Boston Dynamics Adds Wheels to Atlas Humanoid (therobotreport.com) 58
Boston Dynamics has equipped its Atlas humanoid with wheels, allowing the robot to rapidly roll across flat terrain while still using its articulated legs to navigate obstacles and uneven surfaces. "The 'experiment' combined classical model-based motor control with whole-body reinforcement learning to maintain dynamic stability across unstructured, real-world terrain," reports The Robot Report. From the report: In addition to learning how to move with wheels for feet, Atlas also performs pirouettes and various aggressive actuation maneuvers. This demonstrates that Atlas can smoothly switch between fast rolling on flat terrain and articulated legged motion over obstacles or irregular surfaces.
[Andy Wu, controls engineer for Atlas at Boston Dynamics] trained the system in sim. The system learned all of the behaviors in sim and executed them in real life, with surprising accuracy, according to Wu. Wu said converting Atlas' locomotion style was "wonderfully simple." A single custom adapter mounts an off-the-shelf scooter rim onto Atlas' existing ankle actuator, and that's the only new part in the whole build. Because Atlas was designed from the start to be produced at high volume and to be easy to service, Wu said the retrofit was about as involved as swapping a wheel on a bicycle. The software side was just as straightforward, he said. Each wheel is modeled as an "ankle" that can spin continuously, so all Wu had to do was take the baseline robot model and swap the feet out for wheels.
The next question that Wu intends to answer is the policy response to failed actuators. This is key to safety and to understanding how the system will respond to a system failure. The primary takeaway for Wu and the controls engineering team at Boston Dynamics is to understand control synergy. The video demonstrates that merging classical motor controls (for baseline stability) with whole-body RL (for reactive adaptation) yields superior real-world agility compared to using either method alone.
[Andy Wu, controls engineer for Atlas at Boston Dynamics] trained the system in sim. The system learned all of the behaviors in sim and executed them in real life, with surprising accuracy, according to Wu. Wu said converting Atlas' locomotion style was "wonderfully simple." A single custom adapter mounts an off-the-shelf scooter rim onto Atlas' existing ankle actuator, and that's the only new part in the whole build. Because Atlas was designed from the start to be produced at high volume and to be easy to service, Wu said the retrofit was about as involved as swapping a wheel on a bicycle. The software side was just as straightforward, he said. Each wheel is modeled as an "ankle" that can spin continuously, so all Wu had to do was take the baseline robot model and swap the feet out for wheels.
The next question that Wu intends to answer is the policy response to failed actuators. This is key to safety and to understanding how the system will respond to a system failure. The primary takeaway for Wu and the controls engineering team at Boston Dynamics is to understand control synergy. The video demonstrates that merging classical motor controls (for baseline stability) with whole-body RL (for reactive adaptation) yields superior real-world agility compared to using either method alone.
One day... (Score:2, Interesting)
How long till all of this is weaponized? Robots hunt you from building to building. Picture this but armed and less friendly looking.
Re: (Score:2)
How long till all of this is weaponized? Robots hunt you from building to building. Picture this but armed and less friendly looking.
They've tried it, but turns out it's too easy to trick them into breaking into dance [youtu.be].
Re: (Score:2)
Dalek tactics. Just run upstairs?
Before someone says it. flying Daleks was just one of the many travesties of this new fangled Doctor Who.
Tom Baker forever!
Re: One day... (Score:2)
Re: (Score:2)
We are squishy and full of lots of complex, delicate biology, but we can still kick ass remarkably well and be regularly defeated.
What we have that robots don't, is self repair, amongst many other things.
In my darker day dreams I see we abandon mechanisms and lean into biology to grow robots. Or create golems, bits of biology and mechanics mashed together.
If Elon's neuro link figures out how to write data to the brain, expect pets that can talk to you, take instructions and a full on planet of the apes, sim
Re: (Score:2)
"Elevate!" [Re:One day...] (Score:2)
They were implied as being able to fly earlier, but the one you're thinking of, the first time they actually were shown on screen flying, is "Remembrance of the Daleks":
https://www.youtube.com/watch?... [youtube.com]
Re: (Score:2)
Yes that's McCoy, he's post the bonkers of Baker. He is also a far better Radaghast. He nailed that.
Re: (Score:2)
Re: (Score:2)
Re: (Score:2)
It already happened. Back in 2017, I was asked -- and declined -- to help build such a system with a basic "move toward face, shoot if face" directive. It was intended to murder anyone indiscriminately that approached within its designated range (approved people would have a way to pause the system to come through). I followed that company's progress for a few years, and it was scary the vids they would put out... and then they went dark when bought up by undisclosed someone. Those sorts of things have onl
Re: One day... (Score:2)
I remember when the Internet was all abuzz about the 1/4 scale predator drone you could buy for a little over a grand which could fly waypoints and take pictures. I think those guys became defense contractors.
Re: One day... (Score:2)
A T-1000 on wheels could have caught young John Connor on his dirt bike, he almost has him as it was :)
Re: (Score:2)
Re: (Score:2)
Ad a couple more legs and a backpack; you've got a Tachikoma!
Re: (Score:2)
How long till all of this is weaponized? Robots hunt you from building to building. Picture this but armed and less friendly looking.
Dunno why everyone thinks Terminator is prophetic. If you want to see robots hunting people down, there's plenty of footage of that coming out of the Ukraine already, and none of it involves humanoid robots. If you want to know what getting hunted down by a robot will be like in the future, imagine a swarm of mosquitoes coming for you, except instead of trying to extract a little blood, they're trying to inject a little cyanide. No self-respecting military is going to buy expensive human-sized killbots w
Dexterity (Score:3, Insightful)
They should be focussed on hand dexterity. I mean, we need robots to do things like farm-work, repetitive factory, and warehouse work. Things like that. These robots are good for dancing and that's about it. I have yet to see handily it do something of commercial value. SO far the only thing I've seen the robots do are gimmicks like dancing. Tele-operated robots doing housework, which is the dumbest use-case. Anyone who can buy a robot can afford a maid for one tenth the cost. Show me a robot delivering packages or something instead.
Re: (Score:2)
Yes, our hands, feet and brain look like the pinnacles of our evolution and the hardest to replicate in technology. Feet, even my sideshow bobs are incredibly complex and our technical evolution would be impossible without our clever hands.
Robot industry needs an environment built for robots, we see that on a car production line, I guess. There's a new line of fasteners and handles and tools to be made that are manageable with robot and human hands, for when the robot breaks down. Maybe something like a Dre
Re: (Score:2)
Humanoid hands can hold a Dremel just fine.
Re: (Score:2)
Are you sure about that? Dremels have a lot of vibration when in use.
Re: (Score:2)
Sorry, I didn't make my point well.
I suggest we need a new set of hand tools and fastenings that both robots and human can use. Rather than entropy of two sets. Why? I get hung up on redundancy in design. The dark factories in China, highly automated, are literally dark, not lit, they are not environments for humans. Robot and human industry is diverging.
Or maybe don't bother with robot hands?
Robot hands are unlikely to be as sophisticated and adaptable as human ones, we humans just keep getting better with
Re:Dexterity (Score:4, Funny)
They should be focussed on hand dexterity. I mean, we need robots to do things like farm-work, repetitive factory, and warehouse work.
You don't need hand dexterity. You can solve farming by mounting a machine gun on the robot and exterminating humans. There no more farming required.
Re: Dexterity (Score:1)
Elon says to thank you for your support.
Thank you.
Re: (Score:2)
Wrong billionaire:
"Bill Gates: Within 10 years, AI will replace many doctors and teachers—humans won’t be needed ‘for most things’"
Re: (Score:1)
Feel free to make one yourself and show them how it's done instead of being a little slashcunt
Thanks for your input, NPC.
Got any more canned dialogue to spew?
Legs never made sense (Score:2, Insightful)
Like much of the current technological pretense of progress, the legged Atlas never made sense as some sort of potential mass manufacturing/logistics mainstay. There's extremely few situations in which wheels or rails aren't superior and plenty still where any locomotion is unnecessary.
The "our main customer is our investors" contemporary tech companies are working with humanoids at all because the coked out investor class instinctively loves the vibe of a new kind of slave shaped investment vehicle and lac
Re: (Score:2)
Re: (Score:2)
The wheels are just barely possible for evolution to create. It's happened, but only among microbes. What's impossible is the roads and smooth floors. (Well, except indirectly, by having people create them.)
Re: (Score:2)
Re: (Score:2)
Yeah, and tanks and jeeps don't need smooth roads. But they are a lot less efficient. As for "Mars rovers", the problem is we don't have a good engineering of legs yet. Wheels are easier to build. (Just look at the problems of designing a good hand for a robot to get an idea. They've tried all sorts of approaches.)
Actually, a bipedal gait is a difficult problem. Quadrupeds are much easier, but they don't work as way in spaces adapted for people.
Re: (Score:2)
Re: (Score:3)
Like much of the current technological pretense of progress, the legged Atlas never made sense as some sort of potential mass manufacturing/logistics mainstay. There's extremely few situations in which wheels or rails aren't superior and plenty still where any locomotion is unnecessary.
To the contrary. In most locations legs are better.
We're just accustomed to environments that have been reshaped for humans: rooms with flat floors, outdoor locations with roads and walkways, wheelchair ramps so people can get up the steps.
Re: (Score:1)
Your logic makes zero sense. If legs are better, why'd most environments designed for humans be flat? Don't humans have legs?
Reality shows the exact opposite of what you're claiming. Vast majority of our environments are already wheel-friendly, even in cases that see almost exclusively leg based locomotion, in spite of the fact that we have legs; because even legs work the best on flat surfaces, and uneven ground is actually generally considered a hazard, to humans, with legs. The one exception where we do
Re: (Score:2)
Your logic doesn't close.
Indeed you're right, flat environments are easier for legged locomotion. But they are essential for wheeled locomotion.
Legs work in flat and non-flat surfaces. Wheels work poorly in non-flat surfaces. Wheels lose in most locations that aren't engineered by humans to be usable for wheeled locomotion.
Re: (Score:2)
The lack of popularity of legged robots in production environments added to the clear popularity of static robots in production worldwide confirms that.
Another factor is likely that industrial robots don't do the work a single opera
Re: (Score:2)
The Daleks solved the stair problem the same way R2-D2 did, However both of them would be introuble with the EPA for emissions and OSHA for hot work permit violations. Admittedly the Daleks wouldn't care.
Back on earth if I recall from certain movies roller-skated waitresses were common at drive-in burger joints.
Robot roller-derby might be entertaining as well though without the romantic aspect of the Jim Croce song.
Re: Legs never made sense (Score:2)
You still see servers on roller skates occasionally. I used to go to a Sonic in Austin that had them.
Re: Legs never made sense (Score:2)
Not really - legs do make sense because they take a robot everywhere, work, stairs, home, jungle, or clifftop. They just do not take it at speed on flat, clean surfaces (roads, factory and warehouse floors), where the bulk of initial deployments lie.
This is really a response to the many Chinese robots out with both legs and wheels. The Boston Dynamics folks were not inspired by anime the same way the Chinese roboticists were.
Re: (Score:2)
While this is going to slightly complicate recalling robots when they inevitably malfunction and start deleting humans at least they don't have jet packs.
Yet...
Re: (Score:2)
- "They fly now."
Prior art (Score:2)
Re: (Score:2)
Chalk that one up to lazy animators. "Hey, now we don't have to draw a walk cycle! Sweet!" All of the Jetson robots were on wheels or anti-gravs; this was not a coincidence.
Venice Beach (Score:1)
Or was that just a street performer dressed as a robot.
Who is buying them ? (Score:4, Interesting)
Boston Dynamics has always made the most capable humanoid robots, but it's not clear what the market is for them. Their dog-like Spot seems to sell to police forces and military and serves a purpose to go where no human can safely go.
Musk is projecting a market in the millions for his TeslaBot, but various Chinese companies like Unitree already have humanoid robots for sale and don't seem to be finding much of a market for them yet outside of quirky hospitality/entertainment usage. Some companies evaluating them for industrial use, but AFAIK no large scale adopters.
wheels are the way to go (Score:3)
Robots should be able to quickly overtake and hunt down humans as silently as possible.
Re: wheels are the way to go (Score:3)
The roads and sidewalks in the USA are a robot defense mechanism. No way anything is rolling silently on either.
Re: (Score:2)
Sorry what, I'm too engrossed in my smartphone right now to listen. ;-)
Revenge (Score:2)
Sounds like we're soon gonna have Mr Puggle's Revenge going on.
make something actually useful (Score:2)
Okay, put those legs and wheels at the bottom of a ladder. Let me drive the ladder around to clean the rain gutters, or cut-in paint along the ceiling. Your stupid robot doesn't do anything but pre-programmed demos.
Re: (Score:2)
put those legs and wheels at the bottom of a ladder
You just made a bunch of lawyers start salivating.
You do know mobile scaffolds already exist.
Re: (Score:2)
I don't want a scaffold and the motorized ones are a joke. Im talking about scissor lift kind of drivability. If those wheels and legs can balance that heavy-ass robot then they can balance me on a ladder too. It can touch down or off on something when it not moving for stability while working.
Aye (Score:2)
And if my grandmother had wheels she'd be a wagon.
Robot Rollerball (Score:2)
Perfect opportunity to create a Robot Rollerball league.
Rollerball (1975 movie) could be done with these type robots an it could be way more fun to watch then the clumsy battlebot competitions.
Awwww! (Score:2)
They bought Atlas a pair of roller shoes
wow (Score:2)
It dances like the whitest guy on the planet, no hips.