Symbolic Perception Risk in Autonomous Driving

Guangyi LiuDisha KamaleCristian-Ioan VasileNader Motee

Mar 2023

摘要

We develop a novel framework to assess the risk of misperception in a trafficsign classification task in the presence of exogenous noise. We consider theproblem in an autonomous driving setting, where visual input quality graduallyimproves due to improved resolution, and less noise since the distance totraffic signs decreases. Using the estimated perception statistics obtainedusing the standard classification algorithms, we aim to quantify the risk ofmisperception to mitigate the effects of imperfect visual observation. Byexploring perception outputs, their expected high-level actions, and potentialcosts, we show the closed-form representation of the conditional value-at-risk(CVaR) of misperception. Several case studies support the effectiveness of ourproposed methodology.

展开全部

图表提取

论文十问

参考文献

被引用

提问与回答

暂无人提供速读十问回答

被引用

笔记

问答

领域