# WAM-OPD: On-Policy Distillation for World Action Models

arXiv: https://arxiv.org/abs/2608.22364 · HTML: https://mlsanity.com/paper/2608.22364/

## Key facts

- Cited by 4 papers in the map (data through 14 Oct 2026).
- Placed in Embodied and robot world models, one of 11 directions in World models.
- Builds on 14 of its 43 references that are in the map.
- First posted on arXiv 23 Aug 2026.

## Placed in

- World models / [Embodied and robot world models](https://mlsanity.com/wm/embodied-and-robot-world-models/)
- Knowledge distillation / [Multimodal distillation](https://mlsanity.com/distill/multimodal-distillation/)

## Builds on (most cited first, 14 of 14)

- [RoboTwin 2.0: A Scalable Data Generator and Benchmark with Strong Domain Randomization for Robust Bimanual Robotic Manipulation](https://mlsanity.com/paper/2506.18088/) (2025)
- [Vision-OPD: Learning to See Fine Details for Multimodal LLMs via On-Policy Self-Distillation](https://mlsanity.com/paper/2605.18740/) (2026)
- [DiffusionOPD: A Unified Perspective of On-Policy Distillation in Diffusion Models](https://mlsanity.com/paper/2605.15055/) (2026)
- [Flow-OPD: On-Policy Distillation for Flow Matching Models](https://mlsanity.com/paper/2605.08063/) (2026)
- [Flash-WAM: Modality-Aware Distillation for World Action Models](https://mlsanity.com/paper/2606.05254/) (2026)
- [π_textttRL: Online RL Fine-tuning for Flow-based Vision-Language-Action Models](https://mlsanity.com/paper/2510.25889/) (2025)
- [VLA-OPD: Bridging Offline SFT and Online RL for Vision-Language-Action Models via On-Policy Distillation](https://mlsanity.com/paper/2603.26666/) (2026)
- [Hi-WM: Human-in-the-World-Model for Scalable Robot Post-Training](https://mlsanity.com/paper/2604.21741/) (2026)
- [WAM-RL: World-Action Model Reinforcement Learning with Reconstruction Rewards and Online Video SFT](https://mlsanity.com/paper/2606.17906/) (2026)
- [Guided Action Flow: Q-Guided Inference for Flow-Matching Vision-Language-Action Policies](https://mlsanity.com/paper/2607.02092/) (2026)
- [WCM: A World Critic Model for Vision-Language-Action Reinforcement Learning](https://mlsanity.com/paper/2607.29613/) (2026)
- [Disturbance observer-based tracking control for roll-to-roll slot die coating systems under gap and pump rate disturbances](https://mlsanity.com/paper/2601.08488/) (2026)
- [SignVLA: A Gloss-Free Vision-Language-Action Framework for Real-Time Sign Language-Guided Robotic Manipulation](https://mlsanity.com/paper/2602.22514/) (2026)
- [WorldSample: Closed-loop Real-robot RL with World Modelling](https://mlsanity.com/paper/2607.02431/) (2026)

## Built on by (most cited first, 4 of 4)

- [World-Action Models for Robot Learning and Control: A Survey](https://mlsanity.com/paper/2609.16074/) (2026)
- [Direct Experience World-Model Optimization: Learning the World Beyond Action Imitation](https://mlsanity.com/paper/2609.37398/) (2026)
- [RealtimeWAM: One-Step Asynchronous World Action Models](https://mlsanity.com/paper/2610.06617/) (2026)
- [OpenWAM: An Open Framework for Composable World-Action Models](https://mlsanity.com/paper/2610.07922/) (2026)

## Abstract (arXiv)

World Action Models (WAMs) generate both future video and robot actions, offering two connected outputs for post-training supervision. How can a pretrained WAM learn from a stronger Teacher on the histories it encounters during execution? We present WAM-OPD, which collects Student rollout histories and queries a Teacher for paired video and action targets. The Student learns from both targets while retaining its one-step video and action generation at deployment. Across 12 RoboTwin 2.0 tasks, WAM-OPD improves average success from 33.8% to 65.7%; across four real-robot tasks, it improves average success from 51.4% to 64.6%. With the collected Student histories held fixed, joint video-action supervision achieves the highest observed success on all three ablation tasks, while either modality alone also improves performance. A separate comparison with Teacher-generated histories finds task-dependent differences between the two history sources. These results demonstrate the value of paired video-action supervision for improving WAM policies without increasing their deployed sampling budget.