Ryan Chan @Ryan_Resolution

Making robotics fun and intuitive to build with. CEO @ MakerMods https://t.co/iqxjzOSRlB makermods.ai San Francisco, CA Joined May 2019

Tweets

701
Followers

2K
Following

1K
Likes

3K

you don’t need a robot to see physical ai in action, you just need a browser. this week I finally tried something I've wanted to do for a while: design an SO-101 scene in three.js, connect it to a real leader arm, collect teleoperation data, train an ACT model, then port it to

30 60 543 86K 441

1 0 9 1K 3

View Details

Ryan Chan @Ryan_Resolution

5 days ago

@IsaacSin12 @LeRobotHF We need to make snacks that look like this cube

0 0 2 172 0

View Details

Isaac Sin @IsaacSin12

5 days ago

Spent a day with @LeRobotHF LeLab and trained a working SmolVLA policy on a brand new setup, 10 episodes, solid model within 1 hour. Huge congrats to the team, this makes Physical AI much easier to pick up Favorite part: you can run inference on the model while it's still training. Watching the policy improve every few thousand steps is addictive Two small things that'd make it even smoother: • the uv env doesn't come with transformers or accelerate, so inferencing a trained SmolVLA needs a manual install of both • you have to recalibration for every new config

8 4 71 5K 25

View Details

Ryan Chan @Ryan_Resolution

7 days ago

Metal Arms bimanual setup timelapse @IsaacSin12 is a great hand model

5 3 43 3K 7

View Details

Isaac Sin @IsaacSin12

a week ago

We trained π0.5 to do a simple task on our Metal Arm with just 20 episodes of data. Everything collected on our own stack teleop with gravity compensation: github.com/makermods-robo… data collection script: github.com/makermods-robo… openpi for the policy. custom implementation of RTC @ 60hz. The smoothness difference speaks for itself. @Ryan_Resolution @makermodsai

8 17 202 16K 150

View Details

Ryan Chan @Ryan_Resolution

a week ago

@KaiLentit Le chaton Lives Matter = LLM

0 0 4 263 0

View Details

NormaCore @norma_core_dev

a week ago

The next batch of ElRobot Kit samples is packed up and ready to ship! 🚀 Huge thanks to MakerMods, @IsaacSin12 , and @Ryan_Resolution for fabricating them. From Madrid with love- we are super excited to send these out to our creators and see what they build. 🤖

2 7 24 1K 3

View Details

Ryan Chan @Ryan_Resolution

a week ago

@IsaacSin12 @makermodsai Check out the MakerMods Metal Arms here: makermods.ai/metal-arm

0 0 2 192 0

View Details

Ryan Chan @Ryan_Resolution

a week ago

3Kg Payload 🦾 MakerMods Metal Arm

5 5 95 13K 13

View Details

Vector Wang @VectorWang2

a week ago

Introduce EgoInfinity: a web-scale (14.6 yrs, 142M clips) data engine that automatically lifts Youtube videos into 4D hand-object interaction, and retargets for robot learning Not only a dataset, but a modular, upgradable engine HF Space: huggingface.co/spaces/Rice-Ro…

20 62 359 106K 315

View Details

Ryan Chan @Ryan_Resolution

a week ago

@shengkun_ye @jerica_kuah @LerWilson Thanks for helping us too!

1 0 4 138 0

View Details

Ryan Chan @Ryan_Resolution

2 weeks ago

@theonlyAyo bro you need to try the MakerMods Metal Arms

1 0 1 125 0

View Details

Ryan Chan @Ryan_Resolution

2 weeks ago

@GZinMetaverse Good work!

1 0 1 702 0

View Details

Grace Zhang @GZinMetaverse

2 weeks ago

For the past 3 months, I've been heads down building and extending our multimodal data capture rig for Physical AI training. What you're seeing here is a single synchronized capture from our system — RGB, depth (stereo IR camera) feed; real-time 3D hand reconstruction, and a 23×20 tactile pressure grid from sensor gloves, all aligned and timestamped together. Getting here took a lot of unglamorous iteration. We reworked the pipeline to deliver clean, gap-free depth at scale after identifying early depth frame drops caused by data throughput bottlenecks in our capture box. We built per-channel calibration models that convert raw resistance signals from our flex sensor gloves into accurate, drift-corrected force values — solving the creep behavior that made a steady grip falsely appear to fade over time. Even though we’ve stayed quiet on socials, word of mouth spread — our rigs have already been deployed in data factories, manufacturing lines, and homes across the US, Europe, China, and Southeast Asia by our customers and ourselves. In the next few days, I'll be sharing more about how we're tackling rights-cleared data capture from the hardware level, and our closed-loop data-model flywheel — using model evaluation results to identify gaps and guide what data our customers need. More soon, stay tuned…

29 27 307 52K 215

View Details

Ryan Chan @Ryan_Resolution

2 weeks ago

@sean_pixel microsd ❌ Picosd ✅

0 0 2 314 0

View Details

Ryan Chan @Ryan_Resolution

2 weeks ago

@theonlyAyo No better way to learn how to build XLeRobot

0 0 1 334 3

View Details

Ayo @theonlyAyo

2 weeks ago

x.com/i/article/2065…

2 6 20 2K 8

View Details

Nikhil Kumar @nikhilkr

2 weeks ago

A small milestone at @GoDrift_ai this week - we received a new robot arm from @makermodsai and had it executing tasks within hours of opening the box. Two things made that possible: > Makermods built hardware that's genuinely easy to assemble. It's rare in this space and worth calling out > Drift is designed so that new hardware becomes useful immediately and not after weeks of integration We'll be running this arm on real-world tasks over the coming weeks and sharing what we learn, the good, the broken and the surprising. Stay tuned!

3 5 22 3K 4

View Details

Ryan Chan @Ryan_Resolution

2 weeks ago

@nikhilkr @makermodsai It was a mad rush on our side to ship the arms over. And @GoDrift_ai executed this in record speed. Really impressed by this, looking forward to partnering up together to make more breakthroughs

1 0 1 202 0

View Details

Ayo @theonlyAyo

2 weeks ago

Introducing Kits by Tnkr 📦 The fastest way to clone a robot. For Maintainers: distribute your project, BOM to kit in one click. For Builders: everything you need to build it, in one box. tnkr.ai