Data and Simulation Infrastructure for Physical AI
Santa Clara, U.S.A.
- Tweets 210
- Following 69
- Followers 2,462
- Likes 137
Lightwheel is heading to #IROS2026. Booth 714, Pittsburgh, Sep 27 to Oct 1.
🔹 Learn: EgoSuite-Open100K, 100K hours data of humans doing diverse and real tasks
🔹 Train: SimFoundry, simulation worlds where robots practice
🔹 Evaluate: RoboFinals, a standard exam for robot policies
🔹 Deploy: RoboStack, from sim to real robot deployment
1
5
357
After hours, join our joint party with @ADI_News and @DexmateAI.
Sep 29, 6 to 10PM, Pittsburgh. Bring your hardest sim-to-real question.
RSVP: lightwheel-iros.netlify.app/…
58
Lightwheel is hiring for 2027.
We build the data and simulation infrastructure for Physical AI. Come help us take it global.
Roles
→ Strategic Product Manager, Early Career (GTM)
→ Strategic Projects Manager (Global Operations)
Who can apply
→ Full-time: graduating Sept 2026 to Aug 2027, in CS, Robotics, Automation, AI, or related fields
→ Internship: current students
📍 SF Bay Area
🔗 lightwheel.ai/careers
#PhysicalAI
3
426
Sim to real. Summit to party. 🎉
The humanoid robotics world is in Seoul next week, and on Wednesday night it's all in one room. Builders, researchers, and investors, off the conference floor and talking sim2real. 🤖🍻
It caps our day at @HumanoidsSummit, after @bgxc's Opening Keynote and demos at Booth 16-21.
🗓 Sep 23, 5–10 PM
📍 Seoul (venue shared upon RSVP)
Spots are going fast. Apply now 👉 lightwheel-seoul.netlify.app…
1
301
Lightwheel is at @HumanoidsSummit Seoul, Sep 22–23 at COEX 🇰🇷
Sep 23, 9:00 AM: @bgxc delivers the Opening Keynote, "Building the Continuous Learning Infrastructure for Physical AI."
Booth 16-21: see the full loop, from human data to deployment. Our team is on-site all week
1
2
6
316
The best conversations at IROS never happen in the sessions. They happen later that night, plate in one hand, with the people who make this field worth being in.
@LightwheelAI x @ADI_News x @DexmateAI are co-hosting a party on September 29, 6-10PM, in Pittsburgh. Good food, a proper bar, four hours off the clock.
Bring your hardest sim-to-real question. We'll handle the rest!
#IROS2026 #Robotics #PhysicalAI #Dexmate #AnalogDevices
2
1
14
855
We started in simulation. Egocentric data is where that led us.
Our co-founder @bgxc explains on @RoboPapers why scaling sim means scaling real data first, and why egocentric data was the only source that could be standardized.
Watch the full episode with @micoolcho and @DJiafei here: piped.video/ZywjONaTHRg?si=bAN5…
#Robotics #Real2Sim
4
1
14
1,312
Learn More:
Project page:
egosuite100k.lightwheel.ai
Blog post:
huggingface.co/blog/Lightwhe…
Data on HuggingFace:
huggingface.co/collections/L…
221
Why does a simulation company open-source 100,000 hours of egocentric data?
Because training data is scaling fast and evaluation isn't. @bgxc explains on RoboPapers: ego data is the most scalable way to train, simulation is the only scalable way to test. EgoSuite-Open100k is half of that; RoboFinals is the other half.
Thanks @micoolcho and @DJiafei for the conversation.
Robotics has a data problem, and egocentric data is a compelling way to solve it because egocentric data is (1) scalable — it’s cheap to collect — and (2) easily captures the breadth and diversity of real human tasks, while (3) being true to real physics. Lightwheel recently released a hundred thousand hours of high-quality egocentric video data, spread across a wide range of human tasks from cleaning and tidying to assembly and construction to retail tasks.
Here to tell us about this tremendous effort is @bgxc, CEO and co-founder of Lightwheel. The data is also available on HuggingFace right now.
Watch Episode 102 of RoboPapers, with @micoolcho and @DJiafei, to learn more!
5
6
72
11,127
Lightwheel retweeted
Full episode dropping soon!
Geeking out with @bgxc on EgoSuite-Open100K egosuite-oepn-100k-test.ligh… @LightwheelAI
Co-hosted by @micoolcho @DJiafei
4
21
6,845
We were on the ground this week at LEAP x DeepFest AI 2026.
Martin Elbs spoke on Building the Continuous Learning System for Physical AI, and how closing the loop between data, simulation, evaluation, deployment, and real-world feedback can accelerate how robots learn.
Louis Lian joined the panel How Generative AI is Redefining Robotics at Scale, discussing how advances in generative AI are changing how we train, evaluate, and ultimately deploy robots at scale.
And the Lightwheel team is at the booth connecting with the Physical AI community. Great conversations and a lot of momentum around the future of robotics.
If you missed us, we'll at @HumanoidsSummit Seoul in a few weeks!
#LEAPxDeepFest2026 #PhysicalAI #Robotics
13
1,151
Scale alone doesn’t make egocentric data valuable. Diversity does.
EgoSuite-Open100K, spans 15k+ tasks and 15k+ scenes across 7 environment categories and 18 task categories.
The first 10k hours are open now, with more coming as we build toward 100K.
egosuite100k.lightwheel.ai/
2
5
28
3,116
Most egocentric data at scale is private. We think that's worth changing.
With @huggingface, we've released the first 10,000 hours of EgoSuite-Open100K — an open egocentric human dataset for Physical AI.
huggingface.co/blog/Lightwhe…
#Robotics
10
26
167
11,688
The full release will total 100k hours across 15k+ tasks and 15k+ real-world scenes.
128 scene types across 7 environments.
Hand pose, body pose, and event-level semantic annotation, released in LeRobot v3 and MCAP.
2
3
628
Not sure where to start?
EgoDemo is a 50-hour sample covering every annotated subset plus two raw-video variants.
Explore the full collection here: huggingface.co/collections/L…
4
260
Lightwheel retweeted
100,000 hours of human hands doing real work. LeRobot format, fully annotated and open.
EgoSuite-Open100K from @LightwheelAI X @huggingface: 15,000+ tasks across 15,000+ real scenes, hand + body pose and wrist cams.
Recorded across 7 environment categories and 128 scene types, from kitchens and bedrooms to warehouse floors and assembly lines.
It's open for academic research and commercial training, and streams straight from the Hub. Always great to see a dataset at this scale released openly on the Hub.
More info: hf.co/collections/Lightwheel…
26
80
728
45,404
The largest fully annotated open egocentric human dataset.
Today we're open-sourcing EgoSuite-Open100K with @huggingface : 100,000 hours of egocentric human data, free for anyone to use.
Physical AI has a scaling law of its own now, and its input is human data. That part is settled. What the field still needs is a shared foundation to build on.
15,000+ tasks. 15,000+ real scenes, from factory floors to retail backrooms. Hand pose, body pose and subtask-level semantics, with wrist camera coverage on part of the set. Licensed for commercial training, not research-only.
We build the data infrastructure for Physical AI. Open100K is the first public layer of it.
The first 10,000 hours are live now, with the rest rolling out in stages.
17
33
314
74,478
Project page: egosuite100k.lightwheel.ai/
Dataset on Hugging Face: huggingface.co/collections/L…
If you train on it, tell us where it falls short. That's what decides what we open next.
2
2
17
4,329