Why is something like walking so effortless for our brains to compute but requires millions of code for a robot...

Why is something like walking so effortless for our brains to compute but requires millions of code for a robot? Is science approaching AI the wrong way?

Other urls found in this thread:

ai.mit.edu/projects/leglab/robots/robots.html
youtube.com/watch?v=n3WhwQ0S-vM
github.com/osudrl/atrias/blob/d874519368fbd5e369b2f006b9151f664da58f66/software/atrias_controllers/atc_slip_walking/src/ATCSlipWalking.cpp
youtube.com/watch?v=VBDysRlrfcY
github.com/osudrl/atrias/blob/d874519368fbd5e369b2f006b9151f664da58f66/software/atrias_controllers/atc_slip_walking/include/atc_slip_walking/ATCSlipWalking.hpp
en.m.wikipedia.org/wiki/Spinal_locomotion
twitter.com/AnonBabble

>effortless for our brains
It's not.

>t. Leglet

Nature had millions of years to work it out. We've been at it for a couple decades.

>biolets

google "globus pallidus" "ventral striatum" "cerebellum" "parkinsons disease" "psychomotor agitation"

motor functions are extremely resource-intensive and fragile, taking up large amounts of our brain's space and metabolism
it seems "effortless" to us because those parts of our brains have existed for as long as vertebrates have, and unlike robots have been time-tested for hundreds of millions of years

>Is science approaching AI the wrong way?
No. Science is actually emulating brains in AI precisely for this reason. Look up neural nets.

thinks what the human brain does is effort less


LOL

Brainlet detected

I'm getting mixed messages here

Read a fucking book on AI ffs

The only reason robots cant walk BIPEDALLY is because they are fucking heavy compared to our meatsuits.

That's not what a neural net is.

>> requires millions of code for a robot?
lol no. You can do basic bipedal walking in less than 10 lines.

life took billions of years to create walking.

Took us a while to pinpoint how it worked. We know this now.

this might be the dumbest fucking thing ive ever read

on a perfectly even, flat surface with decent surface friction coefficient, with a perfectly balanced body free from collisions and at a steady pace, sure. But our brains do so much more

not just that though

on a perfectly even, flat surface with decent surface friction coefficient, with a perfectly balanced body free from collisions and at a steady pace using a robot built to do specifically that using simplified high level code to make it easy for even your grandma to make the commercially available robot walk

this user is so fucking retarded beyond repair

I'm completely serious. Basic bipedal walking is not that hard. Pic related
ai.mit.edu/projects/leglab/robots/robots.html
youtube.com/watch?v=n3WhwQ0S-vM

>Basic bipedal walking is not that hard.

maybe for a brainlet like you

You under estimate the number of operations that your brain does for that and the physiological and biological adaptation necessary for this miracle.

of course, but we don't need all of them to implement basic walking. Here's an example of a walking controller for the ATRIAS robot, the predecessor to the most advanced biped robot in the world:
github.com/osudrl/atrias/blob/d874519368fbd5e369b2f006b9151f664da58f66/software/atrias_controllers/atc_slip_walking/src/ATCSlipWalking.cpp
It's not a million lines long, and really once you get down to it, it's just a simple state machine of several different controllers
Here is ATRIAS in action
youtube.com/watch?v=VBDysRlrfcY

I am not convinced by Atrias. Put it on sand.

>if we tether it to the ceiling so it can't fall over we can do waking in 10 lines of code
Unless you count loading libraries as a line of code.

Don't forget the included header file.
github.com/osudrl/atrias/blob/d874519368fbd5e369b2f006b9151f664da58f66/software/atrias_controllers/atc_slip_walking/include/atc_slip_walking/ATCSlipWalking.hpp

it's not hanging from the ceiling brainlet

How can you comment on something you know nothing about?
Why is Veeky Forums full of ignorant teenagers?

basic bipedal walking isn’t substantial, also no

nigger we have dedicated hardware for it, it's much more complex than some control system.

The cerebellum is a pretty substantial dedicated motor processor. Discrete "Lines of code" do not easily map to a system that is like an analog computer in many ways either, with behavior better modeled in terms of dynamical system theory and chaos perhaps.

It takes a child years to learn how to walk properly and people still fall down all the time and snap their bones.

en.m.wikipedia.org/wiki/Spinal_locomotion

>Why is Veeky Forums full of ignorant teenagers?
bcoz mods are fckn useless

>walking so effortless
... that is takes years to achieve.
gtfo

because walking is so effortless and unconscious we don't have a good understanding of what's going on.

>a billion neurons with a trillion dendrites

Vs

>3 lines in keras

t. complet