Self-Driving Cars’ All-Seeing LIDAR Blinded By $60 Raspberry Pi-Powered Pulse Laser

With vehicle makers focusing on making our future travel autonomous, a major concern arises: is it going to be safe and secure? If companies expect people to trust their lives with their technology, it goes without saying that it must be bullet-proof. There can be no compromises. Uber is a company that understands this very well, as it recently beefed-up its security brain-power by hiring two people who've proven that all of the technology lacing our vehicles could prove to be a serious detriment if it's insecure.

It's a good thing that autonomous vehicles haven't littered the market quite yet, as it's clear that there is still much work to be done. Thanks to Jonathan Petit, Principal Scientist at Security Innovation, we have all the proof we need of that. At Black Hat Europe, to be held in November, Jonathan is going to highlight how a vehicle's "LIDAR" system - the one entirely responsible for building a picture of what's around it - could be fooled with a mere $60 worth of off-the-shelf components.

Lidar Example
An example of LIDAR technology, courtesy of Leidos

LIDAR works by beaming invisible lasers all around a vehicle to gain an understanding of the obstacles around it - either vehicles in front or back, pedestrians crossing the street, or what-have-you. For autonomous driving to work well, LIDAR also needs to work well, and have its signals uninterrupted.

With Jonathan's DIY device, his own lasers can interfere with LIDAR, in effect allowing him to place or move objects that are not even there. Done right, this could cause a DDoS type of attack for a vehicle, forcing it to remain in place because it believes there are obstacles in front of it.

It's said that the system has an accuracy of up to 350 meters, and that it could spoof "thousands" of objects to interfere with LIDAR. Clearly, if the DIY device is powerful enough, this could cause some serious headaches for drivers. Jonathan does admit though that there's only one particular system right now that's susceptible to this kind of attack, so it appears there are systems out there that are in fact bulletproof. At least for now, that is.


Via:  IEEE
Show comments blog comments powered by Disqus