I'f you put a fake road sign somewhere people are probably going to obey it as well.
Hacked billboards make Teslas see phantom objects causing them to swerve or stop
151–160 of 264 posts
Re: Hacked billboards make Teslas see phantom objects causing them to swerve or stop
#152Earlier quoted context omitted.
Freedom of speech is meant to be respected by government. Private businesses aren’t obligated to respect anyone’s free speech.
Today they silence the voices of everyone you hate, tomorrow they silence your voice. Its a very slippery slope and it is dangerous. Corporations should never pretend to be the guardians of truth. Bias will just spew from there about whats supposedly real and whats supposedly fake.
The root cause here is that we have corporations big enough that their censorship matters.
Re: Hacked billboards make Teslas see phantom objects causing them to swerve or stop
#153Earlier quoted context omitted.
> They should only detect driveable space. I'm interested to hear how you decide what's drivable without detecting things that decide if roads are undrivable.
You don't need to know it's a "car" or a "cow" or a "corvette" to know that the flat surface you are on ends in less than your stopping distance. Semantically, it's a meaningless distinction. But in implementation it's much easier to detect the shape of the upcoming terrain using, say, lidar or regular old multi-camera disparity. Flat and following your map: that's driveable. I'm guessing many take this approach, but…
A one-way street is also a drivable flat surface. As is a road with no lane markings.
Can you guess why blindly driving down each of these will be dangerous?
Re: Hacked billboards make Teslas see phantom objects causing them to swerve or stop
#154There are a lot of similar failure modes in humans. Sometimes they're are used for good, like making drivers slow down: https://www.insider.com/optical-illusions-3d-crosswalk-drivi... But other times they cause crashes: https://imgur.com/a/kYr94
I wonder if these self driving cars would recognise those 3D crossings as the same as the normal thing. I could see it looking quite different to a neural network or something looking for features and a bit overfitted.
Re: Hacked billboards make Teslas see phantom objects causing them to swerve or stop
#155I'f you put a fake road sign somewhere people are probably going to obey it as well.
That is likely, if the fake street sign were to resemble a real one both in appearance and context.
If it were a stop sign in a billboard ad, humans would ignore it, though.
Re: Hacked billboards make Teslas see phantom objects causing them to swerve or stop
#156Cars should not detect "obstacles". They should detect roads. They should not seek _exceptions_ to driveable space through classification- or segmentation-first strategies. They should only detect driveable space. Full stop. You either detect a full-stopping-distance's worth of clear road through reliable (non visual) ranging with dense enough samples to preclude wheel-catching voids ... or you will continue to run i…
This wouldn’t work well with moving obstacles. The road could be clear until a person enters the road within your full-stopping distance.
Re: Hacked billboards make Teslas see phantom objects causing them to swerve or stop
#157Earlier quoted context omitted.
You don't need to know it's a "car" or a "cow" or a "corvette" to know that the flat surface you are on ends in less than your stopping distance. Semantically, it's a meaningless distinction. But in implementation it's much easier to detect the shape of the upcoming terrain using, say, lidar or regular old multi-camera disparity. Flat and following your map: that's driveable. I'm guessing many take this approach, but…
A one-way street is also a drivable flat surface. As is a road with no lane markings. Can you guess why blindly driving down each of these will be dangerous?
Right, but can you guess why stopping in the middle of traffic for a fake stop sign is bad?
Walk before you run is the essence of the argument here.
Re: Hacked billboards make Teslas see phantom objects causing them to swerve or stop
#158Anything with software sadly becomes a weapon. Especially when its heavy bulky and potentially deadly.
There will be a lot of money thrown on adversarial attack mitigation. But how much is too much? At some point it might become so expensive, that it will be cheaper to switch to guiderail-based self driving, which can be overseen and secured centrally. And self-driving planes are probably going to be a nonstarter
Re: Hacked billboards make Teslas see phantom objects causing them to swerve or stop
#159Earlier quoted context omitted.
The idea to remove all billboards for teslas is really funny. I would say that this billboard hack means anyone could install a banner on their balcony to get a tesla to swerve onto a neighbour’s house. However, I believe this will be fixed in a later update.
> I believe this will be fixed in a later update Maybe. I hope so. But from my understanding, context is the hardest part of computer vision.
Yes, human reasoning is not just pattern matching but mixed with deductive thinking. A picture of a stop sign is not a stop sign, neither is a reflection of one. Trying to teach a neural network all of these exceptions seems like a terrible approach and would likely not work very well.
Re: Hacked billboards make Teslas see phantom objects causing them to swerve or stop
#160I think the concept of the line of death, from browser security UI design, is relevant here. The space outside of the highway right of way is completely uncontrolled and the Tesla should know better than to look at a billboard hundreds of feet up in the air for regulatory signs. The MUTCD has rules about where signs are to be placed--look only in those places. [1] https://textslashplain.com/2017/01/14/the-line-of-dea…
The problem is that they don’t have lidar and therefore can’t reliably measure depth, right? A billboard far away high up in the air may appear in the same direction as a closer road sign if you don’t have depth perception. You should be able to infer it from a camera with the right model but those methods are more fallible. Similar root cause to that time the guy got beheaded by a white truck perpendicular to the highway that looked like the sky.