Xinhao Liu*,
Jiaqi Li*,
Youming Deng,
Ruxin Chen,
Yingjia Zhang,
Yifei Ma,
Li Guo,
Yiming Li,
Jing Zhang,
Chen Feng
Caution
Dataset Update In Progress — Known issues found in our dataset. Fixed version expected by early June 2026.
- Simulation Code Release
- [Jan 2026] 3D Reconstruction Benchmark Release
- [Jan 2026] Data Release
Wanderland is a comprehensive framework that consists of many different components. Each component is relatively independent so please refer to the README.md file in each subfolder for more details. Note that all components use uv for dependency management. Please install uv first if you haven't done so.
curl -LsSf https://astral.sh/uv/install.sh | sh
- To download data from Hugging Face, check download
- To benchmark 3D reconstruction methods, check 3d_recon_benchmark
- To generate navigation episodes / videos, check navigation
- To reproduce our reconstruction pipeline, check reconstruction
Each component have its own virtual environment. Remember to deactivate and reactivate the venv when switching between different components.
@inproceedings{liu2026wanderland,
title={Wanderland: Geometrically Grounded Simulation for Open-World Embodied AI},
author={Liu, Xinhao and Li, Jiaqi and Deng, Youming and Chen, Ruxin and Zhang, Yingjia and Ma, Yifei and Guo, Li and Li, Yiming and Zhang, Jing and Feng, Chen},
booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
pages={1041--1052},
year={2026}
}
Real-to-Sim:
- Vid2Sim: Realistic and Interactive Simulation from Video for Urban Navigation, CVPR 2025
- Gauss Gym: A Geometrically Grounded Simulation Environment for Embodied AI, arXiv 2025
- BEHAVIOR-1K: A Human-Centered, Embodied AI Benchmark with 1,000 Everyday Activities and Realistic Simulation, arXiv 2024
3D Reconstruction:
- VGGT: Visual Geometry Grounded Transformer, CVPR 2025
- π3 : Permutation-Equivariant Visual Geometry Learning, arXiv 2025
- MapAnything: Universal Feed-Forward Metric 3D Reconstruction, arXiv 2025
- Depth Anything 3: Recovering the Visual Space from Any Views, arXiv 2025
Navigation:
- DeepExplorer: Metric-Free Exploration for Topological Mapping by Task and Motion Imitation in Feature Space, RSS 2023
- CityWalker: Learning Embodied Urban Navigation from Web-Scale Videos, CVPR 2025
- NaVILA: Legged Robot Vision-Language-Action Model for Navigation, RSS 2025
- Learning to Drive Anywhere with Model-Based Reannotation, RA-L 2025
- From Seeing to Experiencing: Scaling Navigation Foundation Models with Reinforcement Learning, arXiv 2025