Skip to content

xl-sr/CAL

master
Switch branches/tags

Name already in use

A tag already exists with the provided branch name. Many Git commands accept both tag and branch names, so creating this branch may cause unexpected behavior. Are you sure you want to create this branch?
Code

Latest commit

 

Git stats

Files

Permalink
Failed to load latest commit information.
Type
Name
Latest commit message
Commit time
 
 
 
 
 
 
 
 
 
 
 
 

🌟 We updated the repo with training code! 🌟

Conditional Affordance Learning

In CoRL 2018 [Paper] [Video] [Talk]

Axel Sauer 1, 2, Nikolay Savinov 1, Andreas Geiger 1, 3
1 ETH Zurich, 2 TU Munich, 3 MPI for Intelligent Systems and University of Tubingen

If you use this implementation, please cite our CoRL 2018 paper.

@inproceedings{Sauer2018CORL,
  author={Sauer, Axel and Savinov, Nikolay and Geiger, Andreas},
  title={Conditional Affordance Learning for Driving in Urban Environments},
  booktitle={Conference on Robot Learning (CoRL)},
  year={2018}
}

Installation

# install anaconda if you don't have it yet
wget https://repo.continuum.io/archive/Anaconda3-5.3.0-Linux-x86_64.sh
bash Anaconda3-5.3.0-Linux-x86_64.sh
source ~/.profile
# or use source ~/.bashrc - depending on where anaconda was added to PATH as the result of the installation

Now we will:

  1. create a conda environment named CAL and install all dependencies
  2. download the binaries for CARLA version 0.8.2 [CARLA releases]
  3. download the model weights
git clone https://github.com/xl-sr/CAL.git
cd CAL

# create the training environment
conda env create -f requirements.yml
source activate CAL

# run download script
./download_binaries_and_models.sh

Run the Agent

In CARLA_0.8.2/ start the server with for example: (see the CARLA documentation for details)

./CarlaUE4.sh Town01 -carla-server -windowed -benchmark -fps=20 -ResX=800 - ResY=600

Open a second terminal, and run:

python python_client/driving_benchmark.py -c Town01 -v -n test --corl-2017

If you want to run the CORL 2017 benchmark add '--corl-2017'

python python_client/driving_benchmark.py -c Town01 -v -n test --corl-2017

If you want to continue an experiment, you can add the 'continue-experiment' flag.

Training

cd training/

# download and untar the dataset
wget https://s3.eu-central-1.amazonaws.com/avg-projects/conditional_affordance_learning/dataset.tar.gz
tar -xzvf dataset.tar.gz

Now, open training_CAL.ipynb. The notebook walks you through the steps to train a network on the dataset.

About

[CoRL'18] Conditional Affordance Learning

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published