It's strange how these Boston Dynamics videos were already going "viral" almost 20 years ago when I was in school. [quod pod/dog walking terrain outside]
Back then, you also thought practical robotics would be almost there.
Really makes you think about today's videos, and whether it's just hype all over again.
If you have the data/background, feel free to expand how these demos are different from back then in relation to robotics becoming actually deployable in real arbitrary environments.
I was thinking the opposite. I was thinking (after seeing another company selling a pop-up button for $8) that here is a company that is actually making shit — and constantly (perhaps slowly) making it better.
From dog to human. And, as I understand it, from hydraulic to electric. Looking in particular at video #3, I get a sense that there has had to be a lot of very expensive iteration to get to something that has nearly exactly the degree of motion of a human.
Maybe the problem with the videos is that we have come to expect this as "normal" and no longer viral. Now, that's kind of a wild thought.
But to your question about when they will be deployable — I would hazard to guess that cost is the issue. If these came down a few orders of magnitude in cost I suspect we will start seeing them deployed.
And that's the part that makes me wonder if BD will ever "make it", starting 20 years ago they probably had to make quite specialized algorithms with a ton of hardcoding to make their robots work (and it was impressive then).
Just starting 15, 10 or even 5 years ago neural networks have been acing these kinds of "balancing" problems in other areas but with an entirely different architecture.
In 2025 with the rapid expansion of other "AI" fields, I would expect a robot to weld a scaffold from loose parts and THEN cartwheel off it, go to a human handler and pick up an rifle and then hit a target 100 meters away (or similarly "different" task from the welding to show versatility).
It's robotics, which is an actually difficult technology. What Boston Dynamics has achieved in the past two decades is impressive.
It took decades for computers, after they appeared in government projects, to become ubiquitous. There were many problems to solve.
In our era, we're just used to average companies hyping up piddling, niche products by calling them new technologies. Sure, if the 'technology' is an electric juicebag squeezer or something, someone can get it from research to market within a year.
All that is aside from whether it's a good idea to thrust robots into the world as quickly as possible. Between robots and AI, we could lose a billion jobs in a short time period.
For me, the problem is that while I find these videos extremely fascinating, I'm missing the whole interaction part - it is unclear to me how much of these movements are preprogrammed and how far away the engineers are from having these robots being able to respond to complex requests. This mapping from human language to motor action needs to be the next step (something that Nvidia is working towards).
Those quad dogs are delivering rations and ammo to Ukrainian trenchlines.
There was talk about them being employed even further forward a few times, but I think aerial drones are a lot less complicated & expensive to build, harder to shoot, and most importantly easier to maintain a long-distance wireless connection to.
Dogs, humanoids can better interact with the environment but even back in the day the Boston Dynamics dog marketing was to a large extent automated infrastructure monitoring with things like thermal cams and less so the actual intervention which means you can just get drones with thermal cameras and skip the whole walking issue/cost.
On the other side if you need interaction you can just get an arm. My current employer is looking at the humanoid robots more so for the marketing that they can do things "magically"/without careful preprogramming and adapting to conditions but if you can do that with a humanoid, you can do it with an arm for cheaper still.
The space where you need monitoring, interaction and movement to me seems very limited and even undesired. There's a reason we avoid having people do the entire process from start to finish and like putting them in as short as possible loops..
Or wired connection. There are drones that use fiber to resist to electronic attacks. Of course, it's basically for suicide missions only where you fly directly into deep enemy-controlled territory.
I remember thinking that the humanoid robotics revolution was imminent after seeing ASIMO walk around 2010[0], but after observing the slapstick comedy that was the DARPA Robotics Challenge around 2015[1], it seemed that we had actually made zero progress. But I haven't felt that way in years!
My feeling at this point is that the best humanoid robots have, from a hardware perspective, surpassed the "minimally viable product" threshold. By this I mean that they are physically capable of completing enough tasks that would make them worthwhile a worthwhile product. They are physically capable of walking the dog, folding the laundry, picking up the living room, bringing in the groceries, etc. It's possible that there are reliability issues to address, but given that the newest generation of bots are fully electric, it's hard to imagine that this is a more difficult problem than those they've already solved.
At this point, we're just waiting on the software to catch up. And surely this problem seems far more doable today than it did 20 years ago, given all the progress we've made in machine vision and machine learning!
We'll know that the robotics revolution is truly imminent when the big players start pouring capital into development. Tesla has already joined the fray. I expect more will follow in the next few years.
> Really makes you think about today's videos, and whether it's just hype all over again.
That's my gut reaction when I see these announcements. That, and that I don't want a robot that can do cartwheels, because cartwheels don't wash dishes or mow the lawn. I don't see why they aren't focusing on demonstrating practical tasks.
When a robot is moving itself it has a very good idea of what each part of itself weighs, how big it is, and has sensors to tell it what angle all its joints are at. These are all much more difficult with an object the robot is holding in its hand. But you can still find videos of Atlas doing practical tasks, just not as impressively as it jumps around.
A mentor of mine in robotics talks about how the people at his research group would watch a video from an old Julia Childs TV show of her chopping vegetables and doing other cooking tasks and they would try to figure out how long until a robot could do those things.
I don’t see any robots changing bedpans, fixing plumbing/electrical, picking fruit and vegetables, cleaning bathrooms, making beds, folding laundry, etc.
For fruits and veggies actually there are already more or less automated harvesters. They are just not humanoid. Also I guess that their cost might slow down adoption.
Huh? You know dishwashers and lawnmowers already exist right?
I was at a friend's house (I don't have a lawn) a couple of weekends back and the plates and cutlery and so on we used is just "magically" clean in the morning so long as we remember to put them in the right place, you hear cutting, and outside there's a machine wandering about cutting the lawn periodically.
We don't need a humanoid robot to wash dishes, that's a waste of resources, like when people thought we should have crews of people typing documents rather than just... not.
Nobody is buying a $20k home robot to replace a home dishwasher.
But they might to replace a commercial manual dish washing position.
Which begs the question - why don't restaurants use actual dish washers? They have a "Commercial dish washer machine", but that's actually just a final sterilizing step, and restaurant dishes are instead washed by hand.
> why don't restaurants use actual dish washers? They have a "Commercial dish washer machine", but that's actually just a final sterilizing step, and restaurant dishes are instead washed by hand.
Not the ones I've seen, dishwashers are pretty common. Are you seriously saying that the dishes are handwashed in the Restaurants you know?
A commercial dishwasher does not wash "3D debris" like a home dishwasher does, it's a reserve of boiling water and fairly caustic chemical cleaners that removes lipid films and sterilizes the surfaces, and it does this in a matter of minutes. From what I can tell basically every restaurant employs people to wash/rinse dishes prior to this final step, using a high pressure nozzle, bleach or quat rags, and a series of sinks. The people who end up employed in this role and who actually show up every day are frequently given a lot of leeway as far as eccentricities, history, and interpersonal problems (Nazi facial tattoos and a felony history? Can you start Tuesday?), because nobody else wants to do it.
This one is far better than the robots they had back then. Which, moreover, weren't using machine learning for control but handcrafted algorithms, which don't have the potential to scale as far as machine learning. The robot in the current video was explicitly trained with reinforcement learning.
Their videos were a hit 42 years ago when I was in school. Of course back then Atlas's great, great, great grandfather could only hop in a circle with a center tether.
So when you see the full history, you realize the progress has been fucking amazing.
they do actually have deployed big dogs in certain industrial settings, and have for a few years at least, though I'm not sure what the bar for success is
Aside from looking a bit sluggish when accelerating, and having too large oscillations in the leg control loop, it looks like motion capture. Very impressive.
If I think about robotic autonomy, I imagine a sort of layer cake:
1. Intent and Planning -> What am I doing, What will happen when I do it
2. Macro Implementation -> What are instructions down to all my parts
3. Local Implementation -> Given Instructions, let’s move my leg
4. realtime / reactivity -> OK, I’m 3 degrees off expected on this foot, need to rebalance everything.
Figure et al are demoing 1-4, but slow and without motion - just heads and hands.
Tesla’s robot day was demoing 4 only, as far as I can tell — engineers with controllers did all the macro planning and movement requesting - but there’s no way you can rebalance a robot live with a handheld controller - that seems certain to have been live from code.
BD has always existed in this kind of liminal space to me, where it’s not clear that they’re doing anything more than 4 with their demos. Extremely impressive mobility demos, to be clear. But, I’ve never seen a humanoid demo from BD that’s not in what looks a pre-scanned and planned room.
And we never hear claims about 1, 2 or 3 for humanoid robots from BD as far as I’m aware.
I think for years the low level mobility part has been so hard, BD has felt way ahead, the planning and environment responsiveness was just science fiction. Today we see that the top part of that layer can be gotten with vision models, and we’re seeing dual architectures where those vision models instruct small fast models for layers below. My guess is that this architecture is going to be relatively resilient, changing the cost structure and opportunities for robot companies.
So, I guess if I add all this up, we have a race between mobility experts and agile ML teams with robotic experience. And that adds up to some early launch products; I’d imagine they will be clumsy, and that BD will be a late launcher if it ever goes after consumer at all; it may just decide to license. Anyway I’d like to buy one, so the next few years should be interesting.
If America's imperialist death machines can't do superhero landings and strike cool poses in front of burning villages what are we even doing as a culture?
I still don't see why you would need a humanoid robot for those tasks. Folding laundry could be done by a single robotic arm, and dishes are done by a dishwasher.
Because managing all those dishes still takes a lot of time. Taking heap of dirty dishes, cutlery and whatnot, pre-clean if something nasty is already dried to them, arrange them efficiently, and then afterwards taking them all back to their places.
That's maybe 7 minutes on average combined. It takes probably longer to write some meaningful post here (with some corrections), but we are spoiled these days, pretty irrational, and consider our free time our most important aspect. Some people would prefer eating broccoli or spinach paste to managing dishwasher.
Also, if somebody expects me to fork out 20k for some assistant, there are no limits what I can and should expect out of it. Heck, you can have modern-day slaves for less in most modern societies, not even going into less developed ones.
Even moreso in that you can buy them, today, commercially, for just over $20k. China is several years ahead of us in this space, like they are in most others involving manufacturing, and the gap is widening by the day.
Where are they at with power supply? All these videos are now showing the robots untethered (and they have been for a while), but how long can they operate untethered?
Are they just running on Lithium Ion batteries? And the new electric joint motors are more efficient than powering hydraulics?
I would be much more impressed to see a person with a controller (and just directional controls) telling the bipedal robot where to go in a normal human environment - for example, just crossing the street.
These seem very curated, on perfectly flat, grippy surfaces.
Perhaps it just shows how truly difficult the human method of bipedal locomotion really is that this doesn't seem to have been achieved yet.
It's all fantastic until that one rotation of the waist...then it flys straight through the uncanny valley. It goes from 'observation of movement of an organic thing' to 'I'm calculating the most direct route between steps A and B'...to horrific effect.
What surprises me with all of the robots is that the engineers had the freedom to create the best machine for all purposes, yet they keep creating a "humanoid" with 1 body, head, 2 arms and 2 legs. Is our human body form really the best for all tasks?
I was pretty amused when I saw the headline, thought to myself, "oh, there must have been some reveal for the new Atlas robot" and then noticed the url was newatlas.com.
Modern tech "progress" feels like you're drowning in a lake and someone throws you a bag of m&m's. Sure I like m&m's but it really isn't anywhere near the top of my list right now.
Do "we" need anything else than food and sleep? Humanoid automata trope was invented in ancient times, and it's everywhere in sci-fi since about a century ago. Is it really surprising that someone actually tries to make it happen just for the sake of it?
Human-induced climate change is drastically worsening the living conditions for billions of people who will fight to survive in the coming decades and centuries.
And we still have a chance to handle the situation in peaceful and equitable ways - by transitioning from a competitive to a cooperative (non-commercial, trade-free, open access, open source) commons economy!
Once human labour is not competitive any more (2027-28?), we have to change our economic framework anyways. That's why I'm rooting for the aforementioned option with all the zest left in my aging body..
Is it just me, or do these new "leaps" from Boston Dynamics feel tiresome?
People would surely appreciate automation that helps with household tasks like cleaning, chopping vegetables, and ironing clothes. But such delicate activities don’t seem to be part of BD’s vision....not even on the periphery.
What is the end goal that BD has in mind? Yet another Police/Military toy?
I wonder what this will cost to build and how often they break. I doubt it's going to be able to compete with the price of human labour anytime soon. So it's either for going places people can't go (rescue bot?), or doing things people really don't want to do, like walk towards gunfire.
Whenever I read something about Boston Dynamics now, I can't help but remember that scene from Silicon Valley (HBO) where Eric kicks BigDog https://www.youtube.com/watch?v=BFnMBW95RZI
Back then, you also thought practical robotics would be almost there.
Really makes you think about today's videos, and whether it's just hype all over again.
If you have the data/background, feel free to expand how these demos are different from back then in relation to robotics becoming actually deployable in real arbitrary environments.