Target-driven Visual Navigation Model using Deep Reinforcement Learning

This is implementation of http://web.stanford.edu/~yukez/papers/icra2017.pdf in PyTorch. It attempts to achieve the same results as the Tensorflow implementation, which can be found here: https://github.com/zfw1226/icra2017-visual-navigation.

Introduction

This repocitory provides a Tensorflow implementation of the deep siamese actor-critic model for indoor scene navigation introduced in the following paper:

Target-driven Visual Navigation in Indoor Scenes using Deep Reinforcement Learning
Yuke Zhu, Roozbeh Mottaghi, Eric Kolve, Joseph J. Lim, Abhinav Gupta, Li Fei-Fei, and Ali Farhadi
ICRA 2017, Singapore

Setup and run

This code is implemented in Pytorch 0.4. It uses Docker to automate instalation process. In order to run this code, I recommend pulling it from my dockerhub repository.

In order to start training, run those commands:

git clone https://github.com/jkulhanek/visual-navigation-agent-pytorch
docker-compose run train

Scenes

To facilitate training, we provide hdf5 dumps of the simulated scenes. Each dump contains the agent's first-person observations sampled from a discrete grid in four cardinal directions. To be more specific, each dump stores the following information row by row:

• observation: 300x400x3 RGB image (agent's first-person view)
• resnet_feature: 2048-d ResNet-50 feature extracted from the observations
• location: (x, y) coordinates of the sampled scene locations on a discrete grid with 0.5-meter offset
• rotation: agent's rotation in one of the four cardinal directions, 0, 90, 180, and 270 degrees
• graph: a state-action transition graph, where graph[i][j] is the location id of the destination by taking action j in location i, and -1 indicates collision while the agent stays in the same place.
• shortest_path_distance: a square matrix of shortest path distance (in number of steps) between pairwise locations, where -1 means two states are unreachable from each other.

  Acknowledgements

  I would like to acknowledge the following references that have offered great help for me to implement the model.

• "Asynchronous Methods for Deep Reinforcement Learning", Mnih et al., 2016
• David Silver's Deep RL course
• muupan's async-rl repo
• miyosuda's async_deep_reinforce repo
• miyosuda's async_deep_reinforce repo
• Pytorch A3C implementation repo

Citation

Please cite our ICRA'17 paper if you find this code useful for your research.

@InProceedings{zhu2017icra,
  title = {{Target-driven Visual Navigation in Indoor Scenes using Deep Reinforcement Learning}},
  author = {Yuke Zhu and Roozbeh Mottaghi and Eric Kolve and Joseph J. Lim and Abhinav Gupta and Li Fei-Fei and Ali Farhadi},
  booktitle = {{IEEE International Conference on Robotics and Automation}},
  year = 2017,
}

License

MIT