Jeff Bingham

Miscalculating the gap between algorithm and physics isn't a new-found mistake by practitioners of generative AI; it bit the math-y controls boffins and vision researchers before them.

One perpetual mistake in robotics is to confuse observations with reality and assume that if you can generate an action, you can execute the action.

Fresh perspectives bring excitement and ignorance that tackle old problems in new ways. There is both remarkable success and frustrating repetition of the same mistakes.

The massive influx of data science into robotics is delightful and harrowing.

I suppose it is the beauty and curse of the human imagination. A demonstration is a promise of what could be; a small white lie that motivates the realization of the dream.

Nearly 60 years since Shakey, but many cool things in-between.

YouTube video by Google DeepMind Gemini Robotics Demo: Timing Belt

Lots of fun stuff shown today with the GDMR blog post, but one of my favorites is this NIST board. And Todor's dashing sweater, of course. youtu.be/2AAFiuEP7iE?...

Introducing Gemini Robotics and Gemini Robotics-ER, AI models designed for robots to understand, act and react to the physical world. Introducing Gemini Robotics and Gemini Robotics-ER, AI models designed for robots to understand, act and react to the physical world.

More brains for robots!

Really looking forward to how this progresses:
deepmind.google/discover/blo...

I am teaching a class on #FoundationalModels for #robotics and Scaling #DeepRL algorithms. This class expands on last year's class and my generalist robotics policies tutorial and code. I plan to share the lectures and code assignments. Starting with the first lectures below.

Use the force! Interesting use of kinesthetic teaching with diffusion policy. Hoping there will be more details on the exact action space and training.

Paper: arxiv.org/abs/2501.10356

Videos: clairelc.github.io/dexforce.git...

This block picking demo is a pretty rad example. It makes me so happy that these tools are accessible!

MuJoCo Playground An open-source framework for GPU-accelerated robot learning and sim-to-real transfer

Dreaming in sim and playing in reality. Pretty slick process with the new JAX enabled tooling in MuJoCo.

playground.mujoco.org

I'd like to introduce what I've been working at @hellorobot.bsky.social: Stretch AI, a set of open-source tools for language-guided autonomy, exploration, navigation, and learning from demonstration.

Check it out: github.com/hello-robot/...

Thread ->

Neat course on robot learning published by Marc Toussaint

www.user.tu-berlin.de/mtoussai/tea...

the BigVision repo is my current reference impl for gemma and ViT. such an underrated repo @giffmana.bsky.social and team are doing the lord's work

github.com/google-resea...

github.com/google-resea...

The Beetle Roboticists: A Parable - Matthew T. Mason How do we assess ideas about humanoids, without our anthropocentric bias? Maybe if the ideas were being pushed by another species. Beetles?

From Matt Mason: "Anthropophilia versus technophilia. How can we judge this debate fairly? Since all humans have an anthropocentric bias, we need judges from some other species! And that is the silly idea that led to this parable."

mtmason.com/the-beetle-r...

GitHub - SLAM-Handbook-contributors/slam-handbook-public-release: Release repo for our SLAM Handbook Release repo for our SLAM Handbook. Contribute to SLAM-Handbook-contributors/slam-handbook-public-release development by creating an account on GitHub.

We are in the process of editing a SLAM handbook, to be published by Cambridge University Press, with many *stellar* contributors. Part 1 is available as an online draft for public comments. Help us find bugs/problems!
Link to release repo is here: lnkd.in/gZhTkaxb

Let me also put the link to the wide baseline stereo blog.
It started as PhD thesis in a blog format, then expanded around Image Matching Challenge. I should write there more though

ducha-aiki.github.io/wide-baselin...

AGI : neat :: physical-intelligence : rad.

Here is a simulation to watch of a passive walker evolving as you ponder if an embodiment is important for artificial general intelligence: bingjeff.github.io/walkman/