• Classified by Topic • Classified by Publication Type • Sorted by Date • Sorted by First Author Last Name • Classified by Funding Source •
Reward (Mis)design for Autonomous Driving.
W Bradley Knox, Alessandro Allievi,
Holger Banzhaf, Felix Schmitt, and Peter Stone.
arXiv preprint arXiv:2104.13906,
2021.
Paper webpage
This paper considers the problem of reward design for autonomous driving (AD), with insights that are also applicable to the design of cost functions and performance metrics more generally. Herein we develop 8 simple sanity checks for identifying flaws in reward functions. The sanity checks are applied to reward functions from past work on reinforcement learning (RL) for autonomous driving, revealing near-universal flaws in reward design for AD that might also exist pervasively across reward design for other tasks. Lastly, we explore promising directions that may help future researchers design reward functions for AD.
@article{knox2021reward, title={Reward (Mis)design for Autonomous Driving}, author={W Bradley Knox and Alessandro Allievi and Holger Banzhaf and Felix Schmitt and Peter Stone}, journal={arXiv preprint arXiv:2104.13906}, year={2021}, abstract={ This paper considers the problem of reward design for autonomous driving (AD), with insights that are also applicable to the design of cost functions and performance metrics more generally. Herein we develop 8 simple sanity checks for identifying flaws in reward functions. The sanity checks are applied to reward functions from past work on reinforcement learning (RL) for autonomous driving, revealing near-universal flaws in reward design for AD that might also exist pervasively across reward design for other tasks. Lastly, we explore promising directions that may help future researchers design reward functions for AD. }, wwwnote={<a href="https://www.bradknox.net/reward-misdesign/">Paper webpage</a>} }
Generated by bib2html.pl (written by Patrick Riley ) on Mon Jan 16, 2023 08:56:38