Naveen Appiah

@patrickhyin That makes sense! Thanks 🙂

I would say First-Try Success Rate (Real) vs. Policy Success Rate (Sim) is a more fair comparison for the sim2real gap in these experiments. Because our policies are trained with broad state coverage, they can recover from failures and retry until success. You can see this behavior in the first ~20 seconds of the full, uncut evaluation videos at the bottom of our website: weirdlabuw.github.io/omnireset/

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! weirdlabuw.github.io/omnireset/ (1/N 🧵)

what the ....

@patrickhyin @patrickhyin Amazing work! Why no "on-policy distillation" to RGB observation though ? Shouldn't it be even more robust ?

RL ftw 🚀

To whome it may concern:

@GuanyaShi Small or big. Dent is a dent if there are results and no one else has tried it.

I’m so tired of writing rebuttals to this kind of “lack of novelty” review: “This paper trivially combines A, B, and C, so the algorithmic novelty is limited.” Technically, most (if not all) robotics papers are convex combinations of existing ideas. I still deeply appreciate A+B+C papers—especially when they deliver: - New capabilities: the “trivial combination” unlocks behaviors we simply couldn’t achieve before - Sensible & organic design: A+B+C is clearly the right composition—not some arbitrary A′+B+C′ - Nontrivial interactions: careful analysis of the dynamics, coupling, or failure modes between A, B, C - Rehabilitating old ideas: A was dismissed for years, but paired with modern B/C, it suddenly works—and teaches us why - System-level & "interface" insight: the contribution is not any single piece, but how the pieces talk to each other - Scaling laws or regimes: identifying when/why A+B+C works (and when it doesn’t) - Engineering clarity: making something actually work robustly in the real world is not “trivial” - New problem formulations: sometimes the real novelty is in the reformulation—only under this view does A+B+C make sense. Maybe worth keeping these in mind when reviewing the next A+B+C paper : )

@DrJimFan @GuanyaShi It's probably only me...The moment someone mentions "Imminent AGI", "Pre-AGI" or even just "AGI", I lose to grasp the credibility of the entire statement. 🙈

@GuanyaShi I stopped caring about conference paper reviews a while ago ;). It’s meaningless in the imminent pre-AGI phase we are in.

Amazon makes a big move in the humanoid game. Amazon has acquired Fauna Robotics, a New York-based humanoid robot startup. The transaction closed last week. Fauna Robotics developed Sprout, a compact and approachable humanoid robot designed for safe, everyday interaction in shared human spaces such as homes, offices, and schools. Standing about 3.5 feet tall, Sprout can walk, grasp objects, interact with people, and even dance. The robot was launched in January this year as a humanoid platform for developers, priced at $50,000. Following the acquisition, Fauna’s roughly 50 employees will join Amazon. The company will continue deploying Sprout to outside researchers, and the startup will retain its name while operating as “Fauna, an Amazon company.” $AMZN

@rameshsrivats Sounds like you have become a PM

My workflow - Claude gives me the prompts. I paste it in Claude Code. Claude Code gives me outputs. I screenshot that and paste it in Claude. Claude gives next prompt. And so on. Basically Claude is the manager. Claude Code is the worker. And I'm the mailman.

@peteflorence This is amazing. Love the smooth motion!

The real story behind our GTC demo last week is that we only spent a few days prepping the demo, on a robot we had never touched before! We then ran it live for four days straight. More of the story here👇

Software horror: litellm PyPI supply chain attack. Simple `pip install litellm` was enough to exfiltrate SSH keys, AWS/GCP/Azure creds, Kubernetes configs, git credentials, env vars (all your API keys), shell history, crypto wallets, SSL private keys, CI/CD secrets, database passwords. LiteLLM itself has 97 million downloads per month which is already terrible, but much worse, the contagion spreads to any project that depends on litellm. For example, if you did `pip install dspy` (which depended on litellm>=1.64.0), you'd also be pwnd. Same for any other large project that depended on litellm. Afaict the poisoned version was up for only less than ~1 hour. The attack had a bug which led to its discovery - Callum McMahon was using an MCP plugin inside Cursor that pulled in litellm as a transitive dependency. When litellm 1.82.8 installed, their machine ran out of RAM and crashed. So if the attacker didn't vibe code this attack it could have been undetected for many days or weeks. Supply chain attacks like this are basically the scariest thing imaginable in modern software. Every time you install any depedency you could be pulling in a poisoned package anywhere deep inside its entire depedency tree. This is especially risky with large projects that might have lots and lots of dependencies. The credentials that do get stolen in each attack can then be used to take over more accounts and compromise more packages. Classical software engineering would have you believe that dependencies are good (we're building pyramids from bricks), but imo this has to be re-evaluated, and it's why I've been so growingly averse to them, preferring to use LLMs to "yoink" functionality when it's simple enough and possible.

Codex laughs at your petty guardrails

Computer animation in the 90s was no joke.

@lucasmaes_ Pretty cool! Should there always be a goal observation embedding available to do planning?

JEPA are finally easy to train end-to-end without any tricks! Excited to introduce LeWorldModel: a stable, end-to-end JEPA that learns world models directly from pixels, no heuristics. 15M params, 1 GPU, and full planning <1 second. 📑: le-wm.github.io

"Mid training" dataset

@chris_j_paxton JEPA models are a subset of dynamics models?

Cool work but I feel like such an old man when people call an action-conditioned world model a jepa. Or even a world model. Its a learned dynamics model, isn't it??

@ai_for_success Not clinging to strong opinions

What skill separates those who succeed in the AI era from those who don’t?

@rameshsrivats AI Pro Max will replace senior staff

If AI is going to replace all junior staff, then 10-15 years from now, where will senior staff come from? Tell, tell.

@Ronald_vanLoon @IlirAliu_ Which lab?

Stanford’s Tiny #Robot Could Be the Apple II Moment for #Robotics by @IlirAliu_ #Engineering #ArtificialIntelligence #Innovation #Technology