International Skeptics Forum

International Skeptics Forum (http://www.internationalskeptics.com/forums/forumindex.php)
-   Science, Mathematics, Medicine, and Technology (http://www.internationalskeptics.com/forums/forumdisplay.php?f=5)
-   -   Dancing machine (http://www.internationalskeptics.com/forums/showthread.php?t=348688)

Checkmite 29th December 2020 03:11 PM

Dancing machine
 
Courtesy of Boston Dynamics, naturally:

YouTube Video This video is not hosted by the ISF. The ISF can not be held responsible for the suitability or legality of this material. By clicking the link below you agree to view content from an external website.
I AGREE

Mike Helland 29th December 2020 03:16 PM

Amazing.

Galaxie 30th December 2020 01:21 PM

Wow, they've come a long way! I wonder how many takes it took?

theprestige 30th December 2020 01:34 PM

Quote:

Originally Posted by Galaxie (Post 13342013)
Wow, they've come a long way! I wonder how many takes it took?

What even is a "take" anymore? My guess is they've been trying out the dancing thing for weeks, and recording it on their phones, and this just happened to be a good clip so they bunged it over to the PR department for dissemination.

Thermal 30th December 2020 01:39 PM

Absolutely awesome!

deverett 30th December 2020 01:44 PM

This is what billions buys you. Nothing useful.

rjh01 30th December 2020 02:36 PM

It shows the potential of these robots.

Galaxie 30th December 2020 03:15 PM

Quote:

Originally Posted by theprestige (Post 13342027)
What even is a "take" anymore? My guess is they've been trying out the dancing thing for weeks, and recording it on their phones, and this just happened to be a good clip so they bunged it over to the PR department for dissemination.

A "take" would be get everything set up, start the robots' programs plus the camera(s), and hope nothing goes wrong before you get the footage. Phone recordings? Maybe, but I think they're using steadicams.

jrhowell 30th December 2020 03:26 PM

The first two thirds seems to be a single take, but later on there are a lot of cuts where the light coming in through the windows changes showing that it wasn’t done all at once. Still very impressive!

theprestige 30th December 2020 03:39 PM

Quote:

Originally Posted by deverett (Post 13342045)
This is what billions buys you. Nothing useful.

Haters gonna hate. This is a whole new field of applied engineering. How many billions in today's dollars do you think were spent getting heavier-than-air flight off the ground? There were literally decades of nothing, before it all came together for the first time.

rockinkt 30th December 2020 03:46 PM

Rudimentary motion in machinery induced by specific soundwaves is nothing new. Obviously they just took that same principle and expanded on it. To suggest that this is something that requires some sort of AI is silly.

bluesjnr 30th December 2020 03:50 PM

I've just watched a documentary about the Stark corporation and I'm supposed to be impressed?

Thermal 30th December 2020 03:53 PM

Quote:

Originally Posted by rockinkt (Post 13342173)
Rudimentary motion in machinery induced by specific soundwaves is nothing new. Obviously they just took that same principle and expanded on it. To suggest that this is something that requires some sort of AI is silly.

Yeah...yeah it would. Did you, um...hear any suggestions like that?

bluesjnr 30th December 2020 03:53 PM

Quote:

Originally Posted by rockinkt (Post 13342173)
Rudimentary motion in machinery induced by specific soundwaves is nothing new. Obviously they just took that same principle and expanded on it. To suggest that this is something that requires some sort of AI is silly.

To be fair and to back off, a little, from my original response - where are you seeing the suggestion that what we've watched "requires some sort of AI"?

Mike! 30th December 2020 04:16 PM

Was there more than one humanoid robot or did they do some kind of old fashioned 60's sitcom video magic to make the two dance together?

Mike Helland 30th December 2020 04:19 PM

Quote:

Originally Posted by bluesjnr (Post 13342189)
To be fair and to back off, a little, from my original response - where are you seeing the suggestion that what we've watched "requires some sort of AI"?

These things are AI wonders.

With a remote control car, you just press forward, and the wheels move the car forward.

When you tell one of these things to move forward, it figures out how to do that.

That simple looking dance move where Spot's legs are crossed requires a mountain of computation, and AI is part of that.

https://www.youtube.com/watch?v=R-PdPtqw78k

Trebuchet 30th December 2020 04:26 PM

I find myself wanting the dog thing to have two heads/necks and one back leg. Anybody know why?

pdp 30th December 2020 04:37 PM

Quote:

Originally Posted by Trebuchet (Post 13342228)
I find myself wanting the dog thing to have two heads/necks and one back leg. Anybody know why?

Puppeteer?

theprestige 30th December 2020 04:42 PM

Quote:

Originally Posted by Mike Helland (Post 13342217)
These things are AI wonders.

With a remote control car, you just press forward, and the wheels move the car forward.

When you tell one of these things to move forward, it figures out how to do that.

That simple looking dance move where Spot's legs are crossed requires a mountain of computation, and AI is part of that.

https://www.youtube.com/watch?v=R-PdPtqw78k

I'm pretty sure Boston Dynamic's robots move according to some elegantly simple algorithms for maintaining hysteresis in an unstable system, not mountains of AI computation. Maybe mountains of AI computation went into developing the algorithms?

In fact I think one of their earliest learnings was that it was actually easier to make an unstable robot that constantly shifted its footing to keep from falling over, than it was to make a robot that could just stand still like humans do.

Mike Helland 30th December 2020 05:00 PM

Quote:

Originally Posted by theprestige (Post 13342256)
I'm pretty sure Boston Dynamic's robots move according to some elegantly simple algorithms for maintaining hysteresis in an unstable system, not mountains of AI computation. Maybe mountains of AI computation went into developing the algorithms?

In fact I think one of their earliest learnings was that it was actually easier to make an unstable robot that constantly shifted its footing to keep from falling over, than it was to make a robot that could just stand still like humans do.

Check at the Adam Savage video just before 5 minutes, here's a timestamp:

https://youtu.be/R-PdPtqw78k?t=292

rjh01 30th December 2020 07:27 PM

Quote:

Originally Posted by Mike Helland (Post 13342269)
Check at the Adam Savage video just before 5 minutes, here's a timestamp:

https://youtu.be/R-PdPtqw78k?t=292

That brings up the question, how much autonomy did the robots have doing the dance? Were they listening to the music and acting on it? Or were humans listening to the music themselves and then telling the robots to move in certain ways?

Mike Helland 30th December 2020 07:33 PM

Quote:

Originally Posted by rjh01 (Post 13342365)
That brings up the question, how much autonomy did the robots have doing the dance? Were they listening to the music and acting on it? Or were humans listening to the music themselves and then telling the robots to move in certain ways?

Replace the robots with humans, and what do you do?

Someone choreographed the dance routine. The camera is probably on a mobile robot too.

Now teach the robots the dance routine.

Like the humans, these robots aren't told when to move what muscle and by how much. They are told what kind of maneuver to do (a leg lift), and their software works out the best way to do that in their environment.

I suspect if a 50 gallon barrel rolled through the dance floor, the robots would sense it, avoid it, and get back to dancing. Based on all Boston Dynamic other videos.

Trebuchet 30th December 2020 07:34 PM

Quote:

Originally Posted by pdp (Post 13342246)
Puppeteer?

Got it in one!

Quote:

Originally Posted by theprestige (Post 13342256)
I'm pretty sure Boston Dynamic's robots move according to some elegantly simple algorithms for maintaining hysteresis in an unstable system, not mountains of AI computation. Maybe mountains of AI computation went into developing the algorithms?

In fact I think one of their earliest learnings was that it was actually easier to make an unstable robot that constantly shifted its footing to keep from falling over, than it was to make a robot that could just stand still like humans do.

That's pretty much how human beings do it. Totally unstable while walking, just keep moving. Standing still is pretty much harder for a human; and still extremely dynamic.

theprestige 30th December 2020 07:36 PM

Quote:

Originally Posted by rjh01 (Post 13342365)
That brings up the question, how much autonomy did the robots have doing the dance? Were they listening to the music and acting on it? Or were humans listening to the music themselves and then telling the robots to move in certain ways?

Why do you assume the robots were dancing to the music? What if they were just moving while music was playing?

theprestige 30th December 2020 07:38 PM

Quote:

Originally Posted by Mike Helland (Post 13342269)
Check at the Adam Savage video just before 5 minutes, here's a timestamp:

https://youtu.be/R-PdPtqw78k?t=292

Adam Savage does not work for Boston Dynamics.

Body of work says you're extrapolating without support from an ambiguous source.

Mike Helland 30th December 2020 07:40 PM

Quote:

Originally Posted by theprestige (Post 13342379)
Adam Savage does not work for Boston Dynamics.

Body of work says you're extrapolating without support from an ambiguous source.

Well, I've seen plenty of Boston Dynamic videos of them pushing robots around with hockey sticks. Apparently that's a big no-no now.

Adam Savage got one of their Spot robots and that video I linked does a pretty good job at explaining some of its basic autonomy.

crackers 30th December 2020 08:18 PM

Well, however they do it, itís pretty darn amazing.

rjh01 31st December 2020 03:41 AM

Quote:

Originally Posted by theprestige (Post 13342376)
Why do you assume the robots were dancing to the music? What if they were just moving while music was playing?

Watch the video. They are moving in response to the music. The only question is how much direction from humans have they had?

Robin 31st December 2020 05:24 AM

If someone had brought a set of Boston Dynamic videos back in time to 1968 and said that this is where robotics will be in the second decade of the 21st century, then we would have all been very disappointed.

dann 3rd January 2021 04:52 PM

Quote:

Originally Posted by rjh01 (Post 13342365)
That brings up the question, how much autonomy did the robots have doing the dance? Were they listening to the music and acting on it? Or were humans listening to the music themselves and then telling the robots to move in certain ways?


Amazing video, in particular when you consider how difficult it is for some humans to learn to dance. For instance, when dance instructors say, "forward, back, forward, back," when demonstrating basic steps, they say "forward" when they are already making the step forward, but many people don't understand this simple principle and consider "forward" to be a command and therefore start moving forwards when the instructor is already taking the step back.

I don't think the robots are listening and acting on it. Robots would have been more precise when following the rhythm and keeping the beat, I think. So I think engineers were listening and making the moves after having programmed the robots to emulate them. In Copenhagen, we have quite a lot of engineers learning salsa dancing. When salsa dancers talk about somebody "dancing salsa like an engineer," it isn't usually meant as praise ... :)

zooterkin 3rd January 2021 04:57 PM

Quote:

Originally Posted by rjh01 (Post 13342562)
Watch the video. They are moving in response to the music. The only question is how much direction from humans have they had?

How can you tell thatís what is happening rather than, for instance, the music has been put on afterwards to match the video?

rjh01 3rd January 2021 05:27 PM

Quote:

Originally Posted by zooterkin (Post 13345667)
How can you tell thatís what is happening rather than, for instance, the music has been put on afterwards to match the video?

In theory, they could have told someone to play music to match the robot moves, but I doubt this happened. Much easier to do it the other way round. In any case, I think it is amazing to see the robots do all of these moves.

Nakani 3rd January 2021 10:59 PM

Quote:

Originally Posted by rjh01 (Post 13342562)
Watch the video. They are moving in response to the music. The only question is how much direction from humans have they had?

That would be cool, a robot that could recognize rythm patterns in music and coordinate its movements to match. It could randomly choose from a catalogue of moves and give the impression of dancing.

I think that is within our capabilities but not what is happening here. This looks like robots performing programmed movements that change in time with a popular song.

Segnosaur 4th January 2021 12:56 AM

I for one welcome our new dancing robotic overlords.

rdaneel 4th January 2021 01:02 AM

We all better start practicing, looks like the robot apocalypse is going to be a dance off.

erlando 4th January 2021 01:16 AM

Does anyone really think this is robots autonomously listening and interpreting music on the fly and dancing?

Yeah, just no. Watch closely towards the end and observe the light coming in through the windows at different angles. This is quite clearly multiple takes implying that the music was synced in post.

The tech is impressive. But the robots aren't dancing.

Ziggurat 4th January 2021 02:07 AM

Quote:

Originally Posted by erlando (Post 13346022)
The tech is impressive. But the robots aren't dancing.

The robots are definitely dancing. It is undoubtedly a pre-determined routine, as are many human dances. What they probably are not doing is using external audio as a cue for the timing. But then, why would they? Robots can keep time perfectly internally. Humans aren't very good at that, that's part of why we mostly dance to the music.

rjh01 4th January 2021 02:53 AM

Quote:

Originally Posted by Ziggurat (Post 13346056)
The robots are definitely dancing. It is undoubtedly a pre-determined routine, as are many human dances. What they probably are not doing is using external audio as a cue for the timing. But then, why would they? Robots can keep time perfectly internally. Humans aren't very good at that, that's part of why we mostly dance to the music.

I think doing that is amazing. Programming every move, in all three robots. That is a lot of work.

erlando 4th January 2021 04:58 AM

Quote:

Originally Posted by Ziggurat (Post 13346056)
The robots are definitely dancing. It is undoubtedly a pre-determined routine, as are many human dances. What they probably are not doing is using external audio as a cue for the timing. But then, why would they? Robots can keep time perfectly internally. Humans aren't very good at that, that's part of why we mostly dance to the music.

I don't believe they are using audio at all.

zooterkin 4th January 2021 05:02 AM

Quote:

Originally Posted by rjh01 (Post 13346075)
I think doing that is amazing. Programming every move, in all three robots. That is a lot of work.

Indeed it is. Not as much work as doing what you originally suggested, that the robots were moving in response to the music.


All times are GMT -7. The time now is 01:07 AM.

Powered by vBulletin. Copyright ©2000 - 2021, Jelsoft Enterprises Ltd.
© 2015-20, TribeTech AB. All Rights Reserved.