Lack of robustness of Lidar-based deep learning systems to small adversarial perturbations

Naman Patel, Kang Liu, Prashanth Krishnamurthy, Siddharth Garg, Farshad Khorrami

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

In this paper, we investigate the robustness of LIDAR-based autonomous navigation for unmanned vehicles using Deep Neural Networks (DNN) to adversarial perturbations. A well-trained network robust to sensor noise can yield an undesirable network response (e.g., steering the vehicle in a wrong direction) by maliciously crafted perturbations in sensor data. We show through experimental evaluations on our unmanned ground vehicle (UGV) that small perturbations in some of the LIDAR sensor data (even perturbations smaller than the sensor accuracy) can lead the DNN to generate incorrect outputs. This is somewhat unexpected from a sensor such as LIDAR, which provides very well-defined structural/geometrical information about the environment.

Original languageEnglish (US)
Title of host publication50th International Symposium on Robotics, ISR 2018
PublisherVDE Verlag GmbH
Pages359-365
Number of pages7
ISBN (Electronic)9781510870314
StatePublished - Jan 1 2018
Event50th International Symposium on Robotics, ISR 2018 - Munich, Germany
Duration: Jun 20 2018Jun 21 2018

Publication series

Name50th International Symposium on Robotics, ISR 2018

Other

Other50th International Symposium on Robotics, ISR 2018
CountryGermany
CityMunich
Period6/20/186/21/18

    Fingerprint

ASJC Scopus subject areas

  • Artificial Intelligence
  • Human-Computer Interaction

Cite this

Patel, N., Liu, K., Krishnamurthy, P., Garg, S., & Khorrami, F. (2018). Lack of robustness of Lidar-based deep learning systems to small adversarial perturbations. In 50th International Symposium on Robotics, ISR 2018 (pp. 359-365). (50th International Symposium on Robotics, ISR 2018). VDE Verlag GmbH.