Boston Dynamics’ Atlas tries out inventory work, gets better at lifting
- Thread starter JournalBot
- Start date
You may wish to familiarize yourself with the rest of Boston Dynamics Atlas' capabilities.
Tesla has nothing close.
Atlas Parkour
Upvote
111
(122
/
-11)
“Good news, everyone! Our robot is no longer as good at crushing things in its metal claw! Now its prehensile hands can pick up sophisticated weaponry.”
Upvote
151
(152
/
-1)
I imagine depending on the work, it can be slot in and be plugged in 24 hours a day working and can start off being 1/3 or 1/2 the speed of the person to have a similar cost to the business, with no healthcare costs to worry about.
Upvote
89
(89
/
0)
It might be slower, but:
It can work 24/7 (as long as you provide power).
It never gets tired.
It doesn't go to bathroom breaks.
It doesn't take vacation.
It doesn't get sick.
...
I wouldn't be surprised if this starts replacing humans in some jobs gradually in ~5 years.
I am all for automating anything that can be automated because it increases productivity which leads to a larger pie. The challenge for us as a society is to ensure the a lot of people don't end up without a piece of the pie.
Upvote
130
(131
/
-1)
Oh, on that we totally agree! The average Boston Dynamics demo makes Musk's efforts look like a total joke.
It's just that the article comes off as rather unimpressed.
Upvote
33
(38
/
-5)
Snarky Robot
Ars Legatus Legionis
Well, this one probably isn’t faked, so that helps. Really shows how Hyundai is really more of a tech company than a car company, and their stock should be valued as such.
(For those who don’t know, Hyundai owns Boston Dynamics and is therefore likely the car company with the most advanced robotics investments and research).
Upvote
75
(81
/
-6)
It's amazing how the sudden ubiquity of GenAI this, ChatGPT that, ad nauseum, make me almost warm and fuzzy about the idea of robots taking over the physical heavy lifting for humans.
While there are definitely risks involved in replacing people with machines for physical reasons (policing and military use come to mind), I still far prefer this approach to shoving machines/machine learning up the Bloom's Taxonomy slope.
While there are definitely risks involved in replacing people with machines for physical reasons (policing and military use come to mind), I still far prefer this approach to shoving machines/machine learning up the Bloom's Taxonomy slope.
Upvote
5
(9
/
-4)
Yep It's Me
Ars Centurion
These achievements are always so impressive!
But, I do always have some background thoughts in the ol brain, among them: the end of Battlestar Galactica, the trial of whether or not Data is sentient, and the Bicentennial Man's quest to become a man and be acknowledged.
I wonder what we're creating today and what will be wrough for tomorrow. I hope we'll do the right things and make the right choices. I hope instead of a nightmare dystopia we can find a path to greater good.
But, I do always have some background thoughts in the ol brain, among them: the end of Battlestar Galactica, the trial of whether or not Data is sentient, and the Bicentennial Man's quest to become a man and be acknowledged.
I wonder what we're creating today and what will be wrough for tomorrow. I hope we'll do the right things and make the right choices. I hope instead of a nightmare dystopia we can find a path to greater good.
Upvote
-5
(6
/
-11)
A 1 minute demo is important, because unfortunately the power requirements and cooling requirements are for hydraulics, so it likely has a maximum battery time of 15 minutes or so. So unless they have tethered power and cooling - they probably have to switch to non-hydraulic - which has way less power for the same volume of actuator.
Upvote
10
(11
/
-1)
The video's a lot faster than the gif. Is the gif slowed down, or the video sped up? If the former, then the actual behavior is pretty speedy.
Upvote
9
(9
/
0)
Upvote
16
(16
/
0)
I have a feeling the hand motion would be at least 62% less creepy if instead of the two outer fingers flipping and moving down, the middle finger moved down instead, mimicking an opposed "thumb".
Upvote
14
(14
/
0)
As impressive as this is, and it is impressive, I'd probably be happier with smaller hands that could load my dishwasher.
Upvote
1
(4
/
-3)
A BD partner (Chris Nielson of Levatas) mentioned that this was filmed last year, and hinted that BD is working on a manufacturable version,
View: https://twitter.com/Docneuroeo/status/1754911679698371019
View: https://twitter.com/Chris_S_Nielsen/status/1754912191956893722
https://levatas.com/unlocking-the-power-of-automation-levatas-boston-dynamics/
View: https://twitter.com/Docneuroeo/status/1754911679698371019
View: https://twitter.com/Chris_S_Nielsen/status/1754912191956893722
https://levatas.com/unlocking-the-power-of-automation-levatas-boston-dynamics/
Upvote
14
(15
/
-1)
Eh, depends on your cycle time. IF it can't move fast enough to meet your current cycle time this will slow you down and cost you productivity.
And any factory can already structure shifts so there's around the clock running of the lines if they want
Upvote
-18
(5
/
-23)
Snarky Robot
Ars Legatus Legionis
When robots fully come to the home, I’ll expect they’ll be like Roombas: special, purpose-built devices with limited functionality. For example, a dishwasher and sink directly underneath where dishes are stowed , with smaller arms with less strength, and cameras trained to identify dirt and type of dish.
A humanoid robot is most “friendly” looking, but the human form is not optimized for tasks that aren’t historically necessary for survival. A cooking robot would actively be hampered by the need to follow the human form and use human tools.
Upvote
34
(35
/
-1)
Maybe double-entendre ?
As in the robot flipping off people who think moving to a blue collar job will save them from automation. Joke aside, a 3 point grip is quite stable and should need a bit less force than a clamp grip, There must likely be other technical considerations as to why atlas doesn't just have a typical 3 fingered "vertebrate" hand.
Upvote
1
(1
/
0)
It'll be interesting to see how this sort of stuff pans out. While they'll likely be able to replace people in function in the near future there's a lot of things I haven't heard addressed. How much does the robot cost, what's the maintenance schedule and parts availability, how much do the people who manage the robots cost, etc.
Upvote
11
(11
/
0)
That's not a robot, that's a brazen attempt to overvalue the stock, to benefit some rich asshole dressed like a robot.
This is a dancing robot.
Upvote
24
(29
/
-5)
I wonder if they are planning a serious move into the humanoid robot market, or if this is still a research project. The two are not actually mutually exclusive, but in the past they have made clear Atlas is just for research.
Upvote
8
(8
/
0)
We're probably a few generations away from something with that kind of up time, but eventually, that will likely be true.
Also, labor shortages, there a lot of places where getting anything done, without a person, no matter how slow, is a big win, assuming you don't need to hire even more hard to find people to keep the robots working...
Upvote
6
(6
/
0)
I was under the impression that the tesla robots we've seen to date were actors in a suit on stage, and they have no working prototypes yet.
EDIT: Huh, apparently not, there are videos of walking prototypes as of 8 months or so ago. Lightyears behind Boston Dynamics though, and I'm not convinced the video I saw wasn't your typical rigged demo with shiny looking hardware that can only do one thing, and only if you don't watch for long.
Upvote
-2
(6
/
-8)
I don't see this as an issue. If it's a known item, it just needs to look long enough to recognize it. If it's a new item with a CAD model, it uses that info to make an initial assessment, then updates based on experience. If it's completely unknown, it studies it for a bit to build a model. There was a point in the video where it looked like it was doing that, though I don't know if it really was.
Regardless, all three of those options are what people would do. If I go to pick up something I recognize, I use that knowledge. If it's something big I don't recognize, I estimate its center of mass and total weight to decide where to grab it and whether to use two hands. If it looks small and delicate, I take that into account as well.
Upvote
21
(21
/
0)
I'd bet that the hand would loose about 90% of it's creepiness if they had just switched which side of the clamping arrangement had 2 fingers and which had 1. Then it would be somewhat mentally mappable to how our own hands work.
Upvote
5
(8
/
-3)
Than again they are owned by Hyundai - I wonder if that is the task set they are focusing on....
I personally think robots will be perfect for that Lunar/Mars colony, before any humans get there....
Upvote
4
(4
/
0)
There are WAY too many moving parts and delicate hardware in Atlas to turn it into a viable product. I think it being limited to a research platform and aspirational mascot is just about right. That doesn't mean they aren't incorporating atlas technology into viable products though - I'm sure Spot owes a lot to Atlas, and Spot is definitely a useful product in the right niche, like remote monitoring and data collection.
Upvote
6
(9
/
-3)
A lot of the BD demos are only possible with high force actuators (which aren't generally practical due to high energy usage, high cooling requirements, and reduced service durability). Also many of the demos were just motion transfer preplanned paths and simulations and thus fairly misleading becuase no planning or adaptation was involved. So many of their demo's were closer to animatronics than general purpose robotics.
Upvote
-16
(5
/
-21)