# IMPACT: Attention Is the Interaction Map for Scalable Interaction-Aware World Model Training

arXiv: https://arxiv.org/abs/2609.00161 · HTML: https://mlsanity.com/paper/2609.00161/

## Key facts

- Cited by 1 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 33 of its 45 references that are in the map.
- First posted on arXiv 31 Aug 2026.

## Placed in

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

## Builds on (most cited first, 30 of 33)

- [Qwen2.5 Technical Report](https://mlsanity.com/paper/2412.15115/) (2024)
- [Wan: Open and Advanced Large-Scale Video Generative Models](https://mlsanity.com/paper/2503.20314/) (2025)
- [CogVideoX: Text-to-Video Diffusion Models with An Expert Transformer](https://mlsanity.com/paper/2408.06072/) (2024)
- [HunyuanVideo: A Systematic Framework For Large Video Generative Models](https://mlsanity.com/paper/2412.03603/) (2024)
- [Towards Accurate Generative Models of Video: A New Metric & Challenges](https://mlsanity.com/paper/1812.01717/) (2018)
- [Cosmos World Foundation Model Platform for Physical AI](https://mlsanity.com/paper/2501.03575/) (2025)
- [Genie: Generative Interactive Environments](https://mlsanity.com/paper/2402.15391/) (2024)
- [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)
- [VACE: All-in-One Video Creation and Editing](https://mlsanity.com/paper/2503.07598/) (2025)
- [EgoDex: Learning Dexterous Manipulation from Large-Scale Egocentric Video](https://mlsanity.com/paper/2505.11709/) (2025)
- [Matrix-game 2.0: An open-source real-time and streaming interactive world model](https://mlsanity.com/paper/2508.13009/) (2025)
- [HunyuanVideo 1.5 Technical Report](https://mlsanity.com/paper/2511.18870/) (2025)
- [Pandora: Towards General World Model with Natural Language Actions and Video States](https://mlsanity.com/paper/2406.09455/) (2024)
- [WorldArena: A Unified Benchmark for Evaluating Perception and Functional Utility of Embodied World Models](https://mlsanity.com/paper/2602.08971/) (2026)
- [Geometry-aware 4D Video Generation for Robot Manipulation](https://mlsanity.com/paper/2507.01099/) (2025)
- [Multiplayer Interactive World Models with Representation Autoencoders](https://mlsanity.com/paper/2607.05352/) (2026)
- [Hand2World: Autoregressive Egocentric Interaction Generation via Free-Space Hand Gestures](https://mlsanity.com/paper/2602.09600/) (2026)
- [iWorld-Bench: A Benchmark for Interactive World Models with a Unified Action Generation Framework](https://mlsanity.com/paper/2605.03941/) (2026)
- [WorldVLN: Autoregressive World Action Model for Aerial Vision-Language Navigation](https://mlsanity.com/paper/2605.15964/) (2026)
- [Qwen-RobotWorld Technical Report: Unifying Embodied World Modeling through Language-Conditioned Video Generation](https://mlsanity.com/paper/2606.17030/) (2026)
- [LOME: Learning Human-Object Manipulation with Action-Conditioned Egocentric World Model](https://mlsanity.com/paper/2603.27449/) (2026)
- [STARRY: Spatial-Temporal Action-Centric World Modeling for Robotic Manipulation](https://mlsanity.com/paper/2604.26848/) (2026)
- [PhysisForcing: Physics Reinforced World Simulator for Robotic Manipulation](https://mlsanity.com/paper/2606.28128/) (2026)
- [AirScape: An Aerial Generative World Model with Motion Controllability](https://mlsanity.com/paper/2507.08885/) (2025)
- [Robotic VLA Benefits from Joint Learning with Motion Image Diffusion](https://mlsanity.com/paper/2512.18007/) (2025)
- [AlayaWorld: Long-Horizon and Playable Video World Generation](https://mlsanity.com/paper/2607.06291/) (2026)
- [WorldScape Policy 2.0: Empowering Steerable World Action Modeling with Reasoning-Augmented Memory and In-Context Learning](https://mlsanity.com/paper/2607.18840/) (2026)
- [Open-world Hand-Object Interaction Video Generation Based on Structure and Contact-aware Representation](https://mlsanity.com/paper/2512.01677/) (2025)
- [EA-WM: Event-Aware Generative World Model with Structured Kinematic-to-Visual Action Fields](https://mlsanity.com/paper/2605.06192/) (2026)
- [Worldscape-MoE: A Unified Mixture-of-Experts World Model for Scalable Heterogeneous Action Control](https://mlsanity.com/paper/2607.03964/) (2026)

## Built on by (most cited first, 1 of 1)

- [Code Plans, Diffusion Renders: Open-Ended Generative World Modeling](https://mlsanity.com/paper/2609.26458/) (2026)

## Abstract (arXiv)

World models have made remarkable progress in action-conditioned future prediction for embodied agents, yet still struggle to model physically plausible interactions. Existing approaches address this limitation by constraining the generation process with external representations encoding motion, geometry, or semantics. Obtaining these spatiotemporally dense representations typically requires auxiliary estimators or manual annotations, limiting training scalability. We instead revisit the training objective and identify a supervision-allocation mismatch under the globally averaged mean squared error (MSE) denoising objective: prevalent static content dominates the optimization signal, leaving sparse dynamic-object regions critical to interaction generation disproportionately under-supervised. Motivated by this observation, we introduce IMPACT, a scalable Interaction-aware Model training framework with Prior-guided Attention Calibration and Targeting. IMPACT uses cross-attention associated with manipulated-object tokens as an internal spatiotemporal prior for action-conditioned changes. It samples candidate regions from this prior, calibrates them with detached local prediction errors to construct an interaction map, and uses the map to reweight denoising supervision, requiring neither external representations nor inference-time modifications. Extensive experiments on robot-arm and human-hand manipulation, spanning diverse control modalities and DiT backbones, show that IMPACT consistently outperforms the corresponding MSE-trained baselines, improving interaction fidelity, physical plausibility, and visual quality.