DropoutDAgger: A Bayesian Approach to Safe Imitation Learning

@article{Menda2017DropoutDAggerAB,
  title={DropoutDAgger: A Bayesian Approach to Safe Imitation Learning},
  author={Kunal Menda and Katherine Rose Driggs-Campbell and Mykel J. Kochenderfer},
  journal={CoRR},
  year={2017},
  volume={abs/1709.06166}
}
While imitation learning is becoming common practice in robotics, this approach often suffers from data mismatch and compounding errors. DAgger is an iterative algorithm that addresses these issues by continually aggregating training data from both the expert and novice policies, but does not consider the impact of safety. We present a probabilistic… CONTINUE READING