---
license: apache-2.0
task_categories:
  - visual-question-answering
  - robotics
language:
  - en
tags:
  - embodied-ai
  - vision-language-navigation
---

# MindCraft

Benchmark for spatio-temporal reasoning during vision-and-language navigation, used with [LASAR](https://github.com/YOUR_ORG/LASAR).

## Layout

```
mindcraft/
  000001/
    data.npz
    1.mp4
  ...
```

## Citation

```bibtex
@inproceedings{lasar2026,
  title={LASAR: Towards Spatio-temporal Reasoning with Latent Cognitive Map},
  booktitle={CVPR},
  year={2026}
}
```