Emerson Segura

@ErenChenAI lol, cute

Excited to announce I’m starting the world’s first flying quadruped startup. Currently raising a $50M seed round. DMs are open.

@lucyjcai @anvil_robotics Cool

first time teleoperating our new @anvil_robotics arm!

@ruihangzhang Looks great! How do you generate the cube? Does it work for objects outside of training set

[1/4] 🤯Unlike existing methods, ProxyPose requires only a monocular video and a single query pixel. • No CAD models • No depth information • No object masks 🚀From just these inputs, ProxyPose tracks 6-DoF motion across diverse objects, motions, and scenes.

Introducing 📢📢ProxyPose📢📢 ✨Track 6-DoF motion of any query pixel via video-to-video translation 📰ArXiv: arxiv.org/abs/2607.06555 💻Code: github.com/ruihangzhang97… 🌐Webpage: ruihangzhang97.github.io/proxypose

@foporcher this is something I thought of (re addressing pixel space limitation of current "world action models") but did not have the technical skills to build it.. you did, and it works! So much more to explore in this area

1/ My first PhD paper is out! 🎓 Title: Flow Matching in Feature Space for Stochastic World Modeling tldr: we build stochastic world models directly in high-dimensional DINOv3 feature space, instead of relying on low-dimensional VAE latents.

@foporcher Great work!

@bmontxna @REK probably more fragile than it looks(as are most contemporary humanoids).. more like a Lamborghini than a Humvee:) - impressive looking tho

can’t really see a future of coexistence with this one tbh

@petersteiner10 @bmontxna lol, also for reals: it's surprizing how well wood burns in the rain, (once there is a fire)

@bmontxna And building a fire from wet wood in the rain!!!

it is absolutely a prerequisite for my next boyfriend to not only love camping but be really good at all camping-related tasks. don’t even enter my DMs if you’re not good at grilling and pitching tents I’m serious

@bmontxna lol, true! also don't worry soon robots will do that for you .. by soon I mean 10y to 20y :)

@UnitreeRobotics this is great

We're excited to support BitRobot in open-sourcing the largest humanoid whole-body teleoperation dataset collected in real homes. We hope it accelerates progress toward general-purpose humanoid robots.😉

@ashleybchae @X @nikitabier Myspace was fun!

I miss feeling status updates from myspace days (good ol days). It feels like what's missing here on @X tbh can we bring this back? 🥺 @nikitabier I remember that i loved seeing my friends status updates on their feelings

where the AI money is coming from: cuts to Accenture/Deloittle etc, is probably where CIOs find budgets to pay for llm tokens at OpenAI and Anthropic

@ethanmclark1 Llms benefit from a cheetcode robots don't have:their output "english"or"code" can be consumed "as-is" by people or computers (compilers/interp),as books did b4 llms. Yet,seeing a video is not sufficient to teach how to sucessfully do brain surgery at home,or pilot a space rocket

Working in robotics right now is what I imagine working with language models felt like in 2023. Everyone throwing things at the wall to see what sticks Pixel prediction (Cosmos), action prediction (VLA), reward prediction (TD-MPC), and representation prediction (JEPA). Different paths for the same problem The recipe that won in language was self-supervised pretraining at internet scale then light finetune on top. Only representation prediction runs that playbook. It learns from action-free video data so you can pretrain on YouTube and egocentric data then add a control layer. Everything else needs action-labeled data that doesn't scale As an RL maximalist, I used to hate LeCun's cake. Turns out he was right all along which is how I ended up a JEPA truther

@maxtitovvv @SimonYu_Chinese @dolylupec @wuji_global @TheHumanoidHub @XRoboHub yet in Lena's video the hand is unplugged

@SimonYu_Chinese @dolylupec @wuji_global @TheHumanoidHub @XRoboHub my guess it’s actively backdriveable. so when motor current rises the finger actively moves out the way.

Super excited to visit @wuji_global who's pushing the frontier with their advanced dexterous robotic hand Most traditional robotic hands still rely on tendons/cables, but Wuji Hand uses direct-drive actuators embedded right in the fingers with worm gears. This solves the usual tendon problems by delivering smoother, more precise, and reliable motion with way less sim-to-real gap It also has high dexterity with 20 active degrees of freedom and the weight/size feel almost exactly like an adult human hand Got to test the newly launched Wuji Hand 2 today and was super fascinated! (watch how I handle the hand in the first video😂)

@SimonYu_Chinese @dolylupec @wuji_global @TheHumanoidHub @XRoboHub I have the same question? Its nice they are backdrivable but not normal for regular wormgear mechanism to be able to do this...unless its coupled with something else like a clutch ..also surprized she says the weight is that of a human hand despite all the motors

@dolylupec @wuji_global @TheHumanoidHub @XRoboHub I thought worm gears aren’t back-drivable. We have the first version of the Wuji hand, and those can’t be back-driven. I’m curious what they modified to make it back-drivable

@meigustas @sarahookr @grok he went to school in Switzerland and worked in a Patent office... both of those things might have drained all human emotion - but yeah no excuses are acceptable

@sarahookr @grok How did Einstein treat his first wife

Today I learnt how Albert Einstein treated his first wife. And I can’t unsee it. Whattt how does no one mention this.

@NVIDIAAI @manassharmahere

@manassharmahere 🔥 sweet spot for hack!

Let’s build and sail in SF! My stack for the morning: 1. Nebius AI Cloud - to run Nvidia Cosmos 3 on NV link GPUs 2. Tavily for research Most importantly 3. Everything running on OpenClaw. Also shoutout to Composio - they are gonna be so cool to integrate in my project. But I wanna have something running first. @agentcommunity_ @ship_builders @nebiusai @composio @tavilyai @openclaw @itsaqd @patonchain @DhruvDiddi @alwaysbecolin

@richard41814 Nice work

Talking to other researchers in the "learning from human video" space early this year, a common observation was that it's hard to show transfer from true in-the-wild Internet video, compared to curated research datasets. In most of these datasets, the humans deliberately move like robots, and hands are tracked with clean 3D labels. The usual starting point for Internet video — run a monocular hand pose estimator on YouTube videos, cotrain with robot data — often doesn't work. In our recent work, we study this "YouTube-type" video setting and try to understand what it takes to absorb egocentric Internet videos into a VLA training pipeline. 1/n

@DA_Stockman Note that those Starlink numbers don't count the enormous Capex of making and launching 5y lifespan satelites... the cost is high and they only stay there for 2y to 5y each.. its a reccuring cost to launch new ones and maintain the constelation

Well, here's some math. Starlink is a profitable business with about $11 billion of sales and $3 billion of free cash flow. It might be worth $75 billion at a frisky multiple of 25X free cash flow. The balance----the space launch business and the AI/data centers in space fantasy----has $7 billion of sales and NEGATIVE -$17 billion of free cash flow. So why is it worth anything, unless you are pricing a dream peddled by sell-side hucksters?! In short, after trading up to $2 trillion based on $75 billion of tangible Starlink value, where's the remaining $1.925 trillion of it? This isn't just the classical mania of the crowds. This is sui generis--- mass insanity in a casino that has been giving a lobotomy by three decades of money-printing madness at the Fed and its fellow-traveling central banks around the planet.

@GaryMarcus Hope more US companies release open source models, Google is doing a good job lately. Llama was #1 before. For now, Qwen,etc from well funded companies inside the People's Republic are best oss models, if say Google or Meta were to release better models,those would be adopted.

Zuckerberg and LeCun’s unilateral decision to open source Llama likely (partly) catalyzed China’s AI industry — and may have done truly massive harm to American business interests. We are now starting to see the consequences.

@yacineMTB Lol are you a kid, or on lots of adderrall, making such confident statements. Robotics (interacting with the physical world) is hard,a lot of science in this domain is yet to be discovered. If you think its easy,you've probably not applied sufficient riggour to thinking about it

The fact that robotics isn't unilaterally solved is just an algorithms gap. All of the silicon valley companies doing "robotics" right now with their data collection meme are on the wrong path entirely. It's surprising that robotics isn't really, actually solved. It's not hard