---
title: "DeepMind Lab — paper by Beattie et al. (2016)"
description: "The authors present DeepMind Lab, a first-person 3D platform built on the ioquake3 version of the… 18 references, cited by 66 papers here."
authors: ["Charles Beattie", "Joel Z. Leibo", "Denis Teplyashin", "Tom Ward", "Marcus Wainwright", "Heinrich Kuttler", "Andrew Lefrancq", "Simon Green", "Victor Valdes", "Amir Sadik", "Julian Schrittwieser", "Keith Anderson", "Sarah York", "Max Cant", "Adam Cain", "Adrian Bolton", "Stephen Gaffney", "Helen King", "Demis Hassabis", "Shane Legg", "Stig Petersen"]
year: 2016
venue: "arXiv:1612.03801"
source: "https://arxiv.org/abs/1612.03801"
doi: "10.48550/arxiv.1612.03801"
arxiv: "1612.03801"
licence: "arXiv — non-exclusive license to distribute"
licence_url: "http://arxiv.org/licenses/nonexclusive-distrib/1.0/"
summary_licence: "Elamur Cosmic License"
canonical: "https://research.elamur.ai/free/en/papers/beattie-2016-deepmind-lab/"
---
[Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/papers/beattie-2016-deepmind-lab/)

[home](https://md.research.elamur.ai/free/en/) / [papers](https://md.research.elamur.ai/free/en/papers/) / [2016](https://md.research.elamur.ai/free/en/papers/2016/page/3/) / DeepMind Lab

# DeepMind Lab

**Authors:** [Charles Beattie](https://md.research.elamur.ai/free/en/authors/charles-beattie/), [Joel Z. Leibo](https://md.research.elamur.ai/free/en/authors/joel-z-leibo/), [Denis Teplyashin](https://md.research.elamur.ai/free/en/authors/denis-teplyashin/), Tom Ward, [Marcus Wainwright](https://md.research.elamur.ai/free/en/authors/marcus-wainwright/), [Heinrich Kuttler](https://md.research.elamur.ai/free/en/authors/heinrich-kuttler/), Andrew Lefrancq, [Simon Green](https://md.research.elamur.ai/free/en/authors/simon-green/), Victor Valdes, Amir Sadik, [Julian Schrittwieser](https://md.research.elamur.ai/free/en/authors/julian-schrittwieser/), Keith Anderson, Sarah York, Max Cant, Adam Cain, Adrian Bolton, Stephen Gaffney, Helen King, Demis Hassabis, [Shane Legg](https://md.research.elamur.ai/free/en/authors/shane-legg/) and Stig Petersen · **Year:** 2016 · **Venue:** arXiv:1612.03801 · **Licence:** [arXiv — non-exclusive license to distribute](http://arxiv.org/licenses/nonexclusive-distrib/1.0/)

**Links:** [Publisher's page](https://arxiv.org/abs/1612.03801) · [PDF](https://arxiv.org/pdf/1612.03801) · [DOI 10.48550/arxiv.1612.03801](https://doi.org/10.48550/arxiv.1612.03801) · [arXiv 1612.03801](https://arxiv.org/abs/1612.03801) · [OpenAlex](https://openalex.org/W4394652010)

**Citations:** 68 citations (OpenAlex, 2026-10-14) · [66](https://md.research.elamur.ai/free/en/papers/beattie-2016-deepmind-lab/cited-by/) in our library, of the 149,439 papers whose bibliographies we read

## Summary
Studying embodied intelligence with physical robots is slow and expensive, and building detailed virtual worlds for AI research has required heavy engineering. The authors present DeepMind Lab, a first-person 3D platform built on the ioquake3 version of the Quake III Arena engine, with custom sci-fi assets designed for machine learning. Interaction is lock-stepped: the simulation pauses after each observation until the agent chooses its next action. Agents receive a scalar reward, RGB or RGBD frames, and optional velocity vectors. They can move, jump, crouch, look along two axes, and tag opponents. New tasks can be defined with a Lua level API or with simple Text Levels. Users control the engine through a C API with Python bindings on Linux. Bundled levels cover fruit gathering, fixed and procedurally generated mazes, and laser tag against built-in bots. With GPU rendering, small-resolution simulation approaches 1000 frames per second. The platform matters because it combines pixel-based vision, navigation, memory and fine motor control in one testbed. It already supported DeepMind research on asynchronous RL, auxiliary tasks and navigation.

## Key points
- The platform builds on ioquake3, adds q3map2 and bspc for level generation, and adapts bot scripts from OpenArena. It is compared with the Arcade Learning Environment, VizDoom and Minecraft.
- There are four level families:
  - fruit gathering (seekavoid_arena_01, stairway_to_melon)
  - static mazes (nav_maze_static_01–03, nav_maze_random_goal_01–03)
  - random_maze, with three-minute episodes
  - laser tag (lt_horseshoe_color, lt_chasm, lt_hallway_slope, lt_space_bounce_hard)
- In lt_horseshoe_color and lt_space_bounce_hard, bot colours and textures are randomised each episode, so agents cannot rely on colour to spot bots.
- In the API example, observations are 240×320×3 RGB_INTERLACED arrays and actions are a 7-element integer vector. Each action can be repeated over several frames.
- Benchmarks ran on a 6-core 3.5 GHz Xeon with a Quadro K600 GPU. On nav_maze_static_01 with RGB output:
  - GPU: 996.6 fps at 84×84 and 950.0 fps at 320×240
  - CPU: 199.7 fps at 84×84 and 27.3 fps at 320×240

*Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.*

📚 Shelves (2): [🖥️ Computer Science](https://md.research.elamur.ai/free/en/shelves/computer-science/) · 18,614 works · [🤖 Artificial Intelligence](https://md.research.elamur.ai/free/en/shelves/computer-science/artificial-intelligence/) · 17,965 works

## Cite

Cite as: Charles Beattie, Joel Z. Leibo, Denis Teplyashin, Tom Ward, Marcus Wainwright, Heinrich Kuttler, Andrew Lefrancq, Simon Green, Victor Valdes, Amir Sadik, Julian Schrittwieser, Keith Anderson, Sarah York, Max Cant, Adam Cain, Adrian Bolton, Stephen Gaffney, Helen King, Demis Hassabis, Shane Legg and Stig Petersen. 2016. DeepMind Lab. arXiv:1612.03801.

The paper itself, not this page. BibTeX from the library's record of the work.

```bibtex
@misc{beattie2016deepmind,
  title = {{DeepMind Lab}},
  author = {Charles Beattie, Joel Z. Leibo, Denis Teplyashin, Tom Ward, Marcus Wainwright, Heinrich Kuttler, Andrew Lefrancq, Simon Green, Victor Valdes, Amir Sadik, Julian Schrittwieser, Keith Anderson, Sarah York, Max Cant, Adam Cain, Adrian Bolton, Stephen Gaffney, Helen King, Demis Hassabis, Shane Legg and Stig Petersen},
  year = {2016},
  doi = {10.48550/arxiv.1612.03801},
  eprint = {1612.03801},
  archivePrefix = {arXiv},
  url = {https://doi.org/10.48550/arxiv.1612.03801}
}
```

- [← previous SemEval-2016 Task 7: Determining Sentiment Intensity of English and Arabic Phrases](https://md.research.elamur.ai/free/en/papers/kiritchenko-2016-semeval-2016-task-7-determining/)
- [next → Dialogue Act Classification in Domain-Independent Conversations Using a Deep Recurrent Neural Network](https://md.research.elamur.ai/free/en/papers/khanpour-2016-dialogue-act-classification-domain-independent/)
- [↑ back to papers of 2016](https://md.research.elamur.ai/free/en/papers/2016/page/3/)

## References

18 references, 2 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. Marc G Bellemare, Yavar Naddaf, Joel Veness, and Michael Bowling. [The arcade learning environment: An evaluation platform for general agents](https://md.research.elamur.ai/free/en/papers/bellemare-2012-arcade-learning-environment-evaluation-platform/). Journal of Artificial Intelligence Research, 2012.
2. Rodney A Brooks. Elephants don’t play chess. Robotics and autonomous systems, 6 (1):3–15, 1990. bspc. bspc, 2016. URL [https://github.com/TTimo/bspc](https://github.com/TTimo/bspc). [↗](https://github.com/TTimo/bspc)
3. GtkRadiant. Gtkradiant, 2016. URL [http://icculus.org/gtkradiant/](http://icculus.org/gtkradiant/). [↗](http://icculus.org/gtkradiant/)
4. David Hume. Treatise on human nature. 1739. id software. Quake3, Quake-III-Arena. Max Jaderberg, Volodymyr Mnih, Wojciech Marian Czarnecki, Tom Schaul, Joel Z Leibo, David Silver, and Koray Kavukcuoglu. [Reinforcement learning with unsupervised auxiliary tasks](https://md.research.elamur.ai/free/en/papers/jaderberg-2016-reinforcement-learning-unsupervised-auxiliary-tasks/). arXiv preprint arXiv:1611.05397, 2016.
5. Matthew Johnson, Katja Hofmann, Tim Hutton, and David Bignell. The malmo platform for artificial intelligence experimentation. In International joint conference on artificial intelligence (IJCAI), 2016. [find](https://scholar.google.com/scholar?q=Matthew+Johnson%2C+Katja+Hofmann%2C+Tim+Hutton%2C+and+David+Bignell%2E+The+malmo+platform+for+artificial+intelligence+experimentation%2E+In+International+joint+conference+on+artificial+intelligence+%28IJCAI%29%2C+201)
6. Michał Kempka, Marek Wydmuch, Grzegorz Runc, Jakub Toczek, and Wojciech Jaśkowski. Vizdoom: A doom-based ai research platform for visual reinforcement learning. arXiv preprint arXiv:1605.02097, 2016. — [arXiv:1605.02097](https://arxiv.org/abs/1605.02097) [find](https://scholar.google.com/scholar?q=Micha%C5%82+Kempka%2C+Marek+Wydmuch%2C+Grzegorz+Runc%2C+Jakub+Toczek%2C+and+Wojciech+Ja%C5%9Bkowski%2E+Vizdoom%3A+A+doom%2Dbased+ai+research+platform+for+visual+reinforcement+learning%2E+arXiv+preprint+arXiv%3A1605%2E02097%2C+2016%2E+)
7. Shane Legg and Marcus Hutter. Universal intelligence: A definition of machine intelligence. Minds and Machines, 17(4):391–444, 2007. [find](https://scholar.google.com/scholar?q=Shane+Legg+and+Marcus+Hutter%2E+Universal+intelligence%3A+A+definition+of+machine+intelligence%2E+Minds+and+Machines%2C+17%284%29%3A391%E2%80%93444%2C+2007%2E)
8. Adam Lerer, Sam Gross, and Rob Fergus. Learning physical intuition of block towers by example. arXiv preprint arXiv:1603.01312, 2016. [find](https://scholar.google.com/scholar?q=Adam+Lerer%2C+Sam+Gross%2C+and+Rob+Fergus%2E+Learning+physical+intuition+of+block+towers+by+example%2E+arXiv+preprint+arXiv%3A1603%2E01312%2C+2016%2E)
9. John Locke. An essay concerning human understanding. 1690. A Mahendran, H Bilen, JF Henriques, and A Vedaldi. Researchdoom and cocodoom: Learning computer vision with games. arXiv preprint arXiv:1610.02431, 2016. — [arXiv:1610.02431](https://arxiv.org/abs/1610.02431) [find](https://scholar.google.com/scholar?q=John+Locke%2E+An+essay+concerning+human+understanding%2E+1690%2E+A+Mahendran%2C+H+Bilen%2C+JF+Henriques%2C+and+A+Vedaldi%2E+Researchdoom+and+cocodoom%3A+Learning+computer+vision+with+games%2E+arXiv+preprint+arXiv%3A1610%2E)
10. Giorgio Metta, Giulio Sandini, David Vernon, Lorenzo Natale, and Francesco Nori. The icub humanoid robot: an open platform for research in embodied cognition. In Proceedings of the 8th workshop on performance metrics for intelligent systems, pages 50–56. ACM, 2008. [find](https://scholar.google.com/scholar?q=Giorgio+Metta%2C+Giulio+Sandini%2C+David+Vernon%2C+Lorenzo+Natale%2C+and+Francesco+Nori%2E+The+icub+humanoid+robot%3A+an+open+platform+for+research+in+embodied+cognition%2E+In+Proceedings+of+the+8th+workshop+on+per)
11. Piotr Mirowski, Razvan Pascanu, Fabio Viola, Hubert Soyer, Andy Ballard, Andrea Banino, Misha Denil, Ross Goroshin, Laurent Sifre, Koray Kavukcuoglu, et al. Learning to navigate in complex environments. arXiv preprint arXiv:1611.03673, 2016. [find](https://scholar.google.com/scholar?q=Piotr+Mirowski%2C+Razvan+Pascanu%2C+Fabio+Viola%2C+Hubert+Soyer%2C+Andy+Ballard%2C+Andrea+Banino%2C+Misha+Denil%2C+Ross+Goroshin%2C+Laurent+Sifre%2C+Koray+Kavukcuoglu%2C+et+al%2E+Learning+to+navigate+in+complex+environment)
12. Volodymyr Mnih, Adria Puigdomenech Badia, Mehdi Mirza, Alex Graves, Timothy P Lillicrap, Tim Harley, David Silver, and Koray Kavukcuoglu. Asynchronous methods for deep reinforcement learning. arXiv preprint arXiv:1602.01783, 2016. 1999. URL [https://github.com/id-Software/](https://github.com/id-Software/) [↗](https://github.com/id-Software/)
13. Ludwig Nussel, Thilo Schulz, Tim Angus, Tony J White, and Zachary J Slater. ioquake3, 2016. URL [https://github.com/ioquake/ioq3](https://github.com/ioquake/ioq3). [↗](https://github.com/ioquake/ioq3)
14. OpenArena. The openarena project, 2016. URL [http://www.openarena.ws](http://www.openarena.ws). [↗](http://www.openarena.ws)
15. Weichao Qiu and Alan Yuille. Unrealcv: Connecting computer vision to unreal engine. arXiv preprint arXiv:1609.01326, 2016. — [arXiv:1609.01326](https://arxiv.org/abs/1609.01326) [find](https://scholar.google.com/scholar?q=Weichao+Qiu+and+Alan+Yuille%2E+Unrealcv%3A+Connecting+computer+vision+to+unreal+engine%2E+arXiv+preprint+arXiv%3A1609%2E01326%2C+2016%2E+%E2%80%94+%5BarXiv%3A1609%2E01326%5D%28https%3A%2F%2Farxiv%2Eorg%2Fabs%2F1609%2E01326%29)
16. Stephan R Richter, Vibhav Vineet, Stefan Roth, and Vladlen Koltun. Playing for data: Ground truth from computer games. In European Conference on Computer Vision, pages 102–118. Springer, 2016. [find](https://scholar.google.com/scholar?q=Stephan+R+Richter%2C+Vibhav+Vineet%2C+Stefan+Roth%2C+and+Vladlen+Koltun%2E+Playing+for+data%3A+Ground+truth+from+computer+games%2E+In+European+Conference+on+Computer+Vision%2C+pages+102%E2%80%93118%2E+Springer%2C+2016%2E)
17. Chen Tessler, Shahar Givony, Tom Zahavy, Daniel J Mankowitz, and Shie Mannor. A deep hierarchical approach to lifelong learning in minecraft. arXiv preprint arXiv:1604.07255, 2016. [find](https://scholar.google.com/scholar?q=Chen+Tessler%2C+Shahar+Givony%2C+Tom+Zahavy%2C+Daniel+J+Mankowitz%2C+and+Shie+Mannor%2E+A+deep+hierarchical+approach+to+lifelong+learning+in+minecraft%2E+arXiv+preprint+arXiv%3A1604%2E07255%2C+2016%2E)
18. Figure 7: Example images from the agent’s egocentric viewpoint from several example DeepMind Lab levels. Figure 8: Example images from the agent’s egocentric viewpoint from several example DeepMind Lab levels. [find](https://scholar.google.com/scholar?q=Figure+7%3A+Example+images+from+the+agent%E2%80%99s+egocentric+viewpoint+from+several+example+DeepMind+Lab+levels%2E+Figure+8%3A+Example+images+from+the+agent%E2%80%99s+egocentric+viewpoint+from+several+example+DeepMind+La)

*Updated 2 November 2026*