Conditional Imitation Learning at CARLA

Repository to store the conditional imitation learning based AI that runs on carla. The trained model is the one used on "CARLA: An Open Urban Driving Simulator" paper.

Requirements

tensorflow_gpu 1.1 or more

numpy

scipy

carla 0.8.2

PIL

Running

Basically run:

$ python run_CIL.py

Note that you must have a carla server running .
To check the other options run

$ python run_CIL.py --help

Dataset

The dataset can be downloaded here 24 GB

The data is stored on HDF5 files. Each HDF5 file contains 200 data points. The HDF5 contains two "datasets": 'images_center':
The RGB images stored at 200x88 resolution

'targets':
All the controls and measurements collected. They are stored on the "dataset" vector.

1. Steer, float
2. Gas, float
3. Brake, float
4. Hand Brake, boolean
5. Reverse Gear, boolean
6. Steer Noise, float
7. Gas Noise, float
8. Brake Noise, float
9. Position X, float
10. Position Y, float
11. Speed, float
12. Collision Other, float
13. Collision Pedestrian, float
14. Collision Car, float
15. Opposite Lane Inter, float
16. Sidewalk Intersect, float
17. Acceleration X,float
18. Acceleration Y, float
19. Acceleration Z, float
20. Platform time, float
21. Game Time, float
22. Orientation X, float
23. Orientation Y, float
24. Orientation Z, float
25. High level command, int ( 2 Follow lane, 3 Left, 4 Right, 5 Straight)
26. Noise, Boolean ( If the noise, perturbation, is activated, (Not Used) )
27. Camera (Which camera was used)
28. Angle (The yaw angle for this camera)

Paper

If you use the conditional imitation learning, please cite our ICRA 2018 paper.

End-to-end Driving via Conditional Imitation Learning.
Codevilla, Felipe and Müller, Matthias and López, Antonio and Koltun, Vladlen and Dosovitskiy, Alexey. ICRA 2018 [PDF]

@inproceedings{Codevilla2018,
  title={End-to-end Driving via Conditional Imitation Learning},
  author={Codevilla, Felipe and M{\"u}ller, Matthias and L{\'o}pez,
Antonio and Koltun, Vladlen and Dosovitskiy, Alexey},
  booktitle={International Conference on Robotics and Automation (ICRA)},
  year={2018},
}