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

Bones Studio will release BONES-SEED, a 142,000-animation multimodal motion dataset purpose-built for humanoid robotics, announced at GTC 2026 on March 16, 2026.

Rhea-AI Summary

Bones Studio will release BONES-SEED, a 142,000-animation multimodal motion dataset purpose-built for humanoid robotics, announced at GTC 2026 on March 16, 2026. BONES-SEED includes up to six natural-language descriptions per motion, temporal segmentation with precise timestamps, and skeletal metadata in NVIDIA SOMA and Unitree G1 formats.

The dataset is a curated subset of Bones Studio's larger library and is openly available, with expanded commercial licensing offered.

Argus Mar 17 session

-0.70% close to close 0.9x rel. volume Open Argus

Details

News Market Reaction – NVDA

On Mar 17, the first trading day after this news, NVDA closed 0.70% below the previous close.

Data tracked by StockTitan Argus for the Mar 17 session.

Key Figures

Motion animations: 142,000 animations Natural language descriptions: Up to 6 descriptions Company experience: Over 5 years

Motion animations

142,000 animations

High-fidelity human motion animations in BONES-SEED dataset

Natural language descriptions

Up to 6 descriptions

Per motion sequence in BONES-SEED

Company experience

Over 5 years

Bones Studio experience building multimodal behavior datasets

Historical Context

5 past events · Latest: Mar 16

5 events

1. Mar 16

   AI CPU launch

   24h Move

   +2.2%

   Launch of Vera CPU for agentic AI with stated speed and efficiency gains.

2. Mar 16

   Storage platform launch

   24h Move

   +2.2%

   Unveiling of BlueField-4 STX accelerated storage architecture for agentic AI workloads.

3. Mar 16

   AI platform rollout

   24h Move

   +2.2%

   Introduction of Vera Rubin platform with new chips and rack designs for AI factories.

4. Mar 16

   Gaming AI feature

   24h Move

   +2.2%

   Announcement of DLSS 5 neural rendering for real-time, high-fidelity gaming visuals.

5. Mar 11

   AI cloud partnership

   24h Move

   +0.7%

   Strategic Nebius partnership with a $2B NVIDIA investment to scale AI cloud capacity.

24h Move is the share-price change in the day after each event; other market factors may also have contributed.

Key Terms

multimodal, humanoid robots, motion capture, language-to-action model, +1 more

5 terms

multimodal technical

"enterprise-grade, multimodal datasets of human behavior and motion for AI and robotics"

Multimodal describes an approach, product, or system that uses two or more different types of inputs, methods, or channels — for example combining text, images and audio in a technology product, or blending drugs, devices and therapy in medical care. For investors, multimodal solutions can broaden market reach and competitive differentiation but also add development cost, operational complexity and regulatory hurdles; think of it like a hybrid car that offers more capabilities but requires more parts and oversight.

humanoid robots technical

"researchers and startups building humanoid robots faced a critical challenge"

Machines built with human-like bodies or features—such as a head, arms, and legs—designed to perform tasks, interact with people, or navigate environments similarly to a person. Investors care because these robots can change labor needs, create new markets for products and services, and affect costs and productivity in industries like manufacturing, logistics, healthcare and retail; think of them as programmable workers that can reshape how businesses operate and compete.

motion capture technical

"a curated subset of Bones Studio's broader motion capture library"

Motion capture is a technology that records the movement of people or objects and turns those movements into digital data used to animate characters, analyze performance, or control machines. Think of it as putting sensors on a dancer so their steps become a blueprint for a virtual actor or a rehabilitation program. Investors care because motion capture can drive revenue and efficiency in movies, games, sports science, healthcare and robotics, creating scalable products, licensing opportunities and competitive advantage.

language-to-action model technical

"This isn't raw motion capture - it's everything a language-to-action model needs"

A language-to-action model is a type of artificial intelligence that takes written or spoken instructions and turns them into real-world tasks or decisions, such as sending an email, updating a database, or executing a trading rule. For investors, it matters because these systems can speed operations, cut labor costs and enable new products, but they also introduce execution risk, oversight needs and potential regulatory scrutiny—think of a smart assistant that not only understands you but also presses the buttons.

unitree g1 technical

"Prepared natively in NVIDIA SOMA and Unitree G1 formats"

A Unitree G1 is a four‑legged mobile robot model — a compact, sensor‑equipped machine designed to walk, carry small payloads and run programmed tasks much like a remote‑controlled dog. For investors, such a product signals a company’s move into robotics hardware and software with potential recurring revenue from sales, services and upgrades; its success can indicate market demand, manufacturing scale and competitive position much like a new consumer gadget line reveals a tech firm’s growth prospects.

AI-generated analysis. How Rhea-AI works. Not financial advice.

See more from StockTitan in Google Search and AI answers. Adds StockTitan as a preferred source · opens Google

Add on Google Not now

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.

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 

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

FAQ

AI-generated questions and answers. How Rhea-AI works. Not financial advice.

What is BONES-SEED and when was it announced for NVDA attendees?

BONES-SEED is a 142,000-animation multimodal motion dataset announced at GTC 2026 on March 16, 2026. According to Bones Studio, it provides richly annotated 3D human motion data formatted for NVIDIA SOMA and Unitree G1 to accelerate humanoid robotics research.

How much motion data does BONES-SEED contain and what annotations are included?

BONES-SEED contains 142,000 high-fidelity human motion animations with up to six natural-language descriptions per motion. According to Bones Studio, each sequence has temporal segmentation, precise timestamps, and detailed skeletal metadata for language-to-action models.

Will BONES-SEED be compatible with NVIDIA tools and NVDA platforms?

Yes — BONES-SEED is prepared natively in NVIDIA SOMA formats for NVDA-compatible workflows. According to Bones Studio, the dataset standardizes motion exchange and lets teams train humanoid control models on annotated real-world motion data immediately.

Can companies license expanded BONES-SEED datasets beyond the open release?

Yes — Bones Studio offers expanded commercial licensing beyond the openly available curated subset. According to Bones Studio, organizations can access larger portions of the motion capture library under commercial terms for enterprise or research use.

How does BONES-SEED support language-to-action models for humanoid robots?

BONES-SEED supplies multimodal annotations—natural language descriptions, temporal phases, and skeletal metadata—designed for language-to-action training. According to Bones Studio, this structured labeling gives models the sequence, semantics, and timing needed for whole-body robot control.