# WM-Craftnet: World Synesthesia Model for Generalizable and Robust Dexterous In-Hand Manipulation

arXiv: https://arxiv.org/abs/2609.07002 · HTML: https://mlsanity.com/paper/2609.07002/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in Embodied and robot world models, one of 11 directions in World models.
- Builds on 22 of its 25 references that are in the map.
- First posted on arXiv 7 Sep 2026.

## Placed in

- World models / [Embodied and robot world models](https://mlsanity.com/wm/embodied-and-robot-world-models/)

## Builds on (most cited first, 22 of 22)

- [Dream to Control: Learning Behaviors by Latent Imagination](https://mlsanity.com/paper/1912.01603/) (2019)
- [World Models](https://mlsanity.com/paper/1803.10122/) (2018)
- [Mastering Diverse Domains through World Models](https://mlsanity.com/paper/2301.04104/) (2023)
- [DayDreamer: World Models for Physical Robot Learning](https://mlsanity.com/paper/2206.14176/) (2022)
- [Learning joint reconstruction of hands and manipulated objects](https://mlsanity.com/paper/1904.05767/) (2019)
- [DeXtreme: Transfer of Agile In-hand Manipulation from Simulation to Reality](https://mlsanity.com/paper/2210.13702/) (2022)
- [In-Hand Object Rotation via Rapid Motor Adaptation](https://mlsanity.com/paper/2210.04887/) (2022)
- [General In-Hand Object Rotation with Vision and Touch](https://mlsanity.com/paper/2309.09979/) (2023)
- [Rotating without Seeing: Towards In-hand Dexterity through Touch](https://mlsanity.com/paper/2303.10880/) (2023)
- [Robot Synesthesia: In-Hand Manipulation with Visuotactile Sensing](https://mlsanity.com/paper/2312.01853/) (2023)
- [Robotic World Model: A Neural Network Simulator for Robust Policy Optimization in Robotics](https://mlsanity.com/paper/2501.10100/) (2025)
- [DextrAH-G: Pixels-to-Action Dexterous Arm-Hand Grasping with Geometric Fabrics](https://mlsanity.com/paper/2407.02274/) (2024)
- [ViTacFormer: Learning Cross-Modal Representation for Visuo-Tactile Dexterous Manipulation](https://mlsanity.com/paper/2506.15953/) (2025)
- [MoDem-V2: Visuo-Motor World Models for Real-World Robot Manipulation](https://mlsanity.com/paper/2309.14236/) (2023)
- [AnyRotate: Gravity-Invariant In-Hand Object Rotation with Sim-to-Real Touch](https://mlsanity.com/paper/2405.07391/) (2024)
- [World Model-based Perception for Visual Legged Locomotion](https://mlsanity.com/paper/2409.16784/) (2024)
- [Learning In-Hand Translation Using Tactile Skin With Shear and Normal Force Sensing](https://mlsanity.com/paper/2407.07885/) (2024)
- [DexNDM: Closing the Reality Gap for Dexterous In-Hand Rotation via Joint-Wise Neural Dynamics Model](https://mlsanity.com/paper/2510.08556/) (2025)
- [From Simple to Complex Skills: The Case of In-Hand Object Reorientation](https://mlsanity.com/paper/2501.05439/) (2025)
- [DexReMoE:In-hand Reorientation of General Object via Mixtures of Experts](https://mlsanity.com/paper/2508.01695/) (2025)
- [ViserDex: Visual Sim-to-Real for Robust Dexterous In-hand Reorientation](https://mlsanity.com/paper/2604.11138/) (2026)
- [DreamMimic: Learning Visuomotor Whole-Body Loco-Manipulation via World Model](https://mlsanity.com/paper/2608.22278/) (2026)

## Abstract (arXiv)

Generalizable and robust dexterous in-hand manipulation requires a policy to infer object pose, geometry, contact, and potential slip from partial and noisy observations. Although recent tactile and visuotactile RL methods achieve strong in-hand rotation in controlled settings, their robustness often degrades under pose shifts, force disturbances, and object variation. We propose WM-Craftnet, a world-model-conditioned framework that learns compact action-conditioned latent dynamics from proprioception, depth, tactile sensing, and actions, supervised by multimodal reconstruction and reward prediction. Rather than using the world model for latent imagination or policy optimization, WM-Craftnet uses the learned World Synesthesia Model (WSM) as recurrent task context for an asymmetric actor--critic policy. Importantly, WSM is trained to reconstruct clean depth targets from noisy depth inputs, providing a denoised geometric state for real-robot deployment. Ablations over recurrent baselines, auxiliary heads, tactile masking, and WSM modality heads show that predictive world modeling, clean-depth supervision, and tactile contact cues all shape the learned state. A WSM pretrained on nine (z)-axis objects serves as a reusable prior for (49)-object downstream policy learning. This context improves multi-object rotation, with quantitative and qualitative evidence for unseen-object, perturbation-recovery, and sim-to-real transfer.