Sergey Levine

This account is verified. Learn more

@svlevine

Associate Professor at UC Berkeley
Co-founder, Physical Intelligence

Berkeley, CA rail.eecs.berkeley.edu Joined April 2018

144 Following 143K Followers 2.7K Posts

Latest posts 5

Sergey Levine @svlevine 3d

A VLM can figure out semantics: which object to grasp, where to put it. But how do we turn this knowledge into fluent and dexterous actions?

At CoRL on Tuesday, Andy and Will will talk about Flow Reversal Steering: a method for passing VLM "hints" to a diffusion VLA.

Will Chen @ CoRL 2026 🤠 @verityw_ 3d

@tangerinecoder and I will be presenting Flow Reversal Steering at CoRL on Tue, 5–6:30p @ JW Marriott! We'll be at poster 39 at Lone Star (3rd floor). Come by if you're interested in chatting about robot reasoning and how it can guide efficient RL!
flow-reversal-steering.github.io
View quoted post on X

Sergey Levine @svlevine 3d

Come check out the RLT presentation on Tuesday at CoRL! Charles will talk about how we can adapt VLA models with sample-efficient real-world RL.

Charles Xu ✈️ CoRL Verified account @CharlesXu0124 3d

RLT has been accepted for oral at #CoRL2026!

I’ll be presenting on Tuesday at 9am and come find me by the poster at 5pm!

🎙️Oral O4 (9am-9:30am)
📎Poster #64 (5-6:30pm)
View quoted post on X

Sergey Levine @svlevine 3d

We've used the OmniVLA navigation model for multiple research projects at Berkeley. Earlier this year, Noriaki figured out how to make it run in real time with combined edge + remote inference!
asyncvla.github.io

Come check out Noriaki's presentation about it tomorrow!

Noriaki Hirose @Noriaki_Hirose 3d

Tomorrow, we present AsyncVLA, our work on fast onboard inference for reactive navigation, at the poster session (17:00–18:30)
Come visit us at Booth 142. We’d love to chat and hear your thoughts. Hope to see you there!
asyncvla.github.io

@CatGlossop @shahdhruv_ @svlevine
View quoted post on X

Sergey Levine @svlevine 5d

A really fun project led by @KaiTyrusNelson, @tobykreiman: use simulated experiments to create a prior with state information, then align it to real-world experiments (with partial observation). Oddly enough, the theoretical foundations have a lot in common with inverse RL!

Aditi Krishnapriyan Verified account @ask1729 6d

1/ A grand challenge in science: the simulation-to-experiment gap

Simulators model complex systems but introduce model errors; experiments give real-world data but are costly and noisy

Can we leverage both? Yes! We introduce Adversarial Distribution Alignment (ADA), at #NeurIPS2026

ADA aligns generative models—initially trained on simulations (e.g., molecular dynamics with classical force fields)—to full real-world distributions (e.g., protein structures) using only partial, noisy observations (like cryo-EM images)

ADA goes beyond just matching the expectation of observables: it can match full distributions, even in the presence of multiple correlated, high-dimensional observations

Check out our website: kaityrusnelson.com/ada/
And our paper: arxiv.org/abs/2510.02259

In collaboration with @KaiTyrusNelson, @tobykreiman, and @svlevine

View quoted post on X

Sergey Levine @svlevine Nov 5

When running RL with prior data, a behavior constraint is important to trade off exploration vs staying close to known good parts of the data. But the choice of this constraint is delicate -- we don't want to just match all the prior data, but we don't want to wander too far. In our new paper, we propose a new constraint that combines support constraints with continuous penalties to enable excellent offline to online RL performance.

Charles Xu ✈️ CoRL Verified account @CharlesXu0124 Nov 4

While working on RLT (pi.website/research/rlt) @physical_int, it became clear to me that online RL needs the right constraints to guide exploration without hindering improvement.

Excited to introduce 🌟GSBR🌟, a new behavior regularizer that allows RL actor to both (1) drop bad modes and (2) explore optimal actions outside the base support.

(1/n) A thread below 🧵 👇

View quoted post on X