Being-H0.5: Scaling Human-Centric Robot Learning for Cross-Embodiment Generalization (Jan 2026)

Описание: Title: Being-H0.5: Scaling Human-Centric Robot Learning for Cross-Embodiment Generalization (Jan 2026)
Link: http://arxiv.org/abs/2601.12993v1
Date: January 2026

Summary:
This paper introduces Being-H0.5, a foundational Vision-Language-Action (VLA) model designed to solve the challenges of cross-embodiment generalization in robotics. The authors propose a human-centric learning paradigm supported by UniHand-2.0, the largest embodied pre-training dataset to date, featuring over 35,000 hours of data across 30 distinct robotic platforms. Key technical innovations include a Unified Action Space that maps diverse robot morphologies to a shared physical vocabulary, and a Mixture-of-Flow (MoF) architecture that separates shared motor primitives from specialized experts. To ensure robustness in real-world deployment, the system utilizes Manifold-Preserving Gating for sensory stability and Universal Async Chunking for handling heterogeneous latencies, achieving state-of-the-art performance on LIBERO and RoboCasa benchmarks.

Key Topics:
Vision-Language-Action (VLA) Models
Cross-Embodiment Generalization
Human-Centric Robot Learning
UniHand-2.0 Dataset
Unified Action Space
Mixture-of-Flow (MoF)
Manifold-Preserving Gating
Universal Async Chunking

Chapters:
00:00 - Introduction and Paper Summary
01:28 - Robotics' Tower of Babel Problem
03:00 - Human Motion as Lingua Franca
03:52 - UniHand 2.0 Dataset Scale
05:47 - Unified State Action Space
06:49 - Mixture of Flow Architecture
09:00 - Solving Real-World Deployment
10:59 - Benchmark Results and Analysis
12:27 - Achieving Zero-Shot Transfer
13:43 - Future of Universal Motor Intelligence

Stock video credits:
Pavel Danilyuk - https://www.pexels.com/@pavel-danilyuk
Pixabay - https://www.pexels.com/@pixabay
Google DeepMind - https://www.pexels.com/@googledeepmind
Oleg Gamulinskii - https://www.pexels.com/@oleg-gamulins...
Bedrijfsfilmspecialist.nl - https://www.pexels.com/@bedrijfsfilms...
Pressmaster - https://www.pexels.com/@pressmaster
SHVETS production - https://www.pexels.com/@shvets-produc...
Anete Lusina - https://www.pexels.com/@anete-lusina
Silviu Din - https://www.pexels.com/@silviu-din-16...
Kindel Media - https://www.pexels.com/@kindelmedia
RDNE Stock project - https://www.pexels.com/@rdne
Darli Donizete - https://www.pexels.com/@darlidonizete
Soumya - https://www.pexels.com/@soumya-1446957
Nino Souza - https://www.pexels.com/@ninosouza
Adis Resic - https://www.pexels.com/@adis-resic-29...
Pachon in Motion - https://www.pexels.com/@pachon-in-mot...
Colin Jones - https://www.pexels.com/@larchmedia
Dan Cristian Pădureț - https://www.pexels.com/@paduret
StefWithAnF - https://www.pexels.com/@stefwithanf-1...
Engin Akyurt - https://www.pexels.com/@enginakyurt
crazy motions - https://www.pexels.com/@crazy-motions...
tunnel motions - https://www.pexels.com/@tunnelmotions
Ron Lach - https://www.pexels.com/@ron-lach
Charlie Mounsey - https://www.pexels.com/@charlie-mouns...
Stas Knop - https://www.pexels.com/@stasknop
Tom Fisk - https://www.pexels.com/@tomfisk
Anthony 🙂 - https://www.pexels.com/@inspiredimages
Yaroslav Shuraev - https://www.pexels.com/@yaroslav-shuraev
fauxels - https://www.pexels.com/@fauxels
Trippy Lagoon - https://www.pexels.com/@trippy-lagoon...
cottonbro studio - https://www.pexels.com/@cottonbro