Robotics startup XDOF raises $70M and releases ABC-130K, the largest open-source bimanual teleoperation dataset

VIEWS3.7KBOOKMARKS3LIKES27

This team uniquely combines experience and passion across robotics models, hardware and the grit for data gathering. They have stood up partnerships with all of the major players already! Congrats @philippswu

Philipp Wu@philippswu

Announcing @xdofai: We’ve raised $70 million to build the core robotic infrastructure ecosystem for robot foundation models.

My cofounders Fred (@YideShentu), Nemo (@itsnemojin) and I have been pursuing the dream of general purpose robots for our entire lives. After work at Covariant, Meta and Tesla, it became clear to us that general purpose robots are coming, and we are building XDOF to help make them a reality.

For the last two years, we’ve been working behind the scenes to support major labs and companies deploying robots. In us, they have a partner with full-stack expertise, from hardware to operations to policy training.

As our first public contribution to the space, we are open-sourcing ABC-130K, the largest open source teleoperation dataset, in collaboration with our partners from UC Berkeley, Carnegie Mellon, MIT and Amazon FAR.

Thank you to our customers, partners, collaborators and investors for your trust and conviction in us. Together, we can accelerate the future of robotics!

1h3.7K273

RETWEETS7

Ritvik Singh@ritvik_singh9

Introducing ABC: open data, training, and infrastructure for robotics.

We release the largest teleop dataset to date, and extensively investigate design decisions, pretraining, and post-training techniques.

@arthurallshire @Cinnabar233 @adamrasb @redstone_hong @davidrmcall

1h5.4K6212

REPLIES2

Philipp Wu@philippswu

ABC-130K is the largest open bimanual manipulation dataset. It combines dataset scale (4x larger than the next YAM dataset) and quality on a low-cost embodiment (YAM from I2RT).

This size and depth allows researchers to conduct a wider range of experiments and achieve better policy results.

The dataset is released under the highly permissive Apache 2.0 license and can be accessed below. Let us know what you think!

http://xdof.ai/blog/abc-130k

1h429101

Gokul Swamy@g_k_swamy

Even back in undergrad, Philipp always seemed too cracked to be real. I am very, very excited to see all the good that comes out of the dataset and open-source infra Philipp & co. have built!

Philipp Wu@philippswu

Announcing @xdofai: We’ve raised $70 million to build the core robotic infrastructure ecosystem for robot foundation models.

My cofounders Fred (@YideShentu), Nemo (@itsnemojin) and I have been pursuing the dream of general purpose robots for our entire lives. After work at Covariant, Meta and Tesla, it became clear to us that general purpose robots are coming, and we are building XDOF to help make them a reality.

For the last two years, we’ve been working behind the scenes to support major labs and companies deploying robots. In us, they have a partner with full-stack expertise, from hardware to operations to policy training.

As our first public contribution to the space, we are open-sourcing ABC-130K, the largest open source teleoperation dataset, in collaboration with our partners from UC Berkeley, Carnegie Mellon, MIT and Amazon FAR.

Thank you to our customers, partners, collaborators and investors for your trust and conviction in us. Together, we can accelerate the future of robotics!

51m970102

XDOF@xdofai

We are proud to release ABC-130K, the largest open bimanual manipulation dataset.

ABC-130K includes 130K+ trajectories covering 200 complex manipulation tasks. All are performed on a low-cost bimanual rig (YAM by I2RT).

The data is available under the highly permissive Apache 2.0 license.

1h966174

Philipp Wu@philippswu

We are hiring across all types of engineering, operations and recruiting. Come join us!

https://www.xdof.ai/careers

1h3055

Philipp Wu@philippswu

Thank you to @TimFernholz at @TechCrunch for covering our launch!

https://techcrunch.com/2026/06/17/collecting-robot-training-data-is-dirty-unglamorous-work-some-ai-labs-are-already-paying-xdof-to-do-it/

1h2875

Ritvik Singh@ritvik_singh9

Check out our paper, more videos, and our more than 3.5K hour dataset here https://abc.bot

1h1165

Ritvik Singh@ritvik_singh9

In addition to real data, we have simulation data. Using this data, we show sim/real correlation.

1h985

Ritvik Singh@ritvik_singh9

One interesting observation we find is across different model classes, training flow loss and validation action error correlate with real world success, while we did not observe any statistically significant correlation between validation flow loss and real world success.

1h735

Ritvik Singh@ritvik_singh9

We develop lightweight robot infra, enabling things such as DAgger and intervention data to augment our large scale dataset. This allows us to experiment with post-training our checkpoints to enable dexterous tasks such as airpod insertion

1h545

Ritvik Singh@ritvik_singh9

This would not have been possible without the support from Amazon FAR or our amazing mentors @akanazawa, @GuanyaShi, @pabbeel, @peterxichen, @rocky_duan, @JitendraMalikCV, @phillip_isola

1h525

Philipp Wu@philippswu

@TimFernholz @TechCrunch Check out ABC: Scalable Behavior Cloning with Open Data, Training, and Evaluation below:

1h1785

Ritvik Singh@ritvik_singh9

Shout out to our collaborators @xdofai -- see their tweets here and also congrats on their company announcement:

1h665

Kevin Zakka@kevin_zakka

So proud of Phil, Fred and the xdof team! It’s been inspiring to watch this take shape from the early days in the lab at Berkeley. I'm super excited for what they’re building and for the future of robotics.

Philipp Wu@philippswu

Announcing @xdofai: We’ve raised $70 million to build the core robotic infrastructure ecosystem for robot foundation models.

My cofounders Fred (@YideShentu), Nemo (@itsnemojin) and I have been pursuing the dream of general purpose robots for our entire lives. After work at Covariant, Meta and Tesla, it became clear to us that general purpose robots are coming, and we are building XDOF to help make them a reality.

For the last two years, we’ve been working behind the scenes to support major labs and companies deploying robots. In us, they have a partner with full-stack expertise, from hardware to operations to policy training.

As our first public contribution to the space, we are open-sourcing ABC-130K, the largest open source teleoperation dataset, in collaboration with our partners from UC Berkeley, Carnegie Mellon, MIT and Amazon FAR.

Thank you to our customers, partners, collaborators and investors for your trust and conviction in us. Together, we can accelerate the future of robotics!

23m21040

XDOF@xdofai

@ritvik_singh9 @arthurallshire @Cinnabar233 @adamrasb @redstone_hong @davidrmcall So great to collaborate with you all on this work!

1h424

Vai Viswanathan@vai_viswanathan

@xdofai amazing! does this also have success / failure runs from your evals?

29m13

Philipp Wu@philippswu

Our collaborators published ABC: Scalable Behavior Cloning with Open Data, Training, and Evaluation using the ABC-130K dataset and XDOF services.

https://abc.bot/

Thank you to UC Berkeley lead authors on this work: Arthur Allshire @arthurallshire Himanshu Gaurav Singh @Cinnabar233 Ritvik Singh @ritvik_singh9 Adam Rashid @adamrasb Hongsuk Choi @redstone_hong David McAllister @davidrmcall

And collaborators: Pieter Abbeel @pabbeel Peter Chen @peterxichen Rocky Duan @rocky_duan Phillip Isola @phillip_isola Jitendra Malik @JitendraMalikCV Guanya Shi @GuanyaShi Angjoo Kanazawa @akanazawa

1h1.5K191

XDOF@xdofai

Robotics startup XDOF raises $70M and releases ABC-130K, the largest open-source bimanual teleoperation dataset

Story Overview

Open weights and data lower the barrier for new robot policies

Funding targets the unglamorous data-collection bottleneck