3 ms·
V2V and V2X serve as additional ways for a car to get data, to complement the car's own sensors. They're not command protocols. V2V shouldn't make your car do a
by ACS_Solver 5y ago
V2V and V2X serve as additional ways for a car to get data, to complement the car's own sensors. They're not command protocols. V2V shouldn't make your car do anything dangerous, as all the usual software logic still applies. E.g. your car may get info over V2V that an ambulance with sirens is coming up behind you, so your car slows down to make room, but that's an internal decision of the car, it's not a "slow down" command over V2V channels.
In that vein, rogue traffic signs or other objects designed to confuse a car's inputs are probably more of a threat.
- angry_octet 5y agoFalse V2V inputs could cause a car satnav to divert to an alternative route. If the data fusion is done wrong, or if external visibility is very poor, it could rely on single source data (V2V without confirmation from an on-board sensor) and swerve or engage emergency braking. If an attacker has access to disable/deceive, e.g., the microwave sensors (via software attacks, or by jamming) then it becomes quite possible. There is lots of research in the topic though, so I'm fairly confident most V2V systems will be robust, but it depends on regulation. If they froze capability at a specific 'approved' version then attacks could become serious. Especially for systems using lots of ML, at higher levels of autonomy. At the moment it seems like Looney Tunes attacks (draw a picture of a tunnel with the word TUNNEL on it, paint the road markers towards it) work amazingly well.