Self-driving cars, like human drivers before them, need to see their surroundings to avoid obstacles and drive safely.
The most sophisticated autonomous vehicles typically use lidar, a rotating radar-like device that acts like the eyes of the car. The lidar provides constant information about the distance to objects so that the car can decide what actions to take safely.
But those eyes, it turns out, can be deceived.
New research reveals that expertly timed lasers aimed at an approaching lidar system can create a blind spot in front of the vehicle large enough to completely obscure moving pedestrians and other obstacles. The deleted data makes cars think the road is safe to keep moving forward, endangering anything that may be in the attack’s blind spot.
This is the first time that lidar sensors have had to remove data on obstacles.
The vulnerability was discovered by researchers from the University of Florida, the University of Michigan, and the University of Electro-Communications in Japan. Scientists also provide upgrades that could eliminate this weakness to protect people from malicious attacks.
The results will be presented at the USENIX Security Symposium 2023 and are currently published on arXiv.
Lidar works by emitting laser light and capturing reflections to calculate distances, much like how a bat’s echolocation uses sound echoes. The attack creates false reflections to blur the sensor.
“We mimic lidar reflections with our laser so the sensor doesn’t take into account other reflections from real obstacles,” said Sara Rampazzi, professor of computer science and information science and engineering at the UF that conducted the study. “The lidar still receives authentic data from the obstacle, but the data is automatically discarded because our false reflections are the only ones seen by the sensor.”
The scientists demonstrated the attack on moving vehicles and robots with the attacker placed about 15 feet on the side of the road. But in theory, this could be accomplished from further away with improved equipment. The technology required is fairly basic, but the laser must be perfectly synchronized with the lidar sensor and moving vehicles must be carefully tracked to keep the laser pointing in the right direction.
“It’s mostly a matter of timing the laser with the lidar device. The information you need is usually publicly available from the manufacturer,” said S. Hrushikesh Bhupathiraj, a UF PhD student in Rampazzi’s lab and one of the main authors of the study. .
Using this technique, scientists were able to remove data from static obstacles and moving pedestrians. They also demonstrated, using real world experiments, that the attack could track a slow-moving vehicle using basic camera tracking equipment. In decision-making simulations of an autonomous vehicle, this data suppression caused a car to continue accelerating toward a pedestrian it could no longer see instead of stopping as it should.
Updates to the lidar sensors or the software that interprets the raw data could address this vulnerability. For example, manufacturers could teach software to look for telltale signatures of spoofed highlights added by the laser attack.
“Revealing this responsibility allows us to build a more reliable system,” said Yulong Cao, a Michigan doctoral student and lead author of the study. “In our article, we demonstrate that the previous defense strategies are not enough, and we propose modifications that should remedy this weakness.”
Autonomous vehicles can be tricked into “seeing” non-existent obstacles
Yulong Cao et al, You Can’t See Me: Physical Removal Attacks on LiDAR-based Autonomous Vehicles Driving Frameworks, arXiv (2022). DOI: 10.48550/arxiv.2210.09482. arxiv.org/abs/2210.09482
arXiv
Provided by the University of Florida
Quote: Laser attack blinds autonomous vehicles, deletes pedestrians and confuses cars (2022, October 31) Retrieved November 1, 2022 from https://techxplore.com/news/2022-10-laser-autonomous-vehicles-deleting -pedestrians.html
This document is subject to copyright. Except for fair use for purposes of private study or research, no part may be reproduced without written permission. The content is provided for information only.
#Laser #attack #blinds #selfdriving #vehicles #suppresses #pedestrians #confuses #cars