Boston Spot - now it's released!

Macropos

Macropos

Tourist
BANNED USER
Yes, the public will be able to get one soon, at a hefty sum of course, but could they be worth it! Tell me this thing isn't the sexiest thing on four legs?!? Well, maybe not, but bring some fursuit artists into it, just put a Fleshlight in the right place(s) and add some proper programming, and you get what i mean. Not only can you get the four-legged fantasy animal lover of your dreams, it will be SENTIENT and OBEDIENT to your every command! Well, not actually sentient, but anyway.... i think bestiality itself might be about to enter a whole new phase now thanks to these engineers!

 
This thing would probably cost more then it would cost to care for 3 dogs over there lifetime all at one time. It does look cool though.
 
raks said:
man creates robot and what is the first thing he puts it to use for?

PORN!
Click to expand...
It's long been known that there are *EXACTLY* two uses for new technology: War and porn. And even if it's dedicated to one or the other, which one it'll actually get used for first is open to debate...

Once the warriors and the wankers have started the exploration of how to put it to use, the rest of us MIGHT get to tinker with it and see if there's some other use it can be put to.

Been true all through history - why should it change now?
 
They'll never be able to replace living beings with machines. No matter what they do it won't be the same.
 
Wouldn't say i love them yet, but there's definitely potential here. Can't wait to see what some creative fursuit constructors can do with them....
 
"What do you get when robotics engineers have too much time on their hands?"

It's not like they just said "dance" and there's AI deciding how to make each of those movements. They painstakingly programmed every movement just because. Yes, there's likely a level of AI being used to adjust balance servos to keep them upright through the motions, but the motions are still highly choreographed.

You can tell from the stiffness that they were programmed rather than using any kind of mocap or mocap based machine learning.
 
Macropos said:
Wouldn't say i love them yet, but there's definitely potential here. Can't wait to see what some creative fursuit constructors can do with them....
Click to expand...

Boy, that four-legged yellow one with the extensible neck/head thing positively screams "PIERSON'S PUPPETEER!!!" to me...

(Although to be really accurate, it needs to convert one of the legs into another head)

And if you need to ask "What's a Pierson's Puppeteer?", two things: First, you aren't qualified to appreciate this video, and second, go find yourself a copy one of Larry Niven's various "Known Space" collections, or NIven and Lerner's "Fleet of Worlds" series.

dartel said:
You can tell from the stiffness that they were programmed rather than using any kind of mocap or mocap based machine learning.
Click to expand...

Tell ya what, guy - When you're capable of re-creating this video, I'll consider you qualified to piss and moan about how stiff it looks. Until then, get back in your lane and enjoy it as the EXTREMELY well done proof-of-concept demo that it is.
 
UR20Z said:
Tell ya what, guy - When you're capable of re-creating this video, I'll consider you qualified to piss and moan about how stiff it looks. Until then, get back in your lane and enjoy it as the EXTREMELY well done proof-of-concept demo that it is.
Click to expand...
Umm, I have a degree in animation, so I have created things that look that stiff before. There's a noticeable difference between hand animated (programmed robotics) and mocap performances. Even most games these days use mocap because it's vastly smoother and more realistic.
 
dartel said:
Umm, I have a degree in animation, so I have created things that look that stiff before. There's a noticeable difference between hand animated (programmed robotics) and mocap performances. Even most games these days use mocap because it's vastly smoother and more realistic.
Click to expand...

Drawing - whether with pencil or computer - something dancing is a whole shitload easier to make look smooth and/or realistic than getting a multi-hundred-pound robot operating semi-autonomously to dance, however stiff or unnatural the result might look. This is, I have little doubt, the best that *ANYBODY ON THE PLANET* has managed to do so far. I'd be willing to bet that even if somebody handed you a "blank check" card, and gave you a 10 year deadline, you wouldn't be able to do any better. Nor would I.
 
UR20Z said:
Drawing - whether with pencil or computer - something dancing is a whole shitload easier to make look smooth and/or realistic than getting a multi-hundred-pound robot operating semi-autonomously to dance, however stiff or unnatural the result might look. This is, I have little doubt, the best that *ANYBODY ON THE PLANET* has managed to do so far. I'd be willing to bet that even if somebody handed you a "blank check" card, and gave you a 10 year deadline, you wouldn't be able to do any better. Nor would I.
Click to expand...
And all for no reason or use. Hence "what do you get when robotics engineers have too much time on their hands?".

Programming it to dance doesn't "show off what the technology can do", just how much time the engineers spent programming (and reprogramming it when a particular move made it over reach its parameters and fall over).

A real time mocap would actually be a better demonstration of its potential range of motion than a preprogrammed choreographed performance. Then they could show off if the servo stabilizers are capable of matching, say an acrobat, and/or get a sense of where its redline for failure occurs. It would also show off potential for fluidity of movement better than trying to manually recreate a fluid movement through animation.
 
dartel said:
And all for no reason or use. Hence "what do you get when robotics engineers have too much time on their hands?".

Programming it to dance doesn't "show off what the technology can do", just how much time the engineers spent programming (and reprogramming it when a particular move made it over reach its parameters and fall over).

A real time mocap would actually be a better demonstration of its potential range of motion than a preprogrammed choreographed performance. Then they could show off if the servo stabilizers are capable of matching, say an acrobat, and/or get a sense of where its redline for failure occurs. It would also show off potential for fluidity of movement better than trying to manually recreate a fluid movement through animation.
Click to expand...

Easy to criticize when you have only the vaguest idea what's involved, ain't it? Didja know it took the better part of 10 years worth of development effort to get "Big Dog" (one of the earlier iterations of "Spot") to the point where it was capable of simply walking from point A to point B over griddle-flat terrain without falling over? And another couple of years to get it to the point where it's effectively impossible to knock it over, short of hitting it hard enough with a large enough mass to smash it to pieces? Last time I was paying close attention to their progress, it was impossible for a human to do anything more to Big Dog than nudge it slightly off course, even when it was on the move over terrain a human would be lucky to drag himself out of with "only" a broken ankle.

Regardless of how "stiff" you might think it looks, this is some *SERIOUSLY* high-level shit on display.

"For no use", eh?
Dunno what the specs on the two-legged ones or Spot are, but Big Dog was rated to be able to haul two disabled humans in full battle gear over any terrain that wasn't liquid, at a better speed than what a Jeep in rough country can manage, while keeping a full wine glass balanced on top of them without spilling any of the contents or dropping the humans, and with no more human control than "Meet the medics that'll load you here, bring what they load onto you to the hospital over there - GO!"

That's "no use"?
 
UR20Z said:
That's "no use"?
Click to expand...
That's not what is being displayed in that video. I'm not criticizing the tech or the advancements in the tech, I'm criticizing the time spent programming it for this dance video. The engineers, not the machine. Again, there's better demonstrations of its capabilities than spending tons of time choreographing a dance off.
 
dartel said:
That's not what is being displayed in that video. I'm not criticizing the tech or the advancements in the tech, I'm criticizing the time spent programming it for this dance video. The engineers, not the machine. Again, there's better demonstrations of its capabilities than spending tons of time choreographing a dance off.
Click to expand...

It's a demo. Your basic project "Dog and Pony show" that goes on with EVERY high-tech project so that the dollars to continue the project keep flowing. Perhaps doubling as a sort of "progress report" to those who have already pumped in dollars that have already been spent. Sure - it could be your standard 3000 page report with spreadsheets and flowcharts and what-not, but how impressive would that have been compared to this? Can you imagine how that'd be likely to go over? "Yeah, great - you've spent 15 years and as many million dollars, and all you've got to show for it is this three-ring binder full of paper and ink?" I see only one outcome: Project canceled.

This video at least shows definite, concrete, easily seen, easily measured progress towards the final goal, in a way that anybody who isn't a complete and total "thud" can easily see, understand, and appreciate. Even if that goal is nothing more than getting the darned things to dance - never mind that this is damn-near miraculous when stacked up against the state of the art when the project got started.

These things are semi-autonomous, able to outperform MANY alternatives at many tasks, under conditions that no human could handle, at speeds humans aren't capable of on their best day, with strength that, while fully able to rip chunks out of a concrete wall, is controlled well enough that they can pick up and juggle eggs and light bulbs, and at the end of the demo, still have uncracked eggs, and light bulbs that work.

Which doesn't even begin to address the fact that your supposed "tons of time" isn't nearly as much as you're probably thinking. Back in the day, yes - it did indeed take hours, if not days or even weeks, of time to program a 3-axis machine to do a motion that took a whole 4-5 seconds to perform. With the stuff they've been doing at Boston Dynamics, I'm betting that the "coding" for this looked more along the lines of "Move left foot from start to position A while moving right arm from start to position B", and the 'bots "just do it" - none of the "break it down into a bazillion steps" malarkey it used to take to make something happen "the old way".

*THAT* capability is truly what's on display here, whether you realize it or not.

"No use"? A waste of time? I think not.
 
dartel said:
"What do you get when robotics engineers have too much time on their hands?"

It's not like they just said "dance" and there's AI deciding how to make each of those movements. They painstakingly programmed every movement just because. Yes, there's likely a level of AI being used to adjust balance servos to keep them upright through the motions, but the motions are still highly choreographed.

You can tell from the stiffness that they were programmed rather than using any kind of mocap or mocap based machine learning.
Click to expand...

Not to mention it would be cleaner if they just had an animation artist design the movements. That difference is a hefty part of what sets apart some old and new animatronics' lifelikeness in amusement parks.
I can imagine there's a decent amount of limitation needed to take into account given that it's physical and has to maintain balance though, so I understand why they'd hardcode them in for the time being.
 
You must log in or register to reply here.
Back
Top