Bones Studio (@TheBonesStudio) / X

Bones Studio

51 posts

Bones Studio

@TheBonesStudio

The Source of ground-truth multimodal human motion & behavior data for Physical AI. From world models to humanoid control | 5+ years, trusted by leading labs

Joined October 2025

Pinned
Bones Studio
@TheBonesStudio
Mar 16
Training humanoid robots? You need motion data. Real, high-fidelity, human motion data. And until now - there was no open dataset purpose-built for humanoid robotics. For 5 years, we've been building the largest enterprise-grade human motion and behavior datasets for embodied
00:00
49K
Bones Studio
@TheBonesStudio
May 23
Boxes: solved. Spheres: TBD. Deformables: don't ask. Episode 02 of Ground Truth is live!
00:00
1.5K
Bones Studio
@TheBonesStudio
May 21
Pick and Place is easy. Until it isn't... Episode 02 of Ground Truth drops soon. A lot of confidence about to be misplaced.
00:00
769
Bones Studio
@TheBonesStudio
May 9
Welcome to Ground Truth! A series where in a not-so-serious way we share our experience from building datasets for Physical AI. In this first episode, two robots learn that the real world is messy, unpredictable, and full of... stairs that are all slightly different :)
00:00
8.1K
Bones Studio
@TheBonesStudio
May 7
SONIC just crossed the embodiment barrier. @solo_tech ported NVIDIA's SONIC to the @AGIBOTofficial X2 Ultra – a completely different humanoid morphology - using ~2,550 clips from BONES-SEED. Same data. Different robot. It works. This is why we built BONES-SEED as a universal
Dhruv Diddi
@DhruvDiddi
May 7
🚀 @Solo__Tech breakthrough: NVIDIA Sonic goes beyond the G1. 🌏 We’ve achieved a global first: migrating NVIDIA Sonic to a completely different humanoid morphology, the AGIBOT X2 🥇 This is a massive leap for transferable humanoid intelligence. We are moving away from
00:00
1.5K
Bones Studio
@TheBonesStudio
Apr 29
SONIC. Kimodo. Now MotionBricks. Three @nvidia breakthroughs, all powered by our motion data. From research to real-time production at 15,000 FPS. From digital characters to Unitree G1 robots. This is what happens when great data meets great research. Congratulations
tingwu.wang
@TingwuWang
Apr 29
What is missing to bring real-time motion research into AAA games and real-world robotics? We present MotionBricks, a step toward bridging this gap with two key components: - a single generative latent motion backbone covering 350,000+ motion skills, running at 15,000 FPS with 2
00:00
1.3K
Bones Studio
@TheBonesStudio
Apr 16
Great piece by @oyhsu at @a16z. A sharp, coherent breakdown of why Physical AI is entering its own scaling regime. The new input isn't language. It's multimodal physical data. Nobody was collecting it - there was no reason to. Now there is. Synchronized. Automated. At
Oliver Hsu
@oyhsu
Apr 15
Article
Frontier Systems for the Physical World
Robot learning, autonomous science, and new interfaces as instances of an emerging paradigm for physical AI The dominant paradigm in AI today, insofar as it is used in production-ready settings, is...
579
Bones Studio
@TheBonesStudio
Apr 11
First the Kimodo model. Now the Kimodo Motion Generation Benchmark! @davrempe keeps building 🚀 Here's what's exciting: anyone training motion generation models can now benchmark them against BONES-SEED ground truth data. This is how open ecosystems are built – one building
Davis Rempe
@davrempe
Apr 10
Are you training motion gen models on @TheBonesStudio’s SEED dataset? Evaluate and compare your models on our new Kimodo Motion Generation Benchmark! It includes an extensive text-to-motion test suite and code for computing standardized metrics. github.com/nv-tlabs/kimod…
AI Datasets | Bones Studio
From bones.studio
728
Bones Studio
@TheBonesStudio
Apr 3
Humans are the best reference for humanoid robots. The challenge isn't the source - it's the capture. Teleoperation doesn't scale. Generative models do - but only if the real data they're built on is good enough. Captured. Labeled. Transferred.
00:00
66K
Bones Studio
@TheBonesStudio
Apr 2
Love this from @kotaueda999 – teaching a Unitree G1 to dance traditional Japanese Ohara-bushi. Amazing work! The dance motion came from video. But the physical stability? That came from a WBC model pre-trained on BONES-SEED dataset. Without it: 20 hours of training and the
ヒューマノイドロボットに踊りを教える方法 — Unitree G1で鹿児島のおはら節を踊らせるまでの全工程｜Kota Ueda(上田康太)
From note.com
2.4K
Bones Studio
@TheBonesStudio
Mar 28
“Affordances can’t be cleanly inferred from pixels alone. They belong at the object level.” This resonates. Synthetic and estimated data can get you far in pretraining, but at some point building neural objects with real affordances requires ground-truth capture of how humans
Chongjie(CJ) Ye
@ychngji6
Mar 27
Article
Neural Objects: The Missing Abstraction in World Models
There has been significant progress in teaching machines to model the world. From large language models operating over text, to emerging world models that predict pixels or states over time, we are...
848
Bones Studio
@TheBonesStudio
Mar 27
When someone explains what you do better than you can – you just repost. Huge thanks @lukas_m_ziegler! 🇵🇱
Lukas Ziegler
@lukas_m_ziegler
Mar 27
Perfectly synchronized multi-modal data with full 3D spatial context! 🎯 Sounds complicated, but let me explain this in easier terms. First of all, this technology comes from… Poland! 🇵🇱 I'm proud to see a Polish company powering @NVIDIARobotics most advanced robotics
00:00
594
Bones Studio
@TheBonesStudio
Mar 26
Fully agree — Kimodo is amazing 🔥 Huge kudos and congratulations to @davrempe and the whole @nvidia team! Proud that our data helped make this possible 👏
Min Choi
@minchoi
Mar 25
NVIDIA's Kimodo is wild. Text, keyframes, and 2D paths can now generate high-quality 3D human and robot motion at scale Demo in comments👇
Victor M
00:00
Victor M
534