Write a Blog >>
ICSE 2020
Wed 24 June - Thu 16 July 2020
Tue 7 Jul 2020 15:12 - 15:24 at Baekje - A1-Autonomous Driving Systems Chair(s): Donghwan Shin

Deep Neural Networks (DNNs) have been widely applied in autonomous systems such as self-driving vehicles. Recently, DNN testing has been intensively studied to automatically generate adversarial examples, which inject small-magnitude perturbations into inputs to test DNNs under extreme situations. While existing testing techniques prove to be effective, particularly for autonomous driving, they mostly focus on generating digital adversarial perturbations, e.g., changing image pixels, which may never happen in the physical world. Thus, there is a critical missing piece in the literature on autonomous driving testing: understanding and exploiting both digital and physical adversarial perturbation generation for impacting steering decisions. In this paper, we propose a systematic physical-world testing approach, namely DeepBillboard, targeting at a quite common and practical driving scenario: drive-by billboards. DeepBillboard is capable of generating a robust and resilient printable adversarial billboard test, which works under dynamic changing driving conditions including viewing angle, distance, and lighting. The objective is to maximize the possibility, degree, and duration of the steering-angle errors of an autonomous vehicle driving by our generated adversarial billboard. We have extensively evaluated the efficacy and robustness of DeepBillboard by conducting both experiments with digital perturbations and physical-world case studies. The digital experimental results show that DeepBillboard is effective for various steering models and scenes. Furthermore, the physical case studies demonstrate that DeepBillboard is sufficiently robust and resilient for generating physical-world adversarial billboard tests for real-world driving under various weather conditions, being able to mislead the average steering angle error up to 26.44 degrees. To the best of our knowledge, this is the first study demonstrating the possibility of generating realistic and continuous physical-world tests for practical autonomous driving systems; moreover, DeepBillboard can be directly generalized to a variety of other physical entities/surfaces along the curbside, e.g., a graffiti painted on a wall.

Tue 7 Jul
Times are displayed in time zone: (UTC) Coordinated Universal Time change

15:00 - 16:00: Paper Presentations - A1-Autonomous Driving Systems at Baekje
Chair(s): Donghwan ShinUniversity of Luxembourg (SnT)
icse-2020-papers15:00 - 15:12
Shafiul Azam ChowdhuryUniversity of Texas at Arlington, Sohil Lal ShresthaThe University of Texas at Arlington, Taylor T JohnsonVanderbilt University, Christoph CsallnerUniversity of Texas at Arlington
Link to publication DOI Media Attached
icse-2020-papers15:12 - 15:24
Husheng ZhouThe University of Texas at Dallas, Wei LiSouthern University of Science and Technology, Zelun KongThe University of Texas at Dallas, Junfeng GuoThe University of Texas at Dallas, Yuqun ZhangSouthern University of Science and Technology, Lingming ZhangThe University of Texas at Dallas, Bei YuThe Chinese University of Hong Kong, Cong LiuUT Dallas
icse-2020-papers15:24 - 15:36
Andrea StoccoUniversità della Svizzera italiana, Michael WeissUniversità della Svizzera Italiana (USI), Marco CalzanaUniversità della Svizzera Italiana (USI), Paolo TonellaUniversità della Svizzera italiana
icse-2020-papers15:36 - 15:48
Claudio MenghiUniversity of Luxembourg, Shiva NejatiUniversity of Ottawa, Lionel BriandSnT Centre/University of Luxembourg, Yago Isasi ParacheLuxSpace
icse-2020-papers15:48 - 16:00
Joshua GarciaUniversity of California, Irvine, Yang FengNanjing University, Junjie ShenUniversity of California, Irvine, Sumaya AlmaneeUniversity of California, Irvine, Yuan XiaUniversity of California, Irvine, Qi Alfred ChenUniversity of California, Irvine