تقرير
Reward (Mis)design for Autonomous Driving
العنوان: | Reward (Mis)design for Autonomous Driving |
---|---|
المؤلفون: | Knox, W. Bradley, Allievi, Alessandro, Banzhaf, Holger, Schmitt, Felix, Stone, Peter |
سنة النشر: | 2021 |
المجموعة: | Computer Science |
مصطلحات موضوعية: | Computer Science - Machine Learning, 91B16, I.2.6, I.2.8, I.2.9 |
الوصف: | This article considers the problem of diagnosing certain common errors in reward design. Its insights are also applicable to the design of cost functions and performance metrics more generally. To diagnose common errors, we develop 8 simple sanity checks for identifying flaws in reward functions. These sanity checks are applied to reward functions from past work on reinforcement learning (RL) for autonomous driving (AD), 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 aid the design of reward functions for AD in subsequent research, following a process of inquiry that can be adapted to other domains. Comment: 14 pages (27 pages with appendix), 4 figures |
نوع الوثيقة: | Working Paper |
URL الوصول: | http://arxiv.org/abs/2104.13906 |
رقم الأكسشن: | edsarx.2104.13906 |
قاعدة البيانات: | arXiv |
الوصف غير متاح. |