# Teaching a Machine to Read Maps with Deep Reinforcement Learning

arXiv: https://arxiv.org/abs/1711.07479 · HTML: https://mlsanity.com/paper/1711.07479/

## Key facts

- Cited by 9 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 20 Nov 2017.

## Built on by (most cited first, 9 of 9)

- [Hierarchical Robot Navigation in Novel Environments using Rough 2-D Maps](https://mlsanity.com/paper/2106.03665/) (2021)
- [ROS-X-Habitat: Bridging the ROS Ecosystem with Embodied AI](https://mlsanity.com/paper/2109.07703/) (2021)
- [A Framework for Incentivized Collaborative Learning](https://mlsanity.com/paper/2305.17052/) (2023)
- [Learning to Navigate in Mazes with Novel Layouts using Abstract Top-down Maps](https://mlsanity.com/paper/2412.12024/) (2024)
- [Assisted Learning for Organizations with Limited Imbalanced Data](https://mlsanity.com/paper/2109.09307/) (2021)
- [Towards self-attention based visual navigation in the real world](https://mlsanity.com/paper/2209.07043/) (2022)
- [Estimating Reaction Barriers with Deep Reinforcement Learning](https://mlsanity.com/paper/2407.12453/) (2024)
- [Confidence-Controlled Exploration: Efficient Sparse-Reward Policy Learning for Robot Navigation](https://mlsanity.com/paper/2306.06192/) (2023)
- [Constructing Ancestral Recombination Graphs through Reinforcement Learning](https://mlsanity.com/paper/2406.12022/) (2024)

## Abstract (arXiv)

The ability to use a 2D map to navigate a complex 3D environment is quite remarkable, and even difficult for many humans. Localization and navigation is also an important problem in domains such as robotics, and has recently become a focus of the deep reinforcement learning community. In this paper we teach a reinforcement learning agent to read a map in order to find the shortest way out of a random maze it has never seen before. Our system combines several state-of-the-art methods such as A3C and incorporates novel elements such as a recurrent localization cell. Our agent learns to localize itself based on 3D first person images and an approximate orientation angle. The agent generalizes well to bigger mazes, showing that it learned useful localization and navigation capabilities.