romesc
Joined 10 years ago
Karma
71
Submissions
30
Member since
Oct 18
Recent submissions
Comment · 17 days ago
Well that settles it! They'll just have to hire us. Then the universe will be back in order XD!
Comment · 17 days ago
If you're interested in some more details, take a look at this post I made (and am admittedly still polishing) that discusses the method in a low-dimensional space. I am trying to make it more accessible to the community since explaining it for higher-dimensional robot control / hybrid-contact dynamical systems is a whole other thing :)
Comment · 18 days ago
Thank you so much for your kind words!
Ensuring the assets are "CAD-level accurate" has a huge impact on making this work in practice. So much so, that our collaborators Yash Narang [1] and Iretiayo Akinola [2] have spent a number of years on just this - we are standing on the shoulders of giants!
Additionally, understanding the hardware well (including things like motor and controller dynamics) is non-negotiable. We learned the hard way that there is, in fact, no free lunch. You really have to get the deep understanding of physics and engineering right.
[1] https://scholar.google.com/citations?user=M3NuG7AAAAAJ&hl=en [2] https://scholar.google.com/citations?user=e1zesfMAAAAJ&hl=en
Comment · 18 days ago
We use sim2real with distillation-based transfer :)
This means no need to do the rollouts on real robots first! Please check out https://sgs-rl.github.io for some transfer examples (albeit a simpler setting then the multi-part NIST board we examine in AlphaNIST).
Comment · 18 days ago
The mechanism is loosely related to curriculum learning, though we argue there are some distinctions. In the methodological companion paper (https://sgs-rl.github.io), we try to get into more details in the related work and baselines including comparison to methods like PLR (also from Deepmind), SFL, and linear handcrafted curriculums over states.
You are giving me some inspiration, though, that it might be good to try to explain the related techniques in a lower-dimensional setting to help teach the concept more effectively! brb
Comment · 18 days ago
Point received! We wonder what a better way to pay that same homage would be, since we followed the similar philosophy, but with considerations for our contact-rich robotic control setting.
Comment · 18 days ago
Hi, one of the authors here!
We can answer questions! But also, I want to highlight that the "AlphaNIST" name is an homage to the very well known work from David Silver's research group @ Deepmind circa 2016->2018. This was when we were coming up very early in our research careers and their work obviously had a big impact on us!
It has taken longer to make some of those ideas work for areas as challenging as robotics where there are physical constraints to deal with (and admittedly in this paper, we do not show the transfer to real hardware quite yet). But, we have some glimmer of evidence it is looking possible here (in another recent paper) that we cranked on for the last 6 months and will be featured at CoRL: https://x.com/octi_zhang/status/2108583315544682914
We are also all mega-lurkers here so we were honored to be featured on the front page!
Comment · 5 months ago
We are seeing increasing evidence that these sort of video world models are terrible options for useful rollouts of the physical dynamics of the environment. It is hypothesized that you can get them to be better than simulators by training on physics simulation data, but then the question becomes, why not use the simulator directly?
There are a lot of areas where predictive models make sense in the robotics stack, but doing it with "video world models" as is trendy this year is likely a bet in the wrong direction according to the evidence we have been amassing in the last 6 months.
Comment · 2 years ago
I had a lot of fun playing this! It's definitely the quirkiness that sets it apart from the typical "GenAI" content. Highly recommend other hackernews readers check it out.
Comment · 2 years ago
thats just not in the BDs ethos. They have been the only company really trying to physically build these kinds of robust, dynamic systems for the last 3 decades (almost to a fault).
Comment · 2 years ago
Thank you for thinking of this and implementing it!
But, I'm wondering is there any major difference between just clicking the "..." next to the song in the default interface and selecting "Download->Audio"?
Comment · 2 years ago
Or prediction is intelligence.
I think we need a more rigorous definition to go beyond tautology
Comment · 2 years ago
Wow, some actual legit content on HN!
Comment · 2 years ago
Congrats to Noam (and the whole team)!
Comment · 2 years ago
Sure A* is awesome, but taking the "star" and immediately attributing it to A* is probably a bridge too far.
Q* or any X* for that matter is extremely common for referring to the optimal function under certain assumptions. (usually cost / reward structure).