Research map / Video / Video generation
EmbodiedVAE: Disentangled Video VAE for Efficient and Controllable Embodied Manipulation
Jiayi Luo, Hanxin Zhu, Chen Gao, Jiankun Wang, Cong Wang, Tianyu He, Jianxin Li, Zhibo Chen
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add arm-mask supervision during training (ablation) | EmbodiedVAE · Agibot-2025 (VAE reconstruction) | PSNR: 31.1812 -> 31.6745 | Improved |
| add arm-mask supervision during training (ablation) | EmbodiedVAE · Bridge (VAE reconstruction) | PSNR: 29.6680 -> 30.6957 | Improved |
| add arm-mask supervision during training (ablation) | EmbodiedVAE · RT-1 (action-conditioned generation, IRASim-L) | PSNR: 23.8044 -> 24.7082 | Improved |
| add arm-mask supervision during training (ablation) | EmbodiedVAE · Bridge (action-conditioned generation, IRASim-L) | PSNR: 23.7306 -> 24.2484 | Improved |
| add OT-based motion consistency loss (ablation) | EmbodiedVAE · Agibot-2025 (VAE reconstruction) | PSNR: 31.0213 -> 31.6745 | Improved |
| add OT-based motion consistency loss (ablation) | EmbodiedVAE · Bridge (VAE reconstruction) | PSNR: 29.9769 -> 30.6957 | Improved |
| add OT-based motion consistency loss (ablation) | EmbodiedVAE · RT-1 (action-conditioned generation, IRASim-L) | PSNR: 23.8187 -> 24.7082 | Improved |
| add OT-based motion consistency loss (ablation) | EmbodiedVAE · Bridge (action-conditioned generation, IRASim-L) | PSNR: 23.9726 -> 24.2484 | Improved |
| add OT-based motion consistency loss (ablation) | EmbodiedVAE · Agibot-2025 (VAE reconstruction) | LPIPS: 0.0835 -> 0.0723 | Improved |
| add OT-based motion consistency loss (ablation) | EmbodiedVAE · Bridge (action-conditioned generation, IRASim-L) | LPIPS: 0.1643 -> 0.1409 | Improved |
| add arm-mask supervision during training (ablation) | EmbodiedVAE · Bridge (action-conditioned generation, IRASim-L) | LPIPS: 0.1705 -> 0.1409 | Improved |
| use EmbodiedVAE instead of Wan-VAE as the LDM VAE | IRASim-L · RT-1 (action-conditioned generation, IRASim-L) | PSNR: 22.7334 -> 24.7082 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 0 of 1 on most shared results.
- VidTwin — VidTwin: Video VAE with Decoupled Structure and Dynamics (CVPR 2025) · better on 0 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- DiT4DiT: Jointly Modeling Video Dynamics and Actions for Generalizable Robot Control (NeurIPS 2026 · similarity 0.89)
- Robots Pre-train Robots: Manipulation-Centric Robotic Representation from Large-Scale Robot Datasets (ICLR 2025 · similarity 0.89)
- DynScene: Scalable Generation of Dynamic Robotic Manipulation Scenes for Embodied AI (CVPR 2025 · similarity 0.89)
- EnerVerse: Envisioning Embodied Future Space for Robotics Manipulation (NeurIPS 2025 · similarity 0.88)
- RotVLA: Rotational Latent Action for Vision-Language-Action Model (NeurIPS 2026 · similarity 0.88)
- VidMan: Exploiting Implicit Dynamics from Video Diffusion Model for Effective Robot Manipulation (NeurIPS 2024 · similarity 0.88)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
