Rishabh Madan

PhD student @Cornell | Building caregiving robots @EmPRISELab | Prev: @ToyotaResearch @uwcse @IITKgp

[Rishabh Madan @madrish02]

Absolutely! Contact prediction + latent magic can unlock contact-rich whole-arm manipulation. More to come out soon 👀

[C. Zhang @ChongZzZhang]

okay, actually yes

[Rishabh Madan @madrish02]

@KanuGulati @chris_j_paxton @gs_ai_ Truly full-stack!

[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)

[Rishabh Madan @madrish02]

Humans rely on contact and force reasoning as a core part of how we perceive and act in the physical world. This reasoning is hierarchical, with each layer abstracting the one below, and spans scales from fingertip to whole body. At times, we subconsciously infer approximate object properties and maintain spatial awareness through touch alone. This is what makes fast, safe manipulation of fragile objects possible. Slip correction and grip adjustments happen in tens of milliseconds, much faster than vision can react, which is why we can crack eggs and effortlessly hold raspberries without crushing them.

Kudos to the @EkaRobotics team for showing these fast demos! Excited to see how they scale this up.

[Pulkit Agrawal @pulkitology]

Eka means unity -- “one,” in Sanskrit and “first” in Finnish.

We’re building intelligence for the physical world in its native language: forces.

Until now, robotics faced a tradeoff — generality or speed. The real world requires both. Robotics also faced a data problem.

Our Vision–Force–Action (VFA) model — the first of its kind — breaks the generality-speed tradeoff and the data barrier.

It's a new foundation uniting performance, generality, and safety for putting capable robots in everyone's hands.

Today, I am excited to share our journey of pushing robots beyond human limits.

Today, dexterity becomes scalable.

Today, I welcome you to the Era of Eka.

Co-founded with @haarnoja, and so thrilled and grateful to be working with a dream team at @EkaRobotics.

Learn more: https://t.co/QYQ6x2Etyi

[Rishabh Madan @madrish02]

The cross-embodiment transfer and language following demos are very impressive!

[Lucy Shi @lucy_x_shi]

1/ We just released π0.7 — a steerable generalist robot model with emergent capabilities.
I want to share a bit of the backstory, because π0.7 taught me something surprising about where robot learning is heading. A thread on bittersweet lessons 🧵

[Generalist @GeneralistAI]

Introducing GEN-1.

Our latest milestone in scaling robot learning.

We believe it to be the first general-purpose AI model to master simple physical tasks.

99% success rates, 3x faster speeds, adapts in real time to unexpected scenarios, w/ only 1 hour of robot data.

More🧵👇

[Rishabh Madan @madrish02]

Hey! Late post but I’m attending HRI’26 in Edinburgh 🏴󠁧󠁢󠁳󠁣󠁴󠁿 and will be in London next week. Interested in caregiving robots, contact-rich manipulation, or getting JEPA-style models to work for real-world pHRI? Let’s grab coffee ☕️

[Rohan Banerjee @rohanbbanerjee]

Hi all! I'm at #HRI2026 in Edinburgh this week, presenting our work: A Human-in-the-Loop Confidence-Aware Failure Recovery Framework for Modular Robot Policies.

Check out our talk on 3/18 at the Trust and Safety 2 session (Session 6A) @ 11:40am! More in the thread below🧵

[Demis Hassabis @demishassabis]

Cool use case of AlphaFold, this is just the beginning of digital biology!

[Rishabh Madan @madrish02]

Proud of @jmoonio and how far he has come in his startup journey. The path is onwards and upwards from here!

[Jonathan Moon @jmoonio]

For the past 12 months, I’ve been heads down building a robot.

Introducing Emma – an autonomous robot that scans farms, detects diseases, and measures yield.

Currently deployed in 14 vineyards and orchards in CA and NY.

[Abrar Anwar @_abraranwar]

@rishabhmadan96 tested it on videos of whole arm manipulation for body repositioning. This was super surprising!
https://t.co/fKcuE1gxqf

[Rishabh Madan @madrish02]

@VilleKuosmanen @TapoBhat The grey one's the robot mounted with a distributed tactile sensor

[Rishabh Madan @madrish02]

I tried Robometer on some of the most OOD robotics experiments (from my upcoming paper w/ @TapoBhat, coming soon 🤓). Pretty wild to see its progress monitor actually perform reasonably well on the side rollover task!

[Jesse Zhang @ ICRA 2026 ✈️ @Jesse_Y_Zhang]

A reward model that works, zero-shot, across robots, tasks, and scenes?

Introducing Robometer: Scaling general-purpose robotic reward models with 1M+ trajectories.

Enables zero-shot: online/offline/model-based RL, data retrieval + IL, automatic failure detection, and more!

🧵 (1/12)

[Rishabh Madan @madrish02]

@Jesse_Y_Zhang Very cool! Curious what the biggest contributor is to the flatline behavior here?

[Rishabh Madan @madrish02]

@ebiyik_ Haha me too! I wonder what kind of data the pretrained VLM is trained on? And if that is doing the heavy lifting here?

[Rishabh Madan @madrish02]

Bonus: I fed it the wrong task prompt to see if it would get tricked. Mostly flatline 🤩

[Rishabh Madan @madrish02]

Also tried limb repositioning and was pleasantly surprised again!

[Jesse Zhang @ ICRA 2026 ✈️ @Jesse_Y_Zhang]

A reward model that works, zero-shot, across robots, tasks, and scenes?

Introducing Robometer: Scaling general-purpose robotic reward models with 1M+ trajectories.

Enables zero-shot: online/offline/model-based RL, data retrieval + IL, automatic failure detection, and more!

🧵 (1/12)

[Rishabh Madan @madrish02]

@yuvaltassa Is the hypothesis that compliance provides enough margin for bridging the sim-to-real gap? Also, in your opinion, would it then be necessary to simulate such compliance (assuming it could be active and/or passive)?

[Kushal @kushalk_]

🤖 Can a single robot policy manipulate diverse tools without ever seeing them before?

Introducing SimToolReal 🔨 : a generalist dexterous manipulation policy that transfers zero-shot sim→real to unseen tools + unseen tasks

All videos are 1x speed (60 Hz control) 🧵👇

[Siddhant Haldar @haldar_siddhant]

Robot foundation models are limited by costly real data, while simulation data is plentiful but visually mismatched to reality. We present Point Bridge, a method that enables zero-shot sim-to-real transfer for robot learning with minimal visual alignment.

https://t.co/0Zi2PUPbE8