Communications of the ACM
WARNING: Objects in Driverless Car Sensors May Be Closer Than They Appear
Researchers have shown that a popular method to secure LiDAR sensors against "naive attacks" is still vulnerable at longer distances and only works at short distances.
Credit: Duke University Pratt School of Engineering
Duke University researchers have identified an attack strategy that can trick industry-standard autonomous vehicle sensors into believing nearby objects are closer or further than they appear.
This involves using a laser gun to strategically place three-dimensional LiDAR data points within a certain area of the vehicle camera's two-dimensional field of view.
The researchers determined the vulnerable area extends out in front of the camera's lens in a frustum shape.
"This so-called frustum attack can fool adaptive cruise control into thinking a vehicle is slowing down or speeding up," said Duke's Miroslav Pajic. Pajic suggested adding redundancy in the form of "stereo cameras" with overlapping fields of view to better estimate distances and detect LiDAR data that does not match their perception, or the developing systems that allow cars in close proximity to share some of their data.
From Duke University Pratt School of Engineering
View Full Article
Abstracts Copyright © 2022 SmithBucklin, Washington, DC, USA
No entries found