samrat

georgia tech alum & stanford // dallas mavs fan // I sometimes tweet about rl and robots

[samrat @samratsahoo_]

Task and motion planning (TAMP) excels at multi-step, semantic, and geometric reasoning. VLAs excel at deformable object manipulation, which TAMP can't do. Can one VLA do both?

DATAFARM fine-tunes VLAs on TAMP demos aligned to the pre-training distribution.

(1/9)

[Bowen Li @Bw_Li1024]

Failures are inevitable for robots that are deployed in the *big* world. We introduce ReSYNC, a system that leverages large-scale RL sim2real to recover from local failures and concept discovery to explain/avoid new failures. Check it out here: https://t.co/rSK8p4WpXZ

[Satvik Sharma @satviks107]

Want to know how well agents can really control robots directly?

In RoboARCade, we systematically evaluate and diagnose frontier agents controlling robots directly using tens of thousands of dollars’ worth of agent trajectories. RoboARCade contains diverse, precise, and long-horizon tasks. Paper, code, assets and every trace are out today. https://t.co/WCCWpc4V4C

[Harsh Gupta @hgupt3]

Per-joint Gaussian noise is a terrible way to explore a 20+ DoF hand

We introduce EigenDEXplore, which explores along coordinated patterns from human hand motion, while keeping full control of every joint

Better learning, from tool use 🔨 to reorientation 🔄 to bimanual 👐

[Nishanth Kumar @nishanthkumar23]

Very cool new work; especially cool to see the resulting vla do better than TAMP and the original VLA! I think this overall direction of auto-generating training data for VLAs is very promising.

Also great to see TiPToP being useful for other research (this was our hope when we built it!) 😄

[Tom Silver @tomssilver]

@samratsahoo_ I also want to give a big shout-out to TiPToP (by @WillShenSaysHi @nishanthkumar23 @MIT_LISLab), the TAMP system we build on. B/c of TiPToP, @samratsahoo_ was able to set up and run TAMP on a new robot on his first day at Princeton. Inconceivable a few years ago!

[Yichao Liang @yichao_liang]

Agentic real-to-sim is getting increasingly good at rebuilding scenes in a physics simulator. But looking at a scene does not reveal how glue cures, how water heats, or how a fan's wind pushes things. And what if the simulator cannot model these mechanisms at all?

We present EMPIRIC, an agent that learns a residual world model: a physics engine extended with code for the missing mechanisms.

It solves all 25 runs across five simulated domains, where coding-agent baselines solve 14 to 16, and on a real robot it learns wind and domino masses from two gusts.

https://t.co/0QTGqZtMAW

[1/9]

[samrat @samratsahoo_]

I am grateful to my collaborators @YixuanHuang13 and @tomssilver for all the feedback and support on this project!

(9/9)

[samrat @samratsahoo_]

Task and motion planning (TAMP) excels at multi-step, semantic, and geometric reasoning. VLAs excel at deformable object manipulation, which TAMP can't do. Can one VLA do both?

DATAFARM fine-tunes VLAs on TAMP demos aligned to the pre-training distribution.

(1/9)

[samrat @samratsahoo_]

Can you tell DATAFARM from teleop? Take the quiz on our site!
🌐 Project website: https://t.co/V8JVMCvaik
📄 Paper: https://t.co/u65xoKDcdj
💻 Code: https://t.co/5dWRnWLJ7e

(8/9)

[Tom Silver @tomssilver]

I’ve seen robots do plenty of unexpected things. That usually means it’s time to fix a bug. This new project is one of only a few times in my career where I’ve seen robots do things that are unexpectedly clever—that none of us on the project anticipated.

A few examples...

[Anuva Banwasi @anuvabanwasi]

Robots need memory 🧠: remembering where your favorite cup is or which drawer is locked.

How can a robot remember across large spaces, interactions, & fine-grained detail?

Introducing MessyMem (accepted to #CoRL2026 🚀): Learning-from-Doing Memory for Mobile Manipulation. 🧵

[Dian Wang @Dian_Wang_]

Long horizon bimanual mobile manipulation requires reasoning in many coordinate frames: base, L/R hands, etc. In which frame would policy work best?
It really depends, so don’t pick. Mixture of Frames Policy: denoise in multiple frame in parallel.
🌐https://t.co/GnKSeNqzHA (1/9)

[Yixuan Huang @YixuanHuang13]

🤖 Curious about the connection between Task and Motion Planning (TAMP) and robot learning?

📄 Check out our new survey on Learning By and For Task and Motion Planning!

We organize the growing literature on TAMP + learning, introduce a taxonomy of TAMP learning, clarify the state of the art, and highlight promising directions for future research. 🚀

🔗 https://t.co/uqha9QP4gf

[Ria Doshi @riadoshi21]

🤔 Can we train one VLA policy to control multi-robot teams without any explicit communication?

✨ Introducing CHORUS: a single policy for decentralized, multi-embodiment collaboration

🧵⬇️