Chapter 2 Efficient Statistical Validation of Autonomous Driving Systems Handi Yu, Weijing Shi, Mohamed Baker Alawieh, Changhao Yan, Xuan Zeng, Xin Li, and Huafeng Yu 2.1 Introduction The last decade has witnessed tremendous advance in Advanced Driver Assistance System (ADAS) [1] and autonomous driving [2, 3]. Such systems can perform numerous intelligent functions [4] such as collision avoidance, lane departure warning, traffic sign detection, etc. When implementing these systems, machine learning plays an essential role to interpret sensor data and understand surrounding environment. These machine learning systems significantly improve driving safety and comfort, but in turn raise new challenges concerning system robustness and reliability. One main challenge is that most autonomous driving systems heavily rely on visual perception and are sensitive to environmental variations, e.g., rain, fog, etc. H. Yu Duke University, Durham, NC, USA e-mail: hy126@duke.edu W. Shi · M. B. Alawieh Carnegie Mellon University, Pittsburgh, PA, USA e-mail: weijings@cmu.edu; malawieh@cmu.edu C. Yan · X. Zeng Fudan University, Shanghai, P. R. China e-mail: yanch@fudan.edu.cn; xzeng@fudan.edu.cn X. Li () Duke University, Durham, NC, USA Duke Kunshan University, Kunshan, P.R. China e-mail: xinli.ece@duke.edu H. Yu Boeing Research and Technology, Huntsville, AL, USA e-mail: huafeng.yu@boeing.com © Springer Nature Switzerland AG 2019 H. Yu et al. (eds.), Safe, Autonomous and Intelligent Vehicles, Unmanned System Technologies, https://doi.org/10.1007/978-3-319-97301-2_2 5