Skip to content
RobotWorld
← Back to Videos

World2HOSI Generates Long-Horizon Human-Object-Scene Interaction

Loading video
Loading video

World2HOSI combines text prompts and 3D Gaussian scenes so existing motion experts can collaborate in a shared, evolving world without training a new unified motion generator.

World Models3D Gaussian SplattingHuman-Object InteractionMotion GenerationScene InteractionHuman Animation
Category: research
Author: @_cheng_lin
Date: 2026-10-21T00:00:00
Duration: 68.066s
Source: https://x.com/_cheng_lin/status/2112934433053687997
Reference: https://nintreesssss.github.io/World2HOSI/

Related Videos

COSMI: Compositional Human-Object Interaction Dataset and Diffusion Model
0:37

COSMI: Compositional Human-Object Interaction Dataset and Diffusion Model

@GerardPonsMoll1 · 329 views · 2026-10-07
Human-Object InteractionDiffusion Transformer
Sam Altman on AI Scale, Utilities, and One-Person Labs
26:31

Sam Altman on AI Scale, Utilities, and One-Person Labs

@maarcoofdezz · 714 views · 2026-09-24
OpenAISam Altman
Muse Realtime Avatar: Real-Time Embodiment for Expressive AI Characters
2:02

Muse Realtime Avatar: Real-Time Embodiment for Expressive AI Characters

@AIatMeta · 695 views · 2026-09-24
Interactive AvatarDigital Human
PKU Motion-Omni End-to-End Dialogue System
0:08

PKU Motion-Omni End-to-End Dialogue System

@aigclink · 651 views · 2026-09-07
Embodied AIDigital Human
ActionPlan: Future-Aware Streaming Motion Synthesis via Frame-Level Action Planning
0:44

ActionPlan: Future-Aware Streaming Motion Synthesis via Frame-Level Action Planning

@XianghuiXie · 539 views · 2026-03-21
Motion GenerationMotion Diffusion
NVIDIA MotionBricks: real-time generation of 350,000 motion skills
0:37

NVIDIA MotionBricks: real-time generation of 350,000 motion skills

@HowToPrompt__ · 552 views · 2026-08-08
NVIDIAMotionBricks

As an Amazon Associate, we earn from qualifying purchases.