This directory will curate papers on systems that reconstruct, localize, understand, query, update, and act in the physical world. The intended scope includes cross-view geometry, georegistration, 3D and 4D scene representations, open-vocabulary grounding, semantic maps, uncertainty, and embodied planning.
此目录将收录能够在物理世界中完成重建、定位、理解、查询、更新与行动的空间智能论文。重点覆盖跨视角几何、地理配准、3D/4D 场景表征、开放词汇 grounding、语义地图、不确定性与具身规划。
- Treats spatial structure, viewpoint, time, or action as a first-class research problem.
- States the sensing assumptions and evaluates geometric, semantic, temporal, or decision reliability beyond a single image-level score.
- Connects to Geo-grounded 3D Vision-Language Reasoning under Incomplete and Changing Observations or strengthens one of its components.
## Paper Title (Year)
- **Authors:**
- **Venue:**
- **Paper:**
- **Code / Data / Project:**
- **Spatial capability:**
- **Observation assumptions:**
- **Why it matters here:**
- **Missing capability or open question:**See the English README or 中文 README for the current core reading path.