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.
.gitignore
README.md Update README.md Nov 10, 2016
car-blue.png
car-gray.png first commit Aug 7, 2016
car-orange.png first commit Aug 7, 2016
car-purple.png
car-red.png first commit Aug 7, 2016
car-white.png first commit Aug 7, 2016
car-yellow.png first commit Aug 7, 2016
car.py
cone.png first commit Aug 7, 2016
dynamics.py first commit Aug 7, 2016
feature.py first commit Aug 7, 2016
firetruck.png first commit Aug 7, 2016
gen.py
grass.png first commit Aug 7, 2016
irl.py
lane.py first commit Aug 7, 2016
run first commit Aug 7, 2016
trajectory.py first commit Aug 7, 2016
utils.py first commit Aug 7, 2016
vis
visualize.py first commit Aug 7, 2016
world.py

README.md

Toolkit for learning driving models through maximum entropy inverse reinforcement learning, and autonomous vehicle's control through leverageing effects on human actions.

(Companion code to a paper presented at RSS 2016)

Running

To visualize: ./vis {file_name}.pickle

To run an experiment ./run {world_name} where world_name can be any one of the worlds defined in world.py

To run an experiment with irl_ground world: ./run irl_ground

To run the IRL algorithm: ./irl.py data/*.pickle

Modules

  • dynamics.py: This contains code for car dynamics.
  • car.py: Relevant code for different car models (human-driven, autonomous, etc.)
  • feature.py: Definition of features.
  • lane.py: Definition of driving lanes.
  • trajectory.py: Definition of trajectories.
  • world.py: This code contains different scenarios (each consisting of lanes/cars/etc.).
  • visualize.py: This contains the code for visualization (GUI).