Will Huey

incoming phd @uwcse, undergrad @cornell. robots and rl.

[Chuning Zhu @chuning_zhu]

Can we get robots to think before they act? In Latent Memory Palace, we achieve this through a classic principle: variational inference. (1/8)

[Matthew Hong @matthewh6_]

RL can't teach an LLM to solve problems it never solves. Robot policies suffer from exactly the same limitation.

The fix turned out to be almost embarrassingly simple: train the policy with diffusion noise during pre-training.

That's it.

The policy covers a much wider action distribution, RL finally has somewhere to search, and we fine-tune VLAs on real robots in under an hour.

Introducing TMRL.

🧵(1/9)

[Maximus Pace @maxpace1_]

Teaching a robot shouldn't require humans to act like robots.

Human demonstrations contain valuable signal for robot manipulation, but they aren’t directly transferable to robots.

X-Diffusion learns from noisy human demonstrations while staying within the robot’s capabilities.

[Will Huey @willhuey9]

@yu_raymond5 We wouldn’t be anywhere near this point without incredible guidance from @abhishekunique7, as well as super helpful feedback from @mukadammh and Anusha Nagabandi over at Amazon FAR

[Will Huey @willhuey9]

It’s becoming increasingly common to evaluate policies in sim. Policy improvement is harder. We show why vanilla RL learns to exploit the sim2real gap, and provide a simple and principled solution.

[Abhishek Gupta @abhishekunique7]

Policies trained on real robot data via imitation can be surprisingly capable. But for domains like dexterous manipulation, they are often not quite good enough: they move slowly, miss grasps, make unreliable contact, and fail under small perturbations.

Can we improve them without any additional data collection on the real robot?

In SCORE, we show that we can improve real-world diffusion/flow policies cheaply by using simulation to simply learn how to steer them on deployment. This leads to large gains in real-world success and speed across a variety of tasks, without requiring additional real-world experience:

https://t.co/hbhId2qbB1

🧵 (1/10)

[Will Huey @willhuey9]

Shoutout @yu_raymond5 for pushing this project to its absolute limit. Could not have asked for a better co-lead (and he’s somehow an undergrad)

[Raymond Yu @yu_raymond5]

Potentially obvious finding: naively taking a policy trained on real-world data and fine-tuning it with RL in simulation can produce quite dangerous behavior…

https://t.co/e7K9GzLmAU 🧵👇

[Sean Brynjólfsson @jolfss]

Companies are racing to sell “privacy filters” for LLMs, but "detect all names and emails" is not privacy and our paper RedactionBench shows why!

Over 200 realistic docs, 11 domains, 53K annotations and 35 models eval'd:
> Best: Opus 4.6
> Specialist: OpenAI Privacy-Filter 🧵/7

[Tyler Westenbroek @ty_westenbroek]

Real-world RL is still too brittle and data-hungry for long-horizon, contact-rich tasks.

We introduce Simulation Distillation (SimDist), which turns large-scale simulated experience into reusable world-model priors for rapid real-world adaptation.

By combining online planning with dynamics adaptation, SimDist achieves high success rates on tasks requiring precision, force, and reactivity.

Play with our interactive visualization to see for yourself: https://t.co/qFGNySxdAl

(1/n)

[Patrick Yin @patrickhyin]

We’re releasing OmniReset, a framework for training robot policies using large-scale RL and diverse resets for contact-rich, dexterous manipulation.

OmniReset pushes the frontier of robustness and dexterity, without any reward engineering or demonstrations.

Try the policies yourself in our interactive simulator! https://t.co/3hW3nYx2vD

(1/N 🧵)

[Will Huey @willhuey9]

Come see the pitfalls of aligning video frames, and my first ever paper at #ICML2025. Huge thanks to Yuki for helping me with all the nuances of academic research, she was an outstanding mentor and co-lead on this.

[Yuki Wang @YukiWang_hw]

Using OT to define rewards for imitating video demos is popular, but it breaks down when demos are temporally misaligned—a frequent challenge in practice.

We present ORCA at #ICML2025 , which defines rewards by aligning sequences, rather than matching individual frames via OT.