Building Human-embodied Intelligence. CEO @MyoLabAI | Sr. research scientist @OpenAI @GoogleAI @AIatMeta | @berkeley_ai @UWcse #MuJoCo | Ad. Prof. @CMU_Robotic
NewYorkCity
Joined February 2016
- Tweets1.9K
- Following927
- Followers7K
- Likes3.8K
Pinned Tweet
📢Life is a sequence of bets – and I’ve picked my next: @MyolabAI
It’s incredibly ambitious, comes with high risk, & carries unbounded potential. But it’s a version of the #future I deeply believe in.
I believe:
➡️AI will align strongly with humanity - coz it maximizes its own growth & impact
➡️It will transform the world as profoundly as the internet
➡️Like the internet, it will ultimately disappear into the background of our daily lives
Most of what we see today are transient wins - short-term products riding the first waves of capability. Not transients, I’m betting on the signals that will endure.
Just as the cellphone became the personal gateway to the internet era, I believe the future of AI will be 𝐩𝐞𝐫𝐬𝐨𝐧𝐚𝐥𝐢𝐳𝐞𝐝, 𝐜𝐞𝐧𝐭𝐫𝐚𝐥𝐢𝐳𝐞𝐝, & 𝐝𝐞𝐞𝐩𝐥𝐲 𝐡𝐮𝐦𝐚𝐧-𝐜𝐞𝐧𝐭𝐫𝐢𝐜. The interface—the #canvas—of this era is still waiting to be defined.
With MyoLab, I’m placing my bet on the 𝐥𝐢𝐟𝐞𝐥𝐢𝐤𝐞 𝐡𝐮𝐦𝐚𝐧 𝐝𝐢𝐠𝐢𝐭𝐚𝐥 𝐭𝐰𝐢𝐧 as that interface.
We’ve assembled a world-class team with the conviction and grit to make this future real. We’re building a new kind of AI: embodied, personal, and lifelike. Most already believe lifelike digital twins are inevitable. We’re just accelerating the timeline.
Today, we’re releasing an early research preview of the first instantiation of #HumanEmbodiedIntelligence at myolab.ai
We’d love for you to try it and share your feedback.
𝐓𝐡𝐢𝐬 𝐢𝐬 𝐦𝐲 𝐛𝐞𝐭. 𝐖𝐡𝐚𝐭’𝐬 𝐲𝐨𝐮𝐫𝐬?
All forms of intelligence co-emerged with a body, except AI
We're building a #future where AI evolves as your lifelike digital twin to assist your needs across health, sports, daily life, creativity, & beyond...
myolab.ai ➡️ Preview your first #HumanEmbodiedAI
A breath of fresh air
Contact rich manipulation needs a force and physics first approach.
Congratulations @EkaRobotics team, @pulkitology @haarnoja
@EkaRobotics has raised a $150M Series A.
Generality v/s Performance is the fundamental tension at the frontier of robotics. Generality without performance is just a prototype. Performance without generality is just a single-purpose tool.
Customers shouldn’t have to choose. Eka breaks the tradeoff. Powered by our Vision-Force-Action (VFA) model, we deliver the performance and versatility customers need to put robots to work every day.
Eka is built on three convictions:
⚡️Force is the language of manipulation.
🚀 Reinforcement learning is how dexterous manipulation scales.
🎂 System 1—fast, reliable physical execution—is the bottleneck to solve.
With this funding, we will scale our model and accelerate customer deployments.
Grateful to be building alongside a dream team and investors who share our ambition.
Heading to ✈️ #IROS2026 🤖
Looking forward to a few days with the robotics community exploring a question I’ve been obsessed with for two decades:
How do we build physical agents that can learn and act with the richness of human physical intelligence?
We’re bullish on studying (a) how humans move, interact with the physical world, and acquire complex motor skills, and (b) translating those foundations into embodied intelligence for physical agents.
Looking forward to coming back to @CMU_Robotics, reconnecting with old friends, meeting new collaborators, and hearing what everyone is building.
See you in the City of Bridges — Pittsburgh! 🌉
In London for RoboSim Summit at @GoogleDeepMind for next 4 days.
Looking forward to connecting with people building in physicalAI.
✈️ I'm in LONDON🎡 Sept 14-17th
Who should I meet? I'd love to connect with folks building in robotics / physical AI space.
DMs open
Another week, another robotics map! 🇬🇧
This time, we will take a closer look at the busy streets of London and see what robotics companies are located there.
London has excellent engineers and researchers, especially from universities like Imperial College London and UCL, which are well known for robotics, AI, and engineering. Many robotics founders and early employees come directly from these universities.
London is home to @GoogleDeepMind, one of the world’s leading AI labs. Its work on robot learning, control, and general AI has helped push forward how robots learn and adapt in the real world.
The city also has one of Europe’s strongest investor ecosystems. London is a major global finance hub, so it’s easier to find venture capital, corporate investors, and early customers, especially for robotics companies working in areas like logistics, healthcare, and automation.
It is very international and business-friendly. It’s easy to hire talent from around the world, set up a company, and sell globally.
→ @TheHumanoidAI builds general-purpose AI-driven humanoid robots capable of physical tasks across domains.
→ Automata Tech develops easy-to-deploy robotic automation hardware and software for SMBs to reduce manual labor.
→ @shadowrobot creates advanced dexterous robotic hands and manipulation systems for research and industrial automation.
→ Paddington Robotics designs small autonomous robots for retail and service environments to assist staff.
→ @KAIKAKU_AI builds robotics and AI enterprise solutions to optimize warehouse and logistics processes.
→ Automated Architecture (AUAR) develops spatial computing and robotic systems that blend physical and digital environments for construction and design.
→ @recycleye creates AI-powered robotics that identify, sort, and automate recycling and waste processing, and has raised ~$20M+ in funding.
→ @Neuracore_AI builds AI-based perception and planning software for autonomous robots, and has raised $10M+ in venture funding.
→ @apianhealth_ develops autonomous robotic systems for automated medication dispensing and hospital logistics.
→ @SlamcoreLtd offers high-performance SLAM navigation and vision software to help robots map and localize in complex environments, and has raised ~$6M+.
→ @MoleyRobotics builds fully automated robotic kitchen systems (“robotic chef”) and has raised tens of millions in funding (reports ~$30M+).
→ @dexoryHQ develops autonomous warehouse robots and AI software that continuously scan inventory and turn it into real-time operational insights, and has raised $80M Series B and a $165M Series C & growth round.
→ Extend Robotics builds tele-operation technology for robots, and helps with orchestrating the fleets of robots.
It seems that if you wanted to explore London from the perspective of robotic startups, it would take a few days!
What city should I do next? :)
~~
♻️ Join the weekly robotics newsletter, and never miss any news → ziegler.substack.com
✈️ I'm in LONDON🎡 Sept 14-17th
Would love to meet & learn about robotics / physical AI space. DMs open
Incredible results from our colleagues at @GoogleDeepMind. Gemini Robotics 2 is half a step forward.
🙂GOOD:
- Dexterous knot tying is WOW. Especially impressive if this behavior is genuinely emergent rather than heavily represented in the demonstration distribution
- On-device policies are a big deal
- Whole-body + dexterous control is a substantial improvement over previous releases
😔BAD:
- I think there’s still a lot of confusion around the roles of System 2 reasoning (Gemini/ER) vs System 1 task execution (VLA/policy)
- Some of the showcased behaviors look less emergent than the presentation suggests, and more consistent with what one might expect from a policy trained on a rich teleoperation distribution. I’d love to see stronger evidence either way
- And ultimately, the success rates shown still look quite low for the claim of broadly capable robotics
🤮UGLY:
- Google really need to get out of labs
- While much of the robotics community is fighting the ugly, messy problems of deployment in the real world, Google still seems heavily anchored in the GoogleVerse: carefully staged environments, carefully selected tasks, controlled fake objects, controlled scenes
Grapes can be sweet or sour. Real-world robotics is beautifully messy — if you decide to step into it.
✋HELP !
Can Anyone Lend Me a UR?
Our robot deliveries has hit a snag. I need a temp robot in NYC to meet deadlines.
(every lead helps, please repost! 🙏🤖)
Vikash Kumar ✈️IROS2026 retweeted
💪Force is the language of the physical world
Let alone multi-end-effector behaviors, impressive is the GEN-1 peeling the tape off the table. Its demonstrating an understanding of the force. Such a reasoning is refreshing, and very different from what's needed for contact-rich manipulation of small objects.
Kudos @peteflorence @andyzengineer and entire @GeneralistAI team
💪Force is the language of the physical world
Let alone multi-end-effector behaviors, impressive is the GEN-1 peeling the tape off the table. Its demonstrating an understanding of the force. Such a reasoning is refreshing, and very different from what's needed for contact-rich manipulation of small objects.
Kudos @peteflorence @andyzengineer and entire @GeneralistAI team
The first robot hand with -- Heart , Head, and Life Line!
My heart is melting on the attention to details 🫶
Kudos @BerntBornich @radbackwards & @1x_tech team
The biggest gap in robotics is - deployment
Everyone in the Bay is building a robot company. Almost nobody is talking about the real bottleneck.
After 2 years running the Robotics Center of Silicon Valley, seeing hundreds of frontier labs and stealth startups, one pattern is undeniable:
The blocker isn't model capability. It’s the sim-to-real loop.
1. Simulation ≠ Hardware Reality A VLA trained in Isaac Sim or MuJoCo hits a wall when it encounters real-world actuator limits: thermal throttling, joint backlash, and sensor drift. Even two units of the exact same robot model have completely different dynamics.
2. Teleop data is inherently off-policy Human demonstrations bootstrap your model, but your model isn't human. That distribution gap doesn't disappear with more of the same data. It only closes through real-world rollouts, failure capture, and targeted data collection.
3. "Working" is undefined until deployment Teams optimize for lab success rates, then fail on real-world edge cases: a wrinkled shirt, 4 PM glare, or a pallet 3cm off spec. Without a real-world evaluation harness, you are just guessing.
4. The solution? Treat RL² as your core loop Internally, we call this Reinforcement Learning in Real Life.
Deploy ➔ capture failures ➔ evaluate against real criteria ➔ collect targeted data ➔ redeploy. Continuous learning on real hardware.
5. The Takeaway The winners in Physical AI won't just have the biggest models. They'll be the teams that can close the real-world learning loop the fastest. Closing the sim-to-real gap is a 2-year infrastructure problem. That’s why we built our stack at the Robotics Center.
6. Come hang out If you're hitting these exact walls, let’s compare notes.
We host robotics builders at 90 Welsh St (SF) every Friday. Drop by.
Highlighting a few papers from our group at ICML this week
#ICML Two last-authored papers at ICML 2026 this week.
A Very Big Video Reasoning Suite.
Video models today are optimized for one thing: visual quality. Whether they can reason, track causality, follow constraints, execute multi-step plans, has been nearly impossible to study, because the field never had the data. VBVR changes that.
Over 50 researchers and engineers across 30+ institutions built:
1) 150 reasoning tasks grounded in cognitive and developmental psychology
2) 2,015,000 samples and over one million videos, roughly 1,000× larger than the nine prior video-reasoning datasets combined
3) every task procedurally generated with deterministic ground truth, so evaluation is fully rule-based and verifiable, no LLM judges, and it agrees with human judgments
Insights:
1) Fine-tuning an open-source 14B model on VBVR lifts overall performance by 84.6 percent, a new state of the art that surpasses every proprietary system we evaluated, including Sora 2 and Veo 3.1.
2) The learned skills generalize to out-of-domain task families the model never saw, early evidence of emergent generalization in video reasoning.
3) And the honest part: gains saturate with scale, on two different architectures. A persistent gap to human performance remains that data alone cannot close. The bottleneck is architectural.
4) One principle kept surfacing throughout: controllability, doing exactly what is asked — is the bedrock of verifiable video reasoning.
Great thanks to my collaborators @MaijunxianWang @Rui147000038622 @juyi_lin @SaraJiran @thwiedemer @Kaiagaoqy @carrot0817_ @RubyFreax Lianyu Huang Zelong Hong Jiahui Ge Qianli Ma Hang He Yifan Zhou Lingzi Guo Lantao Mei Jiachen li Hanwen Xing Tianqi Zhao Fengyuan Yu Weihang Xiao Yizheng Jiao Jaden Hou Danyang Zhang Pengcheng Xu Boyang Zhong Zehong Zhao Gaoyun Fang John Kitaoka Yile Xu Hua Xu Kenton Blacutt @tin_ng_qn @siyuansong_ @HaoranSunMr Shaoyue Wen @LinyangNeuroAI @RunmingW @Yanzhi_Wang_ @Mengyue_Yang_ @ziqiao_ma @raphaelmilliere @fredahshi Nuno Vasconcelos @DanielKhashabi Alan Yuille @du_yilun @ZimingLiu11 @Brian_Bo_Li Dahua Lin @liuziwei7 @Vikashplus Yijiang Li @drowsyleilei @caizhongang
The second paper — "Vision Language Models Cannot Reason About Physical Transformation" — asks a question Piaget asked children 70 years ago. Pour water into a taller, thinner glass: is it the same amount? Seven-year-olds pass this reliably. We built ConservationBench of 384 real videos of conservation tasks with matched non-conserving controls, and evaluated 112 VLMs across 23,040 trials. Humans score 98 percent; models sit near chance, with 82 of 112 under 10 percent on the strict paired test. The most telling result: models answer "better with no image at all" than with the actual video, and chain-of-thought prompting makes things worse, showing they're leaning on a textual prior, not tracking object state through change.
#videomodel #AI #research #reasoning
🫴 Gifting 5 free desks in NYC.
We just moved into a larger office and have 5 extra desks sitting empty for the rest of the month.
If you're an early-stage founder or startup team and a free workspace in the heart of NYC would help, send me a DM.
No catch—just paying it forward to builders who need it.
Vikash Kumar ✈️IROS2026 retweeted
🙂GOOD: contact rich, WBC, real-time, OSS
😔BAD: staged scenes & tasks, no proof of genrlztn
🤮UGLY: behavior seems pre-canned kinematic movements, lacks task internet, multi-cut video hiding key movements
Impressive progress !!
X-Humanoid unveiled TG-VLA, which it calls the world's first whole-body VLA framework for full-size humanoid robots.
The goal: move humanoids past "mobile dual-arm systems" to human-like, whole-body coordinated operation. Built on three technologies:
- HEX: an open-source full-size humanoid VLA that learns across embodiments and coordinates the whole body. Claimed SOTA on real-world whole-body manipulation
- HAF-VLA: hierarchical action flow that decomposes whole-body motion into subtasks. Best or joint-best on 6 benchmarks
- DSRL-DCT: online RL for high-DoF humanoids in a compressed latent space. 100% success on mobile manipulation
HEX is open-source.
🙂GOOD: contact rich, WBC, real-time, OSS
😔BAD: staged scenes & tasks, no proof of genrlztn
🤮UGLY: behavior seems pre-canned kinematic movements, lacks task internet, multi-cut video hiding key movements
Impressive progress !!
X-Humanoid unveiled TG-VLA, which it calls the world's first whole-body VLA framework for full-size humanoid robots.
The goal: move humanoids past "mobile dual-arm systems" to human-like, whole-body coordinated operation. Built on three technologies:
- HEX: an open-source full-size humanoid VLA that learns across embodiments and coordinates the whole body. Claimed SOTA on real-world whole-body manipulation
- HAF-VLA: hierarchical action flow that decomposes whole-body motion into subtasks. Best or joint-best on 6 benchmarks
- DSRL-DCT: online RL for high-DoF humanoids in a compressed latent space. 100% success on mobile manipulation
HEX is open-source.
😀GOOD: Dexterous, cheap
😔BAD: Tendon driven
🤮UGLY: Realiability, Transmission inertia & friction, Tendon slack & break, hard to repair
I have bought the most @shadowrobot hands on this planet. They have been solving for the 🤮bits for 3 decades. Still not production grade
Today #opensource dexterity became accessible to everyone! 🤯
just follow along the step-by-step assembly video and build your own 17 DOF orcahand v2 from our $1,700 BOM kit
At this point I’m curious why platforms like blender, maya, etc won’t just add a plugin for generative rendering ?
