# UniDexTok: A Unified Dexterous Hand Tokenizer from Real Data

arXiv: https://arxiv.org/abs/2606.10683 · HTML: https://mlsanity.com/paper/2606.10683/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 26 of its 38 references that are in the map.
- First posted on arXiv 9 Jun 2026.

## Builds on (most cited first, 26 of 26)

- [Scalable Diffusion Models with Transformers](https://mlsanity.com/paper/2212.09748/) (2022)
- [Neural Discrete Representation Learning](https://mlsanity.com/paper/1711.00937/) (2017)
- [RT-2: Vision-Language-Action Models Transfer Web Knowledge to Robotic Control](https://mlsanity.com/paper/2307.15818/) (2023)
- [OpenVLA: An Open-Source Vision-Language-Action Model](https://mlsanity.com/paper/2406.09246/) (2024)
- [FAST: Efficient Action Tokenization for Vision-Language-Action Models](https://mlsanity.com/paper/2501.09747/) (2025)
- [DexYCB: A Benchmark for Capturing Hand Grasping of Objects](https://mlsanity.com/paper/2104.04631/) (2021)
- [Cosmos Policy: Fine-Tuning Video Models for Visuomotor Control and Planning](https://mlsanity.com/paper/2601.16163/) (2026)
- [EgoDex: Learning Dexterous Manipulation from Large-Scale Egocentric Video](https://mlsanity.com/paper/2505.11709/) (2025)
- [Embodied Intelligence via Learning and Evolution](https://mlsanity.com/paper/2102.02202/) (2021)
- [DexGraspNet: A Large-Scale Robotic Dexterous Grasp Dataset for General Objects Based on Simulation](https://mlsanity.com/paper/2210.02697/) (2022)
- [OakInk: A Large-scale Knowledge Repository for Understanding Hand-Object Interaction](https://mlsanity.com/paper/2203.15709/) (2022)
- [LEAP Hand: Low-Cost, Efficient, and Anthropomorphic Hand for Robot Learning](https://mlsanity.com/paper/2309.06440/) (2023)
- [ManipTrans: Efficient Dexterous Bimanual Manipulation Transfer via Residual Learning](https://mlsanity.com/paper/2503.21860/) (2025)
- [DexGraspNet 2.0: Learning Generative Dexterous Grasping in Large-scale Synthetic Cluttered Scenes](https://mlsanity.com/paper/2410.23004/) (2024)
- [ControlVLA: Few-shot Object-centric Adaptation for Pre-trained Vision-Language-Action Models](https://mlsanity.com/paper/2506.16211/) (2025)
- [Scaling Diffusion Policy in Transformer to 1 Billion Parameters for Robotic Manipulation](https://mlsanity.com/paper/2409.14411/) (2024)
- [Cross-Embodiment Dexterous Grasping with Reinforcement Learning](https://mlsanity.com/paper/2410.02479/) (2024)
- [Enhancing Generalization in Vision-Language-Action Models by Preserving Pretrained Representations](https://mlsanity.com/paper/2509.11417/) (2025)
- [Latent Action Diffusion for Cross-Embodiment Manipulation](https://mlsanity.com/paper/2506.14608/) (2025)
- [LAP: Language-Action Pre-Training Enables Zero-shot Cross-Embodiment Transfer](https://mlsanity.com/paper/2602.10556/) (2026)
- [UniDex: A Robot Foundation Suite for Universal Dexterous Hand Control from Egocentric Human Videos](https://mlsanity.com/paper/2603.22264/) (2026)
- [Learning Dexterous Manipulation Skills from Imperfect Simulations](https://mlsanity.com/paper/2512.02011/) (2025)
- [Dexora: Open-source VLA for High-DoF Bimanual Dexterity](https://mlsanity.com/paper/2605.18722/) (2026)
- [Cross-Hand Latent Representation for Vision-Language-Action Models](https://mlsanity.com/paper/2603.10158/) (2026)
- [Analyzing Key Objectives in Human-to-Robot Retargeting for Dexterous Manipulation](https://mlsanity.com/paper/2506.09384/) (2025)
- [UniHM: Unified Dexterous Hand Manipulation with Vision Language Model](https://mlsanity.com/paper/2603.00732/) (2026)

## Abstract (arXiv)

Dexterous hands are essential for fine-grained manipulation, but their hardware designs vary substantially across embodiments. Differences in kinematics, joint definitions, and degrees of freedom make it difficult to define a shared state representation compared with parallel grippers. As a result, dexterous-hand data remains fragmented and difficult to use for joint training. In this work, we propose the Unified Dexterous Hand Model (UDHM), which maps human and robot hand states into a shared 22-DoF semantic interface. Based on UDHM, we introduce UniDexTok, a retargeting-free state tokenizer that learns embodiment-conditioned discrete tokens from standardized real joint states. UniDexTok provides a unified representation for heterogeneous dexterous hands without relying on retargeting or simulation data. Compared with the recent baseline UniHM, UniDexTok reduces MPJAE from 15.63 degrees to 0.16 degrees and MPJPE from 18.51 mm to 0.18 mm, corresponding to error reductions of 98.98% and 99.03%, respectively. These results improve reconstruction from centimeter-scale to sub-millimeter accuracy. Experiments further show that data from other embodiments improves target-embodiment reconstruction accuracy, demonstrating the benefit of cross-embodiment tokenization. UniDexTok also shows strong zero-shot and few-shot reconstruction ability when new dexterous hands are introduced.