# NeurIPS 2026 Paper Urbex: Agent-Driven City-Scale 3D Spatial Localization

- Source: CVer · 微信公众号
- Published: 2026-10-21T06:46:35.000Z
- RoboHot: https://robohot.news/en/items/ypv7aqvj4miihem1xoifqxofs
- Original: https://mp.weixin.qq.com/s?__biz=MzUxNjcxMjQxNg==&mid=2247643735&idx=3&sn=f385014cf12f6f78ad2774550a00ccf8&chksm=f80d40bc85b7b09d41a9b44612b09db23c5bffa5aae73bab26dc5c01e21b66fb562e37313a77&scene=126&sessionid=1792593048#rd

## Summary

Urbex is an agentic vision-language framework for city-scale 3D scenes that shifts from static ranking based on pre-generated candidate boxes to active spatial search, progressively narrowing the search area through three tools—landmark query, local zoom-in, and oblique view—and uses GRPO reinforcement learning to optimize tool-invocation strategies.