/driving-preferences

Primary LanguagePython

This code learns reward functions from human preferences in driving scenarios by actively generating scenarios and querying a human expert.

(Companion code to RSS 2017 paper)

Running

To run simply execute run.py

Modules

  • dynamics.py: This contains code for car dynamics.
  • world.py: This contains all of the code for the driving environment and cars (except for the dynamics).
  • visual.py: This contains the code for visualization (GUI).
  • sampling.py: This contains the code for Markov Chain sampling of the distributions.
  • genplots.py: This module contains the code for generating the plots.