4 ms·

I wonder why don't they just program them to ask and follow directions from people. It has served humanity well for almost all of its existence. Of course, it g

by seu 4y ago

I wonder why don't they just program them to ask and follow directions from people. It has served humanity well for almost all of its existence. Of course, it goes against the current trend in technology, which is all about giving individuals complete independence from other individuals, so they can ignore them as much as they want... but hell, wouldn't it make a robot look _way_ more trustworthy and friendly, if you see first that they trust _you_, by asking for your help?

trailingComma 4y ago
I'm sure this would result in a lot of robots being told where to go.
RugnirViking 4y ago
Natural language processing just isn't far enough along to understand human direction giving. source: For my masters degree I wrote a paper on the grounding problem (linking words said to physical objects), specifically the spatial reasoning component ("please give me the book on the table in the corner). We were able to get very simple commands to work, but anything remotely complicated, especially with a temporal aspect or things that can't immediately be seen (keep going until you see the red building, then turn left) is the realm of future research
- amelius 4y ago
  Did you see Google's robocall demo?
- dTal 4y ago
  That's disappointing! SHRDLU was such an impressive demo, all the way back in 1970 - was that research a dead end, then? You'd think we'd have progressed.
  - RugnirViking 4y ago
    A large part invoves the transition between a simulation world, in which the properties of objects is known for certain (size, location, properties like colour) to a world where these properties have to be ascertained from computer vision. Relative size, position, etc can be tricky to determine as they aren't just classification problems but require segmentation and spatial reasoning. Especially taking into account any difference in perspective between the user and the robot, as well as context (what does "above" mean on a map vs in a tower of blocks, or what does "left" mean if you are facing me)
- Closi 4y ago
  Doesn't have to be natural language to get assistance - it could be as simple as 'Can you point to the direction on my screen'. Starship Robotics has a pilot in Milton Keynes in the UK for their delivery robots, and came across the problem of pedestrian crossings where you need to push a button to change the traffic lights & cross. Rather than engineering a solution that required changing the crossings or putting some complex robot-arm on to push the button, they just play a pre-recorded message begging humans to press the button for it. It's easy to devise complex solutions which you can't implement - but it sometimes involves a bit more creativity when you constrain yourself to simple and implementable solutions!
- gwern 4y ago
  > Natural language processing just isn't far enough along to understand human direction giving. https://arxiv.org/abs/2207.04429 https://arxiv.org/abs/2207.04429 https://arxiv.org/abs/2204.01691#google https://arxiv.org/abs/2204.01691#google https://arxiv.org/abs/2207.05608#google https://arxiv.org/abs/2207.05608#google https://arxiv.org/abs/2207.05608#google https://arxiv.org/abs/2207.05608#google https://arxiv.org/abs/2109.01115 https://arxiv.org/abs/2109.01115 etc
RalfWausE 4y ago
"You only have to go this way over there, over there where the shop of the old... what was his name?... you know, i can guarantee you he makes the best chillie... ah, where was i? Yeah, now, go this way..." (etc, etc) This is a fairly accurate description of a "guide" i got from a nice guy after asking him where my hotel may be... and now imagine the software that must be involved to decode THIS...

by seu 4y ago

trailingComma 4y ago

I'm sure this would result in a lot of robots being told where to go.

RugnirViking 4y ago

Natural language processing just isn't far enough along to understand human direction giving. source: For my masters degree I wrote a paper on the grounding problem (linking words said to physical objects), specifically the spatial reasoning component ("please give me the book on the table in the corner). We were able to get very simple commands to work, but anything remotely complicated, especially with a temporal aspect or things that can't immediately be seen (keep going until you see the red building, then turn left) is the realm of future research

amelius 4y ago
Did you see Google's robocall demo?
dTal 4y ago
That's disappointing! SHRDLU was such an impressive demo, all the way back in 1970 - was that research a dead end, then? You'd think we'd have progressed.
- RugnirViking 4y ago
  A large part invoves the transition between a simulation world, in which the properties of objects is known for certain (size, location, properties like colour) to a world where these properties have to be ascertained from computer vision. Relative size, position, etc can be tricky to determine as they aren't just classification problems but require segmentation and spatial reasoning. Especially taking into account any difference in perspective between the user and the robot, as well as context (what does "above" mean on a map vs in a tower of blocks, or what does "left" mean if you are facing me)
Closi 4y ago
Doesn't have to be natural language to get assistance - it could be as simple as 'Can you point to the direction on my screen'. Starship Robotics has a pilot in Milton Keynes in the UK for their delivery robots, and came across the problem of pedestrian crossings where you need to push a button to change the traffic lights & cross. Rather than engineering a solution that required changing the crossings or putting some complex robot-arm on to push the button, they just play a pre-recorded message begging humans to press the button for it. It's easy to devise complex solutions which you can't implement - but it sometimes involves a bit more creativity when you constrain yourself to simple and implementable solutions!
gwern 4y ago
> Natural language processing just isn't far enough along to understand human direction giving. https://arxiv.org/abs/2207.04429 https://arxiv.org/abs/2207.04429 https://arxiv.org/abs/2204.01691#google https://arxiv.org/abs/2204.01691#google https://arxiv.org/abs/2207.05608#google https://arxiv.org/abs/2207.05608#google https://arxiv.org/abs/2207.05608#google https://arxiv.org/abs/2207.05608#google https://arxiv.org/abs/2109.01115 https://arxiv.org/abs/2109.01115 etc

RalfWausE 4y ago

"You only have to go this way over there, over there where the shop of the old... what was his name?... you know, i can guarantee you he makes the best chillie... ah, where was i? Yeah, now, go this way..." (etc, etc) This is a fairly accurate description of a "guide" i got from a nice guy after asking him where my hotel may be... and now imagine the software that must be involved to decode THIS...