Laser beam attacks on lane detection models

With the rise in popularity of autonomous vehicles in the world today, ensuring the safety of these vehicles is of utmost importance. Autonomous vehicles use Autonomous Driving Systems (ADS), which collect inputs from cameras and sensors to be sent through deep neural networks to produce relevant ou...

Full description

Saved in:
Bibliographic Details
Main Author: Tay, Ryan Edward Siang An
Other Authors: Tan Rui
Format: Final Year Project
Language:English
Published: Nanyang Technological University 2023
Subjects:
Online Access:https://hdl.handle.net/10356/166049
Tags: Add Tag
No Tags, Be the first to tag this record!
Institution: Nanyang Technological University
Language: English
Description
Summary:With the rise in popularity of autonomous vehicles in the world today, ensuring the safety of these vehicles is of utmost importance. Autonomous vehicles use Autonomous Driving Systems (ADS), which collect inputs from cameras and sensors to be sent through deep neural networks to produce relevant output for the car to make real-time decisions. As the ADS is susceptible to cybersecurity attacks such as adversarial attacks, more research is required to better prepare these systems against future attacks. This paper will be focusing on one possible method of attack, through the use of a laser beam. The approach taken in this paper was to use different methods of laser beam attacks to test the accuracy of different lane detection models. The test results for each lane detection model determine the type of laser beam attack that the model is vulnerable to.