# RealD2iff: Bridging Real-World Gap in Robot Manipulation via Depth Diffusion

arXiv: https://arxiv.org/abs/2511.22505 · HTML: https://mlsanity.com/paper/2511.22505/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 20 of its 60 references that are in the map.
- First posted on arXiv 27 Nov 2025.

## Builds on (most cited first, 20 of 20)

- [Auto-Encoding Variational Bayes](https://mlsanity.com/paper/1312.6114/) (2013)
- [OpenVLA: An Open-Source Vision-Language-Action Model](https://mlsanity.com/paper/2406.09246/) (2024)
- [π0: A Vision-Language-Action Flow Model for General Robot Control](https://mlsanity.com/paper/2410.24164/) (2024)
- [Learning Fine-Grained Bimanual Manipulation with Low-Cost Hardware](https://mlsanity.com/paper/2304.13705/) (2023)
- [Depth Anything V2](https://mlsanity.com/paper/2406.09414/) (2024)
- [What Matters in Learning from Offline Human Demonstrations for Robot Manipulation](https://mlsanity.com/paper/2108.03298/) (2021)
- [Universal Manipulation Interface: In-The-Wild Robot Teaching Without In-The-Wild Robots](https://mlsanity.com/paper/2402.10329/) (2024)
- [Mobile ALOHA: Learning Bimanual Mobile Manipulation with Low-Cost Whole-Body Teleoperation](https://mlsanity.com/paper/2401.02117/) (2024)
- [Unleashing Large-Scale Video Generative Pre-training for Visual Robot Manipulation](https://mlsanity.com/paper/2312.13139/) (2023)
- [Vision-Language Foundation Models as Effective Robot Imitators](https://mlsanity.com/paper/2311.01378/) (2023)
- [Robot Parkour Learning](https://mlsanity.com/paper/2309.05665/) (2023)
- [3D Diffusion Policy: Generalizable Visuomotor Policy Learning via Simple 3D Representations](https://mlsanity.com/paper/2403.03954/) (2024)
- [Humanoid Parkour Learning](https://mlsanity.com/paper/2406.10759/) (2024)
- [Agile But Safe: Learning Collision-Free High-Speed Legged Locomotion](https://mlsanity.com/paper/2401.17583/) (2024)
- [RoboGSim: A Real2Sim2Real Robotic Gaussian Splatting Simulator](https://mlsanity.com/paper/2411.11839/) (2024)
- [Visual Whole-Body Control for Legged Loco-Manipulation](https://mlsanity.com/paper/2403.16967/) (2024)
- [Generalizable Humanoid Manipulation with 3D Diffusion Policies](https://mlsanity.com/paper/2410.10803/) (2024)
- [GenSim2: Scaling Robot Data Generation with Multi-modal and Reasoning LLMs](https://mlsanity.com/paper/2410.03645/) (2024)
- [Re3Sim: Generating High-Fidelity Simulation Data via 3D-Photorealistic Real-to-Sim for Robotic Manipulation](https://mlsanity.com/paper/2502.08645/) (2025)
- [Bridging the Sim-to-Real Gap for Athletic Loco-Manipulation](https://mlsanity.com/paper/2502.10894/) (2025)

## Abstract (arXiv)

Robot manipulation in the real world is fundamentally constrained by the visual sim2real gap, where depth observations collected in simulation fail to reflect the complex noise patterns inherent to real sensors. In this work, inspired by the denoising capability of diffusion models, we invert the conventional perspective and propose a clean-to-noisy paradigm that learns to synthesize noisy depth, thereby bridging the visual sim2real gap through purely simulation-driven robotic learning. Building on this idea, we introduce RealD2iff, a hierarchical coarse-to-fine diffusion framework that decomposes depth noise into global structural distortions and fine-grained local perturbations. To enable progressive learning of these components, we further develop two complementary strategies: Frequency-Guided Supervision (FGS) for global structure modeling and Discrepancy-Guided Optimization (DGO) for localized refinement. To integrate RealD2iff seamlessly into imitation learning, we construct a pipeline that spans six stages. We provide comprehensive empirical and experimental validation demonstrating the effectiveness of this paradigm. RealD2iff enables two key applications: (1) generating real-world-like depth to construct clean-noisy paired datasets without manual sensor data collection. (2) Achieving zero-shot sim2real robot manipulation, substantially improving real-world performance without additional fine-tuning.