7 ms·

This is exactly why we need to escape the mentality of sensor fusion and make our robots reliable just based on optical data. Then they would be susceptible to

This is exactly why we need to escape the mentality of sensor fusion and make our robots reliable just based on optical data. Then they would be susceptible to attacks that would affect humans too.

ent 8y ago
I took this as an argument on why we shouldn't rely on one type of sensor alone, as it's easier to fool with tricks like this.
pfortuny 8y ago
Afaik equilibrium is a matter of at least vision and audition, is it not?
guav_ 8y ago
Humans with issues in their vestibular system located in the ear can't really use their optical data streams to compensate. When you loose your gyros time to get a walking stick or a beautiful nurse to lean on.
- StavrosK 8y ago
  Can humans with issues in the optical stream use the gyros? I find it harder to balance with my eyes closed, though obviously not impossible.
  - falcolas 8y ago
    Well, the blind and sight-impaired have been walking (and running) for years, so I'd vote "yes". Even people with impressive visual hallucinations are still capable of walking. That said, vertigo (where your eyes and your inner ear are in violent disagreement as to whether you are moving) can impair movement, though you can grow accustomed to the "disagreement".
  - tehbeard 8y ago
    I think the optical stream fits more of an error correction/secondary data stream in this case. The difficulty may be down to you trying to use other senses and overcompensating on the data (e.g. air pressure, which tells you something is near-ish you but not if it's about to contact you so you react)
- jerf 8y ago
  Interestingly, it is possible to learn to walk when you never had a vestibular system, although the gait is visibly distorted and when you know what you are looking for you can easily spot it. Those who suffer from CHARGE, and are left able to walk, can still do so despite sometimes entirely missing their vestibular system. But learning to walk with a vestibular system and then losing it later seems to be very, very difficult to recover from, if it is possible at all.
eatitraw 8y ago
Humans fuse a lot of sensory data though. For example, walking without proprioception is incredibly hard: https://philosophyandpsychology.wordpress.com/2007/10/14/living-without-proprioception/ https://philosophyandpsychology.wordpress.com/2007/10/14/liv...
- noir_lord 8y ago
  I have nerve damage because of a spinal problem, I walk with a slight limp not because I've lost strength but because I can't always accurately 'feel' where my foot is. It's like walking with pins and needles.
verelo 8y ago
Humans use multiple sensors together at one. Your inner ear is a great example. If you see a picture doing one thing, but your inner ear feels another, this is the senstation that causes motion sickness. I cannot find the source right now, but i recall being taught that research dating back to US airforce simulation training suggested that this might be a built in evolutionary function that is designed to make us throw up. The basic narrative being "I'm seeing something I'm not feeling, i must have eaten something that's poisoned me, i should throw up so i don't die". Edit: As a side note, this is why I'm extremely skeptical of things like VR technology in general (as a person that owns a HTC Vive). I'm not sure we're going to be able to easily convince the human brain it's OK for different sensors to feel different things, and not result in us feeling sick.
- freehunter 8y ago
  Re: your side note That's why even though I'm a relatively avid gamer with plenty of disposable income, I've never bought on to the VR stuff. It's not about the cost, or the rapidly improving technology, it's that I don't believe (long term) that I'm going to really enjoy many VR games. I have a racing wheel because I like to play racing games, but most of the time I end up using my controller anyway. Why? Because it doesn't feel real. The wheel ends up feeling farcical because yeah I'm cornering at 1.5G, but I don't feel a thing. In real life you're less likely to corner too fast because you can feel that you're going too fast. You're being shoved around, you're being moved side to side, you're sliding forward as you brake and backwards as you accelerate. In a video game you get none of that, and you become detached. I think there will be plenty of valid VR experiences, once we figure out what's the right fit. Windows Mobile took the desktop and put it on a phone, it wasn't until Android and iOS came out that we figured that smartphones are really actually quite cool, provided you build the right interface. We now know there are just things you can't (or shouldn't) do on a mobile device compared to a desktop, and we build around that. I don't know if playing an FPS in VR is ever going to feel "right" because you're entering the uncanny valley of "I'm walking up the stairs but I'm not actually". At least on Counter Strike you're just pushing buttons on a keyboard and looking at a small rectangle, so your brain knows it's not actually supposed to be happening.
  - 8y ago
nomel 8y ago
To be fair, you trick an optical sensor with a gust of wind and a plastic bag. Ideally all sensors would be redundant, with poorer and poorer (but still safe) performance as each is removed.

This is exactly why we need to escape the mentality of sensor fusion and make our robots reliable just based on optical data. Then they would be susceptible to attacks that would affect humans too.

ent 8y ago

I took this as an argument on why we shouldn't rely on one type of sensor alone, as it's easier to fool with tricks like this.

pfortuny 8y ago

Afaik equilibrium is a matter of at least vision and audition, is it not?

guav_ 8y ago

Humans with issues in their vestibular system located in the ear can't really use their optical data streams to compensate. When you loose your gyros time to get a walking stick or a beautiful nurse to lean on.

StavrosK 8y ago
Can humans with issues in the optical stream use the gyros? I find it harder to balance with my eyes closed, though obviously not impossible.
- falcolas 8y ago
  Well, the blind and sight-impaired have been walking (and running) for years, so I'd vote "yes". Even people with impressive visual hallucinations are still capable of walking. That said, vertigo (where your eyes and your inner ear are in violent disagreement as to whether you are moving) can impair movement, though you can grow accustomed to the "disagreement".
- tehbeard 8y ago
  I think the optical stream fits more of an error correction/secondary data stream in this case. The difficulty may be down to you trying to use other senses and overcompensating on the data (e.g. air pressure, which tells you something is near-ish you but not if it's about to contact you so you react)
jerf 8y ago
Interestingly, it is possible to learn to walk when you never had a vestibular system, although the gait is visibly distorted and when you know what you are looking for you can easily spot it. Those who suffer from CHARGE, and are left able to walk, can still do so despite sometimes entirely missing their vestibular system. But learning to walk with a vestibular system and then losing it later seems to be very, very difficult to recover from, if it is possible at all.

eatitraw 8y ago

Humans fuse a lot of sensory data though. For example, walking without proprioception is incredibly hard: https://philosophyandpsychology.wordpress.com/2007/10/14/living-without-proprioception/ https://philosophyandpsychology.wordpress.com/2007/10/14/liv...

noir_lord 8y ago
I have nerve damage because of a spinal problem, I walk with a slight limp not because I've lost strength but because I can't always accurately 'feel' where my foot is. It's like walking with pins and needles.

verelo 8y ago

Humans use multiple sensors together at one. Your inner ear is a great example. If you see a picture doing one thing, but your inner ear feels another, this is the senstation that causes motion sickness. I cannot find the source right now, but i recall being taught that research dating back to US airforce simulation training suggested that this might be a built in evolutionary function that is designed to make us throw up. The basic narrative being "I'm seeing something I'm not feeling, i must have eaten something that's poisoned me, i should throw up so i don't die". Edit: As a side note, this is why I'm extremely skeptical of things like VR technology in general (as a person that owns a HTC Vive). I'm not sure we're going to be able to easily convince the human brain it's OK for different sensors to feel different things, and not result in us feeling sick.

freehunter 8y ago
Re: your side note That's why even though I'm a relatively avid gamer with plenty of disposable income, I've never bought on to the VR stuff. It's not about the cost, or the rapidly improving technology, it's that I don't believe (long term) that I'm going to really enjoy many VR games. I have a racing wheel because I like to play racing games, but most of the time I end up using my controller anyway. Why? Because it doesn't feel real. The wheel ends up feeling farcical because yeah I'm cornering at 1.5G, but I don't feel a thing. In real life you're less likely to corner too fast because you can feel that you're going too fast. You're being shoved around, you're being moved side to side, you're sliding forward as you brake and backwards as you accelerate. In a video game you get none of that, and you become detached. I think there will be plenty of valid VR experiences, once we figure out what's the right fit. Windows Mobile took the desktop and put it on a phone, it wasn't until Android and iOS came out that we figured that smartphones are really actually quite cool, provided you build the right interface. We now know there are just things you can't (or shouldn't) do on a mobile device compared to a desktop, and we build around that. I don't know if playing an FPS in VR is ever going to feel "right" because you're entering the uncanny valley of "I'm walking up the stairs but I'm not actually". At least on Counter Strike you're just pushing buttons on a keyboard and looking at a small rectangle, so your brain knows it's not actually supposed to be happening.
- 8y ago

nomel 8y ago

To be fair, you trick an optical sensor with a gust of wind and a plastic bag. Ideally all sensors would be redundant, with poorer and poorer (but still safe) performance as each is removed.