Kallol Saha

MSR Student @ CMU RI. I work on hybrid learning-and-planning methods for long-horizon tasks in robotics and beyond.
Previously RA @ RRC, IIITH

[Kallol Saha @_ksaha]

🚨Introducing SPOT: Search over Point Cloud Object Transformations. SPOT is a combined learning-and-planning approach that searches in the space of object transformations.

Website: https://t.co/VwiyDI15FC
Paper: https://t.co/GmdF7hN0FG
Code: https://t.co/YC5sO9OZ0K

[Sriram Krishna ✈️ RSS 2026 @sriramsk99]

Boo! 👻 GHOST has been accepted to #RSS2026!

Can we learn manipulation skills from human video (without action retargeting)?

Yes! GHOST learns generalizable manipulation skills by training a hierarchical policy with an embodiment-agnostic sub-goal predictor + embodiment-specific controller.

🧵(1/7)

[Aditya Agarwal @skymanaditya1]

Excited to be at #ICRA2026 in Vienna this week 🇦🇹 -- I'll be presenting SCOUT as a spotlight talk at the Rigorous Robot Perception workshop on June 01 at VIP Lounge C, and SceneComplete as an Oral talk today (June 02 at TuAT3.5).

[Tom Silver @tomssilver]

As a planning+learning researcher, I’m really excited about KinDER. It clarifies planning (especially TAMP) for outsiders, defines key open challenges for the field, and creates a common ground to compare & combined planning+learning approaches. (1/n)

[Tal Daniel @TalDaniel8]

🚀 #ICLR2026 Oral 💥
How can we design world models that capture object interactions directly from pixels?
Introducing Latent Particle World Models-the first end-to-end self-supervised, object-centric world model, trained from videos, supporting action/img/lang conditioning.
1/n

[Tangible @tangiblerobots]

Almost a year now since we started. Proud of what we've achieved so far and long way to go! 🥳

Merry Christmas from the Tangible team! 🎅

[Kallol Saha @_ksaha]

Manipulating multiple deformable objects like blankets and pillows is a really challenging long-horizon task - pretty hard to do with today's robotic grippers. Eggie from @tangiblerobots can do it with ease - I think this dexterous robot design has unlimited potential!

[Bipasha Sen @bipashasen31]

Long full shot. 2x, teleoperated. en-route to autonomy,

It was fun putting Eggie in a messy, cluttered scene. Clean rooms be boring.

[Kallol Saha @_ksaha]

Eggie is the most homely robot ever!! Can't wait to have this in my home

[Tangible @tangiblerobots]

Thought I heard you say I am cute?

[Kallol Saha @_ksaha]

The fine-grained dexterity here is very impressive!

[Tangible @tangiblerobots]

close up.

[Tangible @tangiblerobots]

Hello, Eggie.

The world was built around humans. Eggie doesn't just look human, Eggie interacts like us.

Dexterous. Mobile. Compliant.

We’re building Eggie to be the smartest robot to ever walk on Earth. Join us.

Built from scratch and with love in California. 🫶

[Kallol Saha @_ksaha]

Check out this recent article from @SCSatCMU covering our work on combining learning and planning for robotic manipulation:

[CMU School of Computer Science @SCSatCMU]

Researchers at CMU’s Robotics Institute have developed a new system that helps robots operate effectively in cluttered, unpredictable environments like kitchens, classrooms and offices — a huge step toward making robots more capable in everyday settings.
https://t.co/EuobBL3qBa

[Jiahui(Jim) Yang @Jiahui_Yang6709]

Super excited to share that Neural MP just won Best Student Paper Award at IROS 2025! Huge congrats to the entire team — it’s been an incredible journey working with all of you!

Neural MP, together with our follow-up work DRP, shows a general recipe for scaling sim data and training reactive, generalizable neural motion planners. We’re genuinely amazed by the policy’s zero-shot capability in arbitrary, unstructured, and dynamic environments.

More details and demos available at:
🔗 Neural MP https://t.co/hwjB3mg9Ud
🔗 DRP https://t.co/FwV3LomfK9

[Kallol Saha @_ksaha]

@ambermli will be presenting our paper SPOT at the Oral Session (2-3PM) at #CoRL2025 today! Come by our poster (A62) from 4:30–6 PM to ask questions or to chat with us!

[Kallol Saha @_ksaha]

🚨Introducing SPOT: Search over Point Cloud Object Transformations. SPOT is a combined learning-and-planning approach that searches in the space of object transformations.

Website: https://t.co/VwiyDI15FC
Paper: https://t.co/GmdF7hN0FG
Code: https://t.co/YC5sO9OZ0K

[Conference on Robot Learning @corl_conf]

Here we go! #CoRL2025

[Chris Paxton @chris_j_paxton]

Biggest CoRL ever

[CMU Robotics Institute @CMU_Robotics]

Robots in the home may one day help with tasks like sorting our dishes & organizing our shelves 👀🤖

But to do so, they need to plan over long sequences– figuring out how to move objects and where to put them.

Check out SPOT: a system that can rearrange objects into any layout!

[Kallol Saha @_ksaha]

Check out @hao_alexis's work on force modulated policies for assistive dressing!

[Alexis Hao @hao_alexis]

Introducing FMVP: a method that adapts to natural arm motions during robot-assisted dressing.

Pre-trained on vision in sim, fine-tuned with limited real-world vision+force data, and tested in a 12-user, 264-trial study, FMVP is robust across garments and motions. #CoRL2025

[Chris Paxton @chris_j_paxton]

I need to do a deep dive on video learning because I'm still honestly not convinced it solves the harsh problems in robotics, which are basically all about physically interacting with the world using perception

[Nikhil Keetha @Nik__V__]

Meet MapAnything – a transformer that directly regresses factored metric 3D scene geometry (from images, calibration, poses, or depth) in an end-to-end way. No pipelines, no extra stages. Just 3D geometry & cameras, straight from any type of input, delivering new state-of-the-art results 🚀

One universal model enables SoTA for:
🔥 Mono Depth Estimation
🔥 Multi-View SfM
🔥 Multi-View Stereo
🔥 Depth Completion
🔥 Registration

… and many more possibilities! – plus everything is metric 🎯

We release code for data processing, training, benchmarking & ablations – everything Apache 2.0!

Details & Links 👇

[Yuke Zhu @yukez]

People who are really serious about robot learning should make their own robot hardware.

[Kallol Saha @_ksaha]

Cool work from @divyamgo10 on geometric red teaming for manipulation!

[Divyam Goel @divyamgo10]

How do we discover a robot's failure modes before deploying it in the real world? Standard benchmarks often don't capture the full picture, leaving policies vulnerable to plausible variations in object shape.

Thrilled that our work, "Geometric Red-Teaming for Robotic Manipulation," has been accepted as an oral presentation at #CoRL2025! We introduce a framework to automatically find these geometric blindspots.
https://t.co/mh84bi7Wbp
🧵