all AI news
Identify, Estimate and Bound the Uncertainty of Reinforcement Learning for Autonomous Driving. (arXiv:2305.07487v1 [cs.AI])
cs.LG updates on arXiv.org arxiv.org
Deep reinforcement learning (DRL) has emerged as a promising approach for
developing more intelligent autonomous vehicles (AVs). A typical DRL
application on AVs is to train a neural network-based driving policy. However,
the black-box nature of neural networks can result in unpredictable decision
failures, making such AVs unreliable. To this end, this work proposes a method
to identify and protect unreliable decisions of a DRL driving policy. The basic
idea is to estimate and constrain the policy's performance uncertainty, which …
application arxiv autonomous autonomous driving autonomous vehicles box decision driving identify intelligent making nature network networks neural network neural networks policy reinforcement reinforcement learning uncertainty