Zahi Kakish

Introducing ✨Posed DROID✨, results of our efforts at automatic post-hoc calibration of a large-scale robotics manipulation dataset.

Try it out at: droid-dataset.github.io

Learn more at:
🌐 arXiv: arxiv.org/pdf/2403.12945
📄 Blog: medium.com/p/4ddfc45361d3

🧵 1/n

The new foundation model from physical intelligence - pi-0.5 - works zero-shot in multiple homes. They tested at airbnbs around san fransisco. I also really love the robot design.

blog: www.physicalintelligence.company/blog/pi05

Nice way of describing how a #robot should generalize in a new setting using VLA approaches.

المسيح قام

The DBH quote about how religion comes to America to die always pops into my mind when thinking of American Christianity...

Ideally, when I unbox a new robot, I just want to teach it the way I would my kid, by giving it a couple demonstrations -- not by teleop, not by endless prompt engineering. So, how can we use human videos to teach robots how to perform tasks in new environments?

www.robot-learning.uk/r-plus-x

Honestly, no one should have to do this. Isn't this what robots and machines are for?

Really a failure of imagination by simpletons in the white house.

Wow! Did they open source this one too?

Blood Meridian is way more violent and gory than I had anticipated...

It def is all marketing

Ramadan Kareem to my Muslim friends!

If inference is good enough this could be great for robotics applications.

One thing I think a lot of tech enthusiasts will need to accept is that the idea of having separate CPU and RAM for small form factors is a thing of the past.

The benefits of bundling them are really outweighing having them separate.

I def see what he means. Lots of robotics researchers I talk to don't really "believe" the results coming from visomotor policies.

Postdocs and PhD students hit hard by Trump’s crackdown on science As US federal grants remain frozen and budget cuts loom, anxiety and fear grip early-career researchers.

The Trump administration's efforts to slash research funding have US graduate students, postdocs and other early-career scientists fearing for their careers. Some might leave the country -- or abandon research altogether.

https://go.nature.com/43ao2zq

Time to label some data...

New golden age of reinforcement learning and this time it seems deserved (as someone who thought it was bs last time we did this)

I noticed gas going up too 😫 Trumps America...

Was it just async and distributed RL that brought these performance increases? If you don't mind me asking Remi, what was your training pipeline?

🚨 BIG MILESTONE! 🚨

@deepseek_ai R1 brings us closer to robots with reasoning abilities 🤖✨

Inspired by RL at scale, we hit highest rewards in just 5 mins of async + distributed RL training on robotics task! 🏆⚡

It used to take hours on H100 before. Game changer. 🚀

Just learned that if Lunar New Year celebrates your zodiac sign then you must wear red on Lunar New Year or something bad would happen to you.

Looks like I have to wear red this week lol

Very cool demonstration of in-context representation learning. This figure says everything.

arxiv.org/abs/2501.00070

it's a wonder**

It's funny how reddit subreddits for certain topics either get super obsessed with the topic/product or hate the topic/product with a passion.

Like going on the Firefox subreddit, it's why these ppl even post in that sub if they hate it this much...

Introducing Stretch AI - a new open-source suite of tools, tutorials, and reference code to explore and build AI-enabled home robot applications with Stretch 3! Check it out at github.com/hello-robot/stretch_ai and learn more from @cpaxton.bsky.social below!

This is what it will take to make America more of a manufacturing hub.

It's remarkable how OSU shits the bed whenever they face Michigan...

Denoising Diffusion-Based Control of Nonlinear Systems We propose a novel approach based on Denoising Diffusion Probabilistic Models (DDPMs) to control nonlinear dynamical systems. DDPMs are the state-of-art of generative models that have achieved success...

Anyone in the control theory community going to IEEE CDC willing to help us out to present a paper for us? Me and our collaborators are having visa issues.

I will provide the slides. The work is at the intersection of diffusion based generative modeling and control theory: arxiv.org/abs/2402.02297

Great list! Would it be any trouble to add my PhD advisor @springberman.bsky.social ?