alt Hacker NewsXiaomi-Robotics-1
Comments
For a bit of context there are some classically very hard things going on in this video, so if you're not impressed, be glad you didn't have to try to build this 10 years ago.
- Coordinating two hands
- Mobile body
- Deformable objects
- Thin affordances (bag zip in particular)
- Multi-object single grasp
Really a host of things that had individually been tackled as either perception or manipulation problems and probably led to more than one person's PhD here on HN.
Watching the two-handed robot has me thinking that a third hand, perhaps with a different set of capabilities (like a tentacle grip) might make many tasks easier. I realise that more limbs = more problems, but even so, having additional ways to hold and turn things would be a big help.
Coining the term: Slopfold
When you Robot folds your clothes but it's kinda wrinkly and sloppy but you accept it anyway because it's easier than precisely folded and pressed clothes.
It's happening so fast now.
AI will absolutely dominate our future. I believe we have only scratched the surface of what is possible.
Maybe I'm pessimistic but I always thought Bill Joy was prescient.
this is what excites me most about the probable AI future. househelp is cheaply and easily accessible for most people in the east, however being able to afford a cleaner or maid in the west is a luxury. i spent a ginormous amounts of time every week keeping my house in shape. i look forward to the day when i will be able to carve out that time and get it back for myself.
The uncut footage is so awesome.
Of course, you can mention all the things they can't do (what if there are stairs, what if the doors are smaller than the robot, etc) but it's mindblowing to see what they can do - and also the chat panel is really nice and makes it more interactive.
Designing a non humanoid robot is the way to go. The design of new tools and appliances for hybrid human/robot use will eventually follow: the zip line for the shoe bag required way too much precision work.
This is amazing! I can see in a few years having Home Assistant use an LLM to coordinate my home while I'm away. Tell the LLM to make an inventory of my home and then send the vacuum and the robot around to clean everything up. A multimodal LLM can inspect pictures for items that need to go in the dishwasher, the laundry or needs polishing and reorganise my closets to best allocate space, make the beds, take care of the plants. It has all day to do it while the solar panels provide ample energy anyway.
Does this require a given hardware platform, or for example, could I run this model with a cheap mini robot with a single camera (i.e. good enough without depth), tracked treads, and an arm?
As a house spouse I want this! I _hate_ folding laundry and always procrastinate on it. Doesn't even need to be able to load the machine like they additionally show, "just" folding it is enough for me!
I love this future. A robot that does all the chores relentlessly of the house while you spend time just working on the stuff you like.
I think this robot is really well-made! Xiaomi seems to be quite controversial these days, but personally, I still really like the brand.
Anyone have any recommendations for robot hardware where I could test this open-source model? It would be a cool/fun project to get it working on a little robot.
Always remember this is the Model T from Ford, imagine what they will do in 10 or 100 years from now
Exciting times we live in
This is very cool, and the paradigm that scales to bringing truly useful robots to virtually all domains of life - just as LLMs have scaled to that degree across virtually all domains of knowledge work.
I picture a lot of perfectly tended gardens growing food.
How are these models compared to world models by Waymo and Tesla, assuming their models are trained for recognising objects that can be further generalized for everyday chores?
Please help me understand. Is it more effective to have a general model like this instead of a state machine that leverages other models and a deterministic set of code?
How dare Xiaomi post a realistic picture of where modern consumer robotics is at.
They will never become trillionaires if they don't pretend they are making a million humanoid robots by end of the year.
Really interesting.
From even before the RAM shortage there was noise in the robotics world of sensor and actuator shortages due to them being bought up in bulk by some of the major AI labs.
feels like the American domestic manufacturing is done, there is no hope to save it.
The first few second of the shirt folding arms trying to put the t-shirt properly on the table to start folding it are at the same time amusing and scaring.
What if the thing retrieves some anger-driven post from Reddit from its training data set and start literally to flip the table??
Neither the Github nor the HF repos have anything on the XiaomiRobotics-1 model or datasets :-( They have a bunch of assets (datasets, code, model, etc.) for XiaomiRobotics-0 however, but I don't know much about that family of models.
Pretty cool!
Only 10B parameters if I correctly understand some of those histograms, which is tiny in LLM world! It bodes well for the future.
The training videos seem to show human hands using the robot grippers to execute the tasks, which must have been a fun but tedious process.
How well do models like these generalize to robot embodiments they weren't trained on?
WAIC has been wild this year. We got Kimi K3 and Qwen 3.8 announced. We got robots boxing and open-sourced robotics foundation models. And we got the announcement of WAICO.
At what point are we gonna start seeing American firms attending WAIC?
In the domain of hardware, what is the equivalent of a harness? How well would this generalise to any robot, versus one that is specifically built the way Xiaomi did?
I think this is the most satisfying video I’ve seen of a robot doing laundry. Not sure if it’s the camera angle, the 3× speed, or the music.
How are there still people who look at all this advancement and the only thing they can utter is "it's not there yet?" It's clearly here, and if it's not to these nerds' expectations, it'll prob meet it in 4-6 months. This was cool asf.
Would love to see a load/unload dishwasher benchmark.
Cue the variations of:
"No wireless, less space than a Nomad. Lame"
I hope I can use this sooner
Cool video but it forgot to pack underwear in the bag!
Of all the tasks I do, folding clothes takes a ridiculously small slice of my time. Is this and all the other robotics videos aimed at H&M staff?
Perhaps what impressed me even more is how functional the page was without JS enabled!
"Did you pack the cases yourself sir?"
Please keep in mind that "uncut footage of ..." is the same as those instagram videos where people throw a basketball backwards 20 meters while drinking beer, straight into the goal.
It doesn't prove this is average performance. In fact, if it was, that's not what they'd be showing.
The fact it didn't do the 2 second tshirt fold trick shows it's just copying trained movements.
This is Automation not AI.
it looks like the robots are not just autonomous but also they do cooperate with each other.
pi.website
DO SOMETHING
What if mum but venture backed lmao
[flagged]
[dead]
It makes me think of why couldnt they just automate ironing and folding. I can stick stuff in the suitcase myself.
The video cuts to a new shot every couple of second instead of one continuous shot like all robotics research. This makes it hard to trust it. The robot probably failed a million team and they stitched the video from 100 experiments.
I'm not sure why some comments here are so pessimistic. I was grinning ear to ear while seeing the video. You're telling me we've finally got robots that can do our laundry? And the model is at least as free as beer?
To quote the lyrics/meme, we used to pray for times like this. This is what robotics should be used for.