Robotics · 2026-10-09

UNITAS: A 3D-Native World Action Model for Embodied Manipulation

Ruixiang Wang, Yongyi Su, Wenlve Zhou, Bo Yue, Hengyan Liu, Dekun Lu, Yuxin Tian, Yihan Fang, Zerui Wu, Xing Hu, Jietao Chen, Yong Guo, Ziyan He, Junbin Yuan, Guiliang Liu, Xiaofen Xing, Kui Jia

arXiv:2610.12099 PDF

AI-assisted summary based on the paper. Consult the linked source for methods and limitations. Benchmark numbers appear after editor verification.

TL;DR

UNITAS is a 3D-native world action model for embodied manipulation that unifies observations, actions, and scene dynamics in a shared metric 3D frame within each interaction. It represents action flow and scene flow as displacement trajectories in 3D, grounded by world-aligned 3D positional embeddings (with or without depth input) and encoded using a physical-time trajectory tokenizer. The authors report leading action-conditioned scene prediction and strong manipulation success, and release code.

Why it matters

The authors argue that robotic interaction happens in metric 3D while video-based world action models represent evolution through view-dependent images/latents, making physical distances and contact relationships implicit. By aligning observations, action trajectories, and scene dynamics in a shared metric 3D frame, UNITAS aims to support task-conditioned behavior, consequence prediction, and executable control using a common interface across robot embodiments and human hands.

Method

3D-native interface: unify observations, action flow, and scene flow in one shared metric 3D frame; action and scene flows are displacement trajectories conditioned on instruction and current observations.
World-aligned 3D positional embeddings ground visual tokens using observed depth or a weighted combination along calibrated camera rays (depth may be absent at inference).
Physical-time trajectory tokenizer encodes an entire point trajectory as one token anchored at its current 3D position; two inference modes: policy execution (action chunk only) and simulator mode (action flow + scene prediction).

Abstract (from arXiv)

World action models (WAMs) aim to answer a coupled physical question: given a task instruction, what motion should the robot execute, and how will that motion change the surrounding world? Most existing WAMs build on pretrained video generators and represent world evolution through images or visual latents. Robotic interaction, however, takes place in metric three-dimensional space, while images are view-dependent projections whose pixel distances do not directly encode physical distances. We introduce UNITAS, to our knowledge the first 3D-native world action model that unifies observations, actions, and scene dynamics in a shared metric 3D frame within each interaction, using a common representation across robot embodiments and human hands. Action flow represents human hands and robot grippers as 3D point trajectories, while scene flow describes scene-point displacements conditioned on these trajectories. World-aligned 3D positional embeddings ground visual tokens with or without depth input, and a physical-time trajectory tokenizer encodes each point trajectory as one token anchored at its current 3D position. This interface supports both direct action execution and action-conditioned scene prediction. With 1.7B parameters, UNITAS achieves the best action-conditioned scene prediction on RoboTwin among the compared methods, with up to 49% lower displacement errors than PointWorld, and state-of-the-art manipulation success, including 99.8% on LIBERO and an average of 85% across real-world tasks. The code is available at https://github.com/DexForce/UNITAS.

Publication record

Bibliographic evidence describes where and when a record appeared. It does not validate scientific claims.

First public release: 2026-10-08 · Indexed by Pub-AI: 2026-10-09 UTC

preprint