It's interesting work. It's a kinda like finding a really weak seemingly impossible to use buffer overflow and now someone has to weaponize it and put it into easy to use metasploit to become just one of 1000s of things to have available.
Personally I'm surprised all these robots don't have microphones yet. Not being able to talk to robots makes them pretty lame.
Hi, first author of the paper here. We also consider this as part of the increasing arsenal of smart-home attacks, which can be opportunistic and long-term. Also given that it's an offline attack, as signal processing / machine learning methods improve, perhaps the lidars signals an attacker collects could eventually become intelligible audio.
I was also surprised that they don't have microphones. I guess the developers would prefer to have that on the companion app instead.
https://umd.app.box.com/s/7qkltjg5xs6cpbjllu8fajpelbs736cm
It's interesting work. It's a kinda like finding a really weak seemingly impossible to use buffer overflow and now someone has to weaponize it and put it into easy to use metasploit to become just one of 1000s of things to have available.
Personally I'm surprised all these robots don't have microphones yet. Not being able to talk to robots makes them pretty lame.