No description, website, or topics provided.
Switch branches/tags
Nothing to show
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Failed to load latest commit information.
models Initial commit Jun 19, 2018
utils Fix print_policy() Jun 20, 2018
.gitignore Initial commit Jun 19, 2018
README.md Update README.md Jun 25, 2018
download_datasets.sh Initial commit Jun 19, 2018
eval.py Initial commit Jun 19, 2018
generate_dataset.py Initial commit Jun 19, 2018
requirements.txt Initial commit Jun 19, 2018
train.py

README.md

Gated Path Planning Networks (ICML 2018)

This is the official codebase for the following paper:

Lisa Lee*, Emilio Parisotto*, Devendra Singh Chaplot, Eric Xing, Ruslan Salakhutdinov. Gated Path Planning Networks. ICML 2018. https://arxiv.org/abs/1806.06408

Getting Started

You can clone this repo by running:

git clone https://github.com/lileee/gated-path-planning-networks.git
cd gated-path-planning-networks/

All subsequent commands in this README should be run from the top-level directory of this repository (i.e., /path/to/gated-path-planning-networks/).

I. Docker container

We provide two Docker containers, with and without GPU support. These containers have Python 3.6.5, PyTorch 0.4.0, and other dependencies installed. They do not contain this codebase or the maze datasets used in our experiments.

To load the container with GPU support:

# PyTorch with GPU support
nvidia-docker pull lileee/ubuntu-16.04-pytorch-0.4.0-gpu:v1
nvidia-docker run -v $(pwd):/home --rm -ti lileee/ubuntu-16.04-pytorch-0.4.0-gpu:v1

To load the container without GPU support:

# PyTorch (CPU-only)
nvidia-docker pull lileee/ubuntu-16.04-pytorch-0.4.0-cpu:v1
nvidia-docker run -v $(pwd):/home --rm -ti lileee/ubuntu-16.04-pytorch-0.4.0-cpu:v1

Here is a speed comparison between the Docker containers for training VIN on a 9x9 maze with 5k/1k/1k train-val-test split:

PyTorch 0.4.0 time per epoch
with GPU support 8.5 sec
without GPU support 32.3 sec

II. Generate a 2D maze dataset

Generate a dataset by running:

python generate_dataset.py --output-path mazes.npz --mechanism news --maze-size 9 --train-size 5000 --valid-size 1000 --test-size 1000

This will create a datafile mazes.npz containing a dataset of 9x9 mazes using the NEWS maze transition mechanism with 5k/1k/1k train-val-test split.

Note:

  • The same maze transition mechanism that was used to generate the dataset must be used for train.py and eval.py. Here, we used --mechanism news to generate the dataset. Other options are --mechanism moore and --mechanism diffdrive.

III. Train a model

You can train a VIN with iteration count K=15 and kernel size F=5 on the datafile mazes.npz by running:

python train.py --datafile mazes.npz --mechanism news --model models.VIN --k 15 --f 5 --save-directory log/vin-k15-f5

This will save outputs to the subdirectory vin-k15-f5/, including the trained models and learning plots.

Similarly, you can train a GPPN by running:

python train.py --datafile mazes.npz --mechanism news --model models.GPPN --k 15 --f 5 --save-directory log/gppn-k15-f5

Notes:

  • --mechanism must be the same as the one used to generate mazes.npz (which is news in this example).
  • --f must be an odd integer.

IV. Evaluate a trained model

Once you have a trained VIN model, you can evaluate it on a dataset by running:

python eval.py --datafile mazes.npz --mechanism news --model models.VIN --k 15 --f 5 --load-file log/vin-k15-f5/planner.final.pth

Similarly for GPPN:

python eval.py --datafile mazes.npz --mechanism news --model models.GPPN --k 15 --f 5 --load-file log/gppn-k15-f5/planner.final.pth

Notes:

  • --mechanism must be the same as the one used to generate mazes.npz (which is news in this example).
  • --f must be the same as the one used to train the model.

Replicating experiments from our paper

I. Download 2D maze datasets

To replicate experiments from our ICML 2018 paper, first download the datasets by running:

./download_datasets.sh

This will create a subdirectory mazes/ containing the following 2D maze datasets used in our experiments:

datafile maze size mechanism train size val size test size
m15_news_10k.npz 15x15 news 10000 2000 2000
m15_news_25k.npz 15x15 news 25000 5000 5000
m15_news_100k.npz 15x15 news 100000 10000 10000
m15_moore_10k.npz 15x15 moore 10000 2000 2000
m15_moore_25k.npz 15x15 moore 25000 5000 5000
m15_moore_100k.npz 15x15 moore 100000 10000 10000
m15_diffdrive_10k.npz 15x15 diffdrive 10000 2000 2000
m15_diffdrive_25k.npz 15x15 diffdrive 25000 5000 5000
m15_diffdrive_100k.npz 15x15 diffdrive 100000 10000 10000
m28_news_25k.npz 28x28 news 25000 5000 5000
m28_moore_25k.npz 28x28 moore 25000 5000 5000
m28_diffdrive_25k.npz 28x28 diffdrive 25000 5000 5000

II. Train VIN and GPPN

Then you can train VIN with the best (K, F) settings for each dataset from our paper by running:

python train.py --datafile mazes/m15_news_10k.npz --mechanism news --model models.VIN --k 30 --f 5
python train.py --datafile mazes/m15_news_25k.npz --mechanism news --model models.VIN --k 20 --f 5
python train.py --datafile mazes/m15_news_100k.npz --mechanism news --model models.VIN --k 30 --f 3

python train.py --datafile mazes/m15_moore_10k.npz --mechanism moore --model models.VIN --k 30 --f 11
python train.py --datafile mazes/m15_moore_25k.npz --mechanism moore --model models.VIN --k 30 --f 5
python train.py --datafile mazes/m15_moore_100k.npz --mechanism moore --model models.VIN --k 30 --f 5

python train.py --datafile mazes/m15_diffdrive_10k.npz --mechanism diffdrive --model models.VIN --k 30 --f 3
python train.py --datafile mazes/m15_diffdrive_25k.npz --mechanism diffdrive --model models.VIN --k 30 --f 3
python train.py --datafile mazes/m15_diffdrive_100k.npz --mechanism diffdrive --model models.VIN --k 30 --f 3

python train.py --datafile mazes/m28_news_25k.npz --mechanism news --model models.VIN --k 56 --f 3
python train.py --datafile mazes/m28_moore_25k.npz --mechanism moore --model models.VIN --k 56 --f 5
python train.py --datafile mazes/m28_diffdrive_25k.npz --mechanism diffdrive --model models.VIN --k 56 --f 3

Similarly, you can train GPPN with the best (K, F) settings for each dataset from our paper by running:

python train.py --datafile mazes/m15_news_10k.npz --mechanism news --model models.GPPN --k 20 --f 9
python train.py --datafile mazes/m15_news_25k.npz --mechanism news --model models.GPPN --k 20 --f 11
python train.py --datafile mazes/m15_news_100k.npz --mechanism news --model models.GPPN --k 30 --f 11

python train.py --datafile mazes/m15_moore_10k.npz --mechanism moore --model models.GPPN --k 30 --f 7
python train.py --datafile mazes/m15_moore_25k.npz --mechanism moore --model models.GPPN --k 30 --f 9
python train.py --datafile mazes/m15_moore_100k.npz --mechanism moore --model models.GPPN --k 30 --f 7

python train.py --datafile mazes/m15_diffdrive_10k.npz --mechanism diffdrive --model models.GPPN --k 30 --f 11
python train.py --datafile mazes/m15_diffdrive_25k.npz --mechanism diffdrive --model models.GPPN --k 30 --f 9
python train.py --datafile mazes/m15_diffdrive_100k.npz --mechanism diffdrive --model models.GPPN --k 30 --f 9

python train.py --datafile mazes/m28_news_25k.npz --mechanism news --model models.GPPN --k 56 --f 11
python train.py --datafile mazes/m28_moore_25k.npz --mechanism moore --model models.GPPN --k 56 --f 9
python train.py --datafile mazes/m28_diffdrive_25k.npz --mechanism diffdrive --model models.GPPN --k 56 --f 11

III. Test Performance Results

Here are the test performance results from running the above commands inside the Docker container lileee/ubuntu-16.04-pytorch-0.4.0-gpu:v1:

VIN GPPN
datafile K F %Opt %Suc K F %Opt %Suc
m15_news_10k.npz 30577.479.0 20996.897.8
m15_news_25k.npz 20583.684.2 201199.099.3
m15_news_100k.npz 30392.692.8 301199.799.8
m15_moore_10k.npz 301186.089.3 30797.098.0
m15_moore_25k.npz 30585.488.1 30998.999.5
m15_moore_100k.npz 30596.997.5 30799.699.8
m15_diffdrive_10k.npz 30398.499.0 301199.199.7
m15_diffdrive_25k.npz 30396.198.5 30998.999.5
m15_diffdrive_100k.npz 30399.099.4 30999.899.9
m28_news_25k.npz 56383.484.2 561196.597.8
m28_moore_25k.npz 56573.381.0 56996.597.9
m28_diffdrive_25k.npz 56382.093.6 561195.398.0

Feel free to play around with different iteration counts --k and kernel sizes --f.

IV. Version differences (Python, PyTorch)

The test performance results above are slightly different from what is reported in our ICML 2018 paper due to version differences in Python (3.6.5 vs. 2.7.12) and PyTorch (0.4.0 vs. 0.3.1).

Below, we provide instructions to exactly replicate the numbers reported in our ICML 2018 paper.

  1. Checkout the Git branch icml2018:
git checkout icml2018
  1. Load the Docker container used in our experiments by running:
# PyTorch with GPU support
nvidia-docker pull lileee/python-2.7-pytorch-0.3.1-custom:latest
nvidia-docker run -v $(pwd):/home --rm -ti lileee/python-2.7-pytorch-0.3.1-custom:latest

This Docker container uses Python 2.7.12 and a custom version of PyTorch 0.3.1 compiled from source at https://github.com/eparisotto/pytorch.

  1. Train a model:
python train.py --datafile mazes/m15_news_25k.npz --mechanism news --model models.VIN --k 20 --f 5

Citation

If you found this code useful in your research, please cite:

@inproceedings{gppn2018,
  author    = {Lisa Lee and Emilio Parisotto and Devendra Singh Chaplot and Eric Xing and Ruslan Salakhutdinov},
  title     = {Gated Path Planning Networks},
  booktitle = {Proceedings of the 35th International Conference on Machine Learning (ICML 2018)},
  year      = {2018}
}

Acknowledgments

Thanks to @kentsommer for releasing a PyTorch implementation of the original VIN results, which served as a starting point for this codebase.