Research map / 3D vision / 3D generation
UrbanVerse: Scaling Urban Simulation by Watching City-Tour Videos
Mingxuan Liu, Honglin He, Elisa Ricci, Wayne Wu, Bolei Zhou
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 |
|---|---|---|---|
| train on UrbanVerse scenes instead of UrbanSim procedurally generated scenes (160 scenes) | PPO navigation policy · CraftBench (sim) | success rate: 9.1 -> 41.9 | Improved |
| train PPO in UrbanVerse vs navigation foundation model MBRA | PPO-UrbanVerse · CraftBench (sim) | success rate: 35.6 -> 41.9 | Improved |
| train PPO in UrbanVerse vs MBRA, collision times | PPO-UrbanVerse · CraftBench (sim) | collision times: 25.6 -> 35.5 | Hurt |
| train directly on test scenes (overfitting reference) instead of diverse UrbanVerse scenes | PPO policy · CraftBench altered routes | success rate: 41.9 -> 26.5 | Hurt |
| increase number of unique UrbanVerse layouts (1 to 32) for training (ablation) | PPO policy · CraftBench | success rate | Improved |
| increase number of unique layouts in procedurally generated UrbanSim scenes (ablation) | PPO policy · CraftBench | success rate | No clear effect |
| increase digital cousins per layout from 1 to 5 (ablation) | PPO policy · AutoBench and CraftBench | success rate | Improved |
| use MASt3R + YoloWorld/SAM2 parser instead of VGGT or GroundedSAM2 (ablation) | UrbanVerse-Gen · KITTI-360 | mAP25: VGGT+GroundedSAM2 7.5 -> MASt3R+YoWorldSAM2 28.2 | Improved |
| zero-shot sim-to-real: UrbanVerse-trained policy vs S2E | PPO-UrbanVerse · 16 real-world routes (Coco/Go2) | success rate: +29.2% (Coco), +31.1% (Go2) | Improved |
Closest papers by meaning
- CityWalker: Learning Embodied Urban Navigation from Web-Scale Videos (CVPR 2025 · similarity 0.87)
- Synthesizing Photorealistic and Dynamic Urban Environments for Multimodal Robot Navigation and Collaboration (NeurIPS 2025 · similarity 0.87)
- Vid2Sim: Realistic and Interactive Simulation from Video for Urban Navigation (CVPR 2025 · similarity 0.87)
- SceneDiffuser++: City-Scale Traffic Simulation via a Generative World Model (CVPR 2025 · similarity 0.87)
- Coarse-to-Real: Generative Rendering for Populated Dynamic Scenes (NeurIPS 2026 · similarity 0.85)
- Towards Autonomous Micromobility through Scalable Urban Simulation (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.
