PR Newswire  | 

Bones Studio to Release BONES-SEED - the First Multimodal Motion Dataset Purpose-Built for Humanoid Robotics

PR Newswire

play Anhören
share Teilen
feedback Feedback
copy Kopieren
newsletter
font_big Schrift vergrößern
Nvidia Corp 182,26 $ Nvidia Corp Chart +0,17%
Zugehörige Wertpapiere:

Until now, researchers and startups building humanoid robots faced a critical challenge: no publicly available, large-scale, annotated motion dataset designed specifically for robotics. At GTC 2026, Bones Studio is closing that gap with NVIDIA technology.

SAN JOSE, Calif., March 16, 2026 /PRNewswire-PRWeb/ -- Bones Studio to Release BONES-SEED – the First Multimodal Motion Dataset Purpose-Built for Humanoid Robotics

"This means any team, anywhere in the world, can now start training humanoid robots on real, annotated human motion data - today."

Until now, researchers and startups building humanoid robots faced a critical challenge: no publicly available, large-scale, annotated motion dataset designed specifically for robotics. At GTC 2026, Bones Studio is closing that gap with NVIDIA technology. BONES-SEED makes 142,000 high-fidelity human motion animations - with rich multimodal annotations - openly availablefor the first time, levelling the playing field for the entire robotics community.

Knockout von Ing Markets Werbung

Passende Knock-Outs

Strategie Hebel
Steigender Kurs
Call
5
10
20
Fallender Kurs
Put
5
10
20
Den Basisprospekt sowie die Endgültigen Bedingungen finden Sie jeweils hier: DE000NB2RNY4 , DE000NB22Q57 , DE000NB5G6E3 , DE000NB26ZX0 , DE000NB4CW69 , DE000NB5QLV9 . Bitte informieren Sie sich vor Erwerb ausführlich über Funktionsweise und Risiken der Produkte. Bitte beachten Sie auch die weiteren Hinweise zu dieser Werbung.

BONES-SEED (Skeletal Everyday Embodiment Dataset) is built from the same data that powered breakthrough SONIC research - a whole-body control model for humanoid robots. The dataset spans locomotion and everyday activities to object interactions and complex whole-body behaviors, all curated specifically for humanoid robotics applications.

What sets BONES-SEED apart is not just the scale and quality of its 3D motion data - it's the depth of its annotations. Each motion comes with up to six natural language descriptions, temporal segmentation that breaks every sequence into meaningful phases with precise timestamps, and detailed skeletal metadata. This isn't raw motion capture - it's everything a language-to-action model needs.

Prepared natively in NVIDIA SOMA and Unitree G1 formats, BONES-SEED standardizes motion data exchange across research and industry - giving the robotics ecosystem a common foundation to build on.

This means any team, anywhere in the world, can now start training humanoid robots on real, annotated human motion data - today.

Learn more about Bones Studio's datasets and access BONES-SEED

With over 5 years of experience, Bones Studio builds enterprise-grade, multimodal datasets of human behavior and motion for AI and robotics. BONES-SEED represents a curated subset of Bones Studio's broader motion capture library, with expanded datasets available for commercial licensing.

Media Contact
Adrian Perdjon, Bones Studio, 48 602670961, adrian@bones.studio, bones.studio 

Cision View original content to download multimedia:https://www.prweb.com/releases/bones-studio-to-release-bones-seed---the-first-multimodal-motion-dataset-purpose-built-for-humanoid-robotics-302714306.html

SOURCE Bones Studio


Für dich aus unserer Redaktion zusammengestellt

Dein Kommentar zum Artikel im Forum

Jetzt anmelden und diskutieren Registrieren Login

Hinweis: ARIVA.DE veröffentlicht in dieser Rubrik Analysen, Kolumnen und Nachrichten aus verschiedenen Quellen. Die ARIVA.DE AG ist nicht verantwortlich für Inhalte, die erkennbar von Dritten in den „News“-Bereich dieser Webseite eingestellt worden sind, und macht sich diese nicht zu Eigen. Diese Inhalte sind insbesondere durch eine entsprechende „von“-Kennzeichnung unterhalb der Artikelüberschrift und/oder durch den Link „Um den vollständigen Artikel zu lesen, klicken Sie bitte hier.“ erkennbar; verantwortlich für diese Inhalte ist allein der genannte Dritte.


Weitere Artikel des Autors

Themen im Trend