Pinned Tweet
A token called $TETHERIA on pump.fun and the account TetherIA_ai are using our former name. Chestnut Robotics (formerly TETHERIA) has never issued any cryptocurrency or token and has no connection to them. This is our only official account. Do not buy or trade it.
Just got back from @HAND_ERC, and the team is on a roll!
We’ll be at @corl_conf this week. Spot someone in this T-shirt? That’s us. Come say hi to try Aero UMI, check out Aero Hand, talk collaboration, or debate whether a rigid, co-designed skeleton is really necessary. Let’s see who wins 😉
We’ll also be at the UMI Arena Device Exhibition on Thursday. Come find us there!
Hot takes from @siyan_hammer at @HAND_ERC. Thanks for having us! We learn from academia and want to give back by building tools researchers can use.
If your lab needs high-fidelity, contact-rich data and a hand designed with deployment in mind from day one, let's talk.
Back from @HAND_ERC! Some hot takes from presentations and conversations with researchers.
1/ Dexterity, robotic hands, and human interfaces are deeply connected.
We can't wait for one problem to be solved before tackling the others. The whole system needs to be designed and improved together from day one.
Please stop asking us whether hands are for research or deployment. Research and deployment are inseparable.
Higher DoF doesn't necessarily mean better manipulation. Before showing off how many degrees of freedom your robot hand has, show us it can perform reliably, again and again.
Reliability first.
Replying to @EmbodiedCoX
7/ Every extra moving part is a liability
A 21-DoF hand worked in the lab and broke on day one on site. @ChestnutRobotic limits degrees of freedom on purpose. Sensors break first
#2 from Mohit, another founding engineer at Chesnut:
"What's the next big wave in robot manipulation?"
Reinforcement learning on top of VLA models. But whether RL happens in simulation or on real hardware, it needs high-quality dexterous data to ground simulators and give policies a strong starting point.
That's why we're building Aero UMI: to make collecting robot-ready dexterous demonstrations scalable.
Let's see.
Starting your week with a few takeaways from our team at #BARS last Friday at @UCBerkeley !
#1 From Kezhou, our founding engineer:
“Real2Sim came up again and again at BARS. Does that mean high-precision real-world data is becoming less important?”
Kezhou's take: Quite the opposite.
The more we rely on simulation to scale robotics data, the more it matters what we capture in the real world and how faithfully we can reproduce it in sim.
Precise contact points and matched kinematics make that bridge much easier to build.
Starting your week with a few takeaways from our team at #BARS last Friday at @UCBerkeley !
#1 From Kezhou, our founding engineer:
“Real2Sim came up again and again at BARS. Does that mean high-precision real-world data is becoming less important?”
Kezhou's take: Quite the opposite.
The more we rely on simulation to scale robotics data, the more it matters what we capture in the real world and how faithfully we can reproduce it in sim.
Precise contact points and matched kinematics make that bridge much easier to build.
Chestnut Robotics ✈️CoRL retweeted
We are still at #BARS
Come to demo our UMI and hand by yourself
Come to talk with us on your research project and see how we can help you
Come to talk with us on job openings
Or just come to save my brain with some random but fun convos
Your next robotics experiment shouldn’t stop at the hardware budget.
We’re launching Chestnut’s Education Partnership to help universities and research labs get started with Aero UMI and Aero Hand at affordable research pricing with direct support from the engineers who built them.
Use Aero in your research or teaching. Tell us what works and what doesn’t. No publication requirement. No obligation to commit to a joint project.
If there’s a good fit on both sides, let’s take the next step into co-research together.
Bring your questions. We’ll help you get started.
What would your lab investigate? Tell us here: lnkd.in/g_iYCd2b
Want to talk in person? @JoeDong233873 will be speaking at #BARS @UCBerkeley tomorrow around 3 PM. Come chat with us. We’d love to hear about your research and explore how we can work together!
#RoboticsResearch #RobotLearning #DexterousManipulation
A duck today. What should we grasp next? Researchers, educators, and students, what would you teach a dexterous robot hand?
Our Education Partnership launches tomorrow. At #BARS this Friday? Come tell us what you’re working on and see how we can make a more dexterous world together.
The blade is thin. The problem isn’t.
Why co-design matters for robot dexterity? We’ll show you at Bay Area Robotics Symposium 2026
This Friday at @UCBerkeley 🔪 Don't miss us
We’d love to collaborate with academia and share what we’ve been building! If your class would welcome an industry perspective on robotics, feel free to reach out.
Spent today at UC Santa Cruz @ucsc talking about how robot hands learn, from simulation to human demonstrations.
At @ChestnutRobotic, we're co-designing the hands, sensing, and tools for learning: Aero Hand Open for sim-to-real research, and Aero Hand + Aero UMI to connect human demonstrations with robot actions.
Thank you to Professor Zhicong Yu for the invitation, and Professor Tae Myung Huh and the Tactile Manipulation Lab for having me over!
events.ucsc.edu/event/ece-29…
We're betting on a rigid world. Oh wait, just the hardware. Not the hearts. ❤️ We like our data structured and our world a little softer.
More on why we're betting on rigid 👇
Another long, busy day at the office.
Time for a little afternoon tea break! Enjoy this clip of our Aero Open playing piano alongside @Valley101_Qian at last week's @thevalley101 summit event. 🎹🤖
Woohoo 🎉 Chestnut Robotics has been selected to showcase our Aero system at the Device Exhibition of the UMI Arena Workshop at @corl_conf 2026!
Excited to bring our co-designed UMI and dexterous robotic hands for a live demo. See you in Austin!
At IROS someone said: “Make things look sexier.”
Fair.
Deeper in robotics, we’re convinced: no unsexy work, no industry progress.
Thermals, durability, calibration.
Demos work once. Unsexy work makes it work the 10,000th time.
That’s why we build Aero UMI.
A hand knows more than it can tell.
Aero UMI lets it show.
#IROS2026



![[BARS Recap] The Hundredth Attempt Is the Real Test](https://pbs.twimg.com/media/HWQpp6TWEAAIfVZ.jpg)









![[IROS Recap #2]Why We Bet on Rigid Data Gloves in a World Going Soft](https://pbs.twimg.com/media/HVQVI_uaYAA4MYi.jpg)

