Vineeth Yeevani

@mattparlmer couldn't agree more. just use more arms if you need more dexterity

Raptor 3 robotic hand

April tag cube. Fully 3d printed! Gonna make a UMI data collection device using this

@antopatrex1 What’s the fix?

VCs are pouring billions into robotics right now and most of it is making things worse. inflated valuations, talent wars that price out real builders, and a culture that rewards demo videos over working hardware. the money isn't the problem. where it goes is.

@JoeClinton02 What’s the reach?

Most low-cost robot arms aren’t good enough for real manipulation research. So I built one that is. This is GEM: the Good Enough Manipulator. <$500 7 DOF 1.2 kg payload 3D printed head + wrist cameras LeRobot support low-cost leader arm Print files, instructions, BOM + LeRobot fork are free: joeclinton.me/gem Thanks to @pollenrobotics for PincOpen and @pepijn2233 for open-arms-mini.

@autobrik @robonet_ just use Apriltags. why do you need a full vr controller for this?

Why do people keep collecting data with teleoperation when a bimanual robot setup costs more than $10k? Isn't there a solution that gives you the same data quality without the robot? At @robonet_, we want to build the Internet of Robotics. As part of that mission, we built HandUMI, a hand-worn data collection device for bimanual arms with parallel-jaw grippers. Specs per unit: - 276.5 grams - $110.68 - Encoder-precision gripper aperture - Integrated wrist camera - Tracking with the VR headset of your choice (Pico/Quest) - More than 5 grippers supported (Piper, Trossen, ARX, Soft gripper, Dream gripper) The best part: all the hardware is open source! Thanks @fdotinc for the hardware lab and the space to make this possible. ft. @alvax64 @leoperzz @raulb4s @mbrq_13 @BryanBRstds @Aryan_Mangla_ , and the rest of the @0xnonhuman team.

why do so many ai companies that have nothing to do with hardware have hardware names/logos. factory, mechanize. please stop doing this, it's hella confusing.

I've been stopped at the airport three times for my 3d connexion space mouse. No-one has questioned my Vision Pro, my random exposed circuit boards, or a full screw driver set. make it make sense please.

5ft tall robot feels different irl than in CAD

first draft umi controllers + compatible feetech its 3215 end effector

extrusions are adult legos

opinion piece: cargo cults, data flywheels, and novelty pumps tl;dr: “deployment data flywheels” miss that scale should provide novelty, and deploying robots on a known task wrings most novelty out of its generated data your ops team’s novelty pump is the only way out 🔗👇

@nymbusjp Get to start with strong foundation models for vision and reasoning. Could just fine tune alpamayo and get pretty far.

I am geniously impressed by Rivian and RJ. Tesla needed over 100,000 H100s running 24/7 inside a custom GPU cluster using the real silicon for bit exact inference during training, years of feedback from their over a million FSD users to improve their models, and over 10 billion supervised miles of data. And Rivian's going to do it. Without any of this. They must be geniuses.

@NickBuilds11 Looks great! How many training samples did this policy take?

And there it is!! Consistent, autonomous crumbled shirt fold complete!

@krea_ai allow uploading 3d models. really disappointing that you don't have this already. would be absolutely killer to pair with realtime for product videos

@BambulabGlobal Please be a way to do continuous carbon fiber.

Ready. Set. neXt. April 14, 2026, 4 PM CEST bambulab.com/launch-0414

@physical_int What about the other direction. Does pre-training large VLA on human data lead to alignment on robot data? Or does this go against the business model

We discovered an emergent property of VLAs like π0/π0.5/π0.6: as we scale up pre-training, the model learns to align human videos and robot data! This gives us a simple way to leverage human videos. Once π0.5 knows how to control robots, it can naturally learn from human video.

Depth anything v3 synthetic data trained teacher paradigm is really interesting.

Idk why I’m always skeptical about sim for robotics. Sim/synthetic powers everything in vision.

Felt the exact same way