We built a robotics data engine with Astra-level performance at 125x lower cost. And we are open-sourcing the weights today! Introducing Flora, a suite of open models to parse the physical world and turn raw footage into rich training signals.
Flora already powers our in-house data engine, labeling 200K hours of videos a day on 500 GPUs. We created it to scale the frontier for robot learning. Now you own the weights too.
Flora gives you:
- High-quality, timestamped captions to curate and filter human egocentric and robot data;
- Dense progress rewards for quality control and RL post-training;
- Rubrics scoring for real-world benchmarking
Try it today, links in thread:
94
107
969
56,347
Damian retweeted
when i have really dumb questions i'll ask chatgpt or gemini because i don't want claude to think i'm stupid
404
3,778
60,237
1,229,993
Damian retweeted
You really really really really really need to make sure you’re not playing poker with anyone using smart glasses.
Zero tolerance.
154
557
12,749
799,510
Damian retweeted
Introducing... Karen Warfare
Built with Opus 5.5 in 1 day
Make the internet fun again
@threejs #threejs #gamedev #indiegame
236
295
3,448
341,670
Damian retweeted
Bugün yoğun bakımdaki babaannemi ziyarete gittim. Oradayken Claude Opus 5.5 kullanarak hastabaşı monitörüne Doom portladım. Oyundaki can yüzdesini hastanın nabzına senkronize ettim; can 0’a düşerse nabız da sıfıra düşer. Bu da oyunu daha heyecanlı hale getiriyor.
405
2,502
29,792
1,380,297
Damian retweeted
suddenly a project written in Bend with 500 stars, I had no idea?
(this actually makes me anxious because there is so much I want to improve NOW, but I'm stuck doing business 😭)
really cool though, anything done in Bend should automatically get exponentially better soon 🫡
27
18
839
43,201
Damian retweeted
Introducing Hugo: the first human companion robot.
At Boston Dynamics, I was Chief Strategy Officer, bringing robots into the real world.
My co-founder, Farbod Farshidian, led research at the RAI Institute.
Together, 15+ years teaching robots to run, jump, and work.
So we asked: what if robots could connect with people?
Hugo runs on the first Vision-Language-Expression (VLE) model.
He links what he sees and hears to movement, expression, and gesture.
He notices you, reads the room, and reacts in real time.
And he's only the first.
Dynamic Creatures is @BostonDynamics' official entertainment and hospitality partner.
We're building a cast of characters with film studios, resorts, and theme parks.
dynamiccreatures.com
389
467
3,162
1,793,032
What if one of the biggest bottlenecks in robot learning isn’t the policy, but the robot arm itself?
In sim, OpenArm misses 64% of the UMI episodes. YAM 42%. Axol 29%.
I searched 5,040 tabletop arm designs (red in the video = a miss). The best one misses just 9%, with 9 to 21× lower tracking error and 6 to 11× lower joint-speed penalties than those three arms.
And even when the other arms can replay a demo, they still pay for it: 8 to 15× higher smoothness penalties and 7 to 33× higher singularity-conditioning penalties.
Caveats: sim only, same controller for every arm, no learned policies tested yet. Full methodology in the article.
Makes me wonder how much arm kinematics biases which behaviors policies can actually learn from human demonstrations.
20
26
269
17,420
Damian retweeted
MOSS the 3D-printed litter-picking robot is NOW open source:
Build your own, or join the TNKR waitlist for a kit.
Cleaner streets, one maker at a time.
Links in the first comment 👇
35
73
629
34,731
Damian retweeted
What might a spatial harness look like for robotics?
In this recording I show how @Auki makes it possible to get multiple robots and devices collaboratively maintaining spatial context, and how we use augmented reality to interface both with the robots and each other.
10
20
121
13,115
literally insane. you can follow someone on this site, so cracked, and he gives away the alpha for FUCKING FREE
FOLLOW THIS CHUD NOW holy shit man
Someone asked me about the BoM cost of the Negotiator II... All these capabilities only cost $3.8 per board in components (100 qty)! Beat that.
PCB $1.3
Components $3.8
Assembly fees $1.5
Shipping and duties $1.4
————————————
Total $8
I am so proud of this card :)
6
4
296
14,686
Damian retweeted
This is the most impressive result I’ve seen in my PhD.
One policy, trained from scratch, solves all 20 NIST assembly tasks. It can even complete 10 tasks in a row, with a single binary reward only after all 10 are done.
@octi_zhang combined diverse resets from OmniReset, task sampling from SGS, and multi-task training to unlock finer dexterity and longer horizons. No extra GPUs needed: one run on the same hardware replaces 20 specialist runs.
Videos and details: alpha-nist.github.io
Proud to be part of this incredible work! More in his thread 👇
(1/6) From @GoogleDeepmind’s AlphaGo to AlphaZero, we saw machines find moves that reached beyond what we’d imagined for them. No one taught them those strategies. They emerged from reinforcement learning: playing, failing, and trying again under a single binary task reward for success or failure. Inspired by them, we couldn’t help but wonder: what might a robot discover if we gave it that same freedom to learn?
That question led us to AlphaNIST. Starting tabula rasa, with that same binary reward plus only a simple control cost, a single policy solves all 20 @NIST assembly tasks at a 98.2% success rate in simulation. More surprising still, AlphaNIST can complete 10 different tasks in a row, receiving that reward only once, after all 10 are done.
Just as AlphaGo’s strategies surprised its creators, AlphaNIST discovered dexterous strategies we couldn’t have come up with ourselves.
Paper: arxiv.org/abs/2610.13354
Website: alpha-nist.github.io
18
59
888
42,783
wagmi.
42 days ago, I left my job as a founding engineer at a @ycombinator robotics startup and moved back to Oregon to start my own robotics company.
Here's how we got here.
I started HIRO (Humanoid and Intelligent Robotics Organization) as a club at Oregon State because I believed advanced robotics should be accessible to everyone, not just a handful of well-funded labs.
I had preordered a robot from @kscalelabs, but when the company shut down, I reached out to its founder, @benjamin_bolte, about buying one of their prototypes. Instead, Ben talked me out of driving down to Palo Alto and encouraged me to build our own robot for the club.
That conversation stuck with me.
Earlier this year, I graduated from Oregon State and moved to San Francisco to work at Nori Robotics.
Over the summer, we went through three iterations of robots and multiple public launches. I learned a ton about what it actually takes to build and ship hardware.
On September 1st, I came back to Oregon and went all-in on @hiroindustries with my cofounder, @ethan_marreel.
In just six weeks, we've brought Origin, our first two-armed robot, to life. From designing and assembling hardware to getting it moving and sharing our first demos.
There's still an enormous amount to figure out. The hardware isn't perfect, the software is a work in progress, and Ethan's teleoperated beaver throw could use some practice.
Six weeks ago, Origin was still an idea. Today, I get to watch it move in my apartment.
1
2
28
2,170
Tips for agentic CAD: tell your agent to add these checkers and run them after every model update (mine is written in Python using build123d).
1. Spec check: measure the important dimensions of the parts and make sure they reasonably agree with your spec (I maintain a YAML with all high level design/decisions).
2. Motion check: move each joint through its full range in hundreds of positions and make sure no moving part touches another part. Use simple shapes and skip pairs whose bounding boxes do not touch so the check stays fast.
3. Geometry check: find very thin faces, very small faces, very short edges, and walls that are too thin. Make sure a symmetric part is really symmetric. Measure wall thickness by casting rays through the part.
4. Edge check: make sure each edge has a radius. If an edge must stay sharp, write it in a list of permitted sharp edges.
5. Manufacturability check: for machining, make sure a cutting tool can reach all surfaces, find inside corners that are too sharp for the tool, and find walls that are too thin. For printing, measure the minimum wall thickness and the area that needs support. For powder printing, find closed spaces that trap powder.
6. Fastener check: use standard screws and inserts from a library (i use bd_warehouse, the parts library for build123d). Make sure each fastener goes through its hole with clearance, aligns with a threaded hole, and engages the thread to the correct depth.
7. Hole check: make sure each hole has a fastener in it. If a hole has no fastener, write the reason for the hole.
8. Picture check: make pictures and sections of the part after each change. Tell the agent to examine the pictures before it says the work is complete.
9. Accepted problems: give each accepted problem a name and a small area. A large area can hide new problems after the shape changes.
10. Test the checkers: give each checker a part with a known error. Use the checker only after it finds the error.
19
31
367
16,730
quick Backplane demo:
full guitar pedal design + layout w/ SPICE simulation to validate
Backplane is a T3Code + KiCad fork w/ a set of really nice skills we put together.
backplane.works/
16
9
250
26,392
Damian retweeted
Introducing ManimGX, the 3D animation engine for agents.
- Blazingly fast: 90× faster than ManimCE
- Rust + wgpu core
- Compatible with Manim's Python API
- Lightweight: one pip install, no system dependencies
Open source: github.com/academa-labs/mani…
Docs: manimgx.academa.ai
For 40 years we built GUIs for humans. Now we build APIs for agents.
Try it. Paste this into your favorite coding agent:
"Follow manimgx.academa.ai/llms.txt to make me a 3D animation of the solar system, and open it."
91
308
3,335
255,031