP · Point prediction | Error (%): 39.13 -> 22.91 | Improved |
| replace predicted point with ground truth point | PointNav · RoomNav validation | RoomNav SPL: 0.35 -> 0.83 | Improved |
Closest papers by meaning
- Learning to Map for Active Semantic Goal Navigation (ICLR 2022 · similarity 0.89)
- Distilling LLM Prior to Flow Model for Generalizable Agent’s Imagination in Object Goal Navigation (NeurIPS 2025 · similarity 0.88)
- Hierarchical Semantic-Augmented Navigation: Optimal Transport and Graph-Driven Reasoning for Vision-Language Navigation (NeurIPS 2025 · similarity 0.88)
- LangMap: A Human-Verified Benchmark for Hierarchical Open-Vocabulary Goal Navigation (NeurIPS 2026 · similarity 0.86)
- Imagine Before Go: Self-Supervised Generative Map for Object Goal Navigation (CVPR 2024 · similarity 0.86)
- Scene Map-based Prompt Tuning for Navigation Instruction Generation (CVPR 2025 · similarity 0.85)
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.
