# SplineWAM: Adaptive Action Horizons for World Action Models via B-Spline Representations

arXiv: https://arxiv.org/abs/2609.39873 · HTML: https://mlsanity.com/paper/2609.39873/

## 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 25 of its 37 references that are in the map.
- First posted on arXiv 30 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, 25 of 25)

- [Flow Matching for Generative Modeling](https://mlsanity.com/paper/2210.02747/) (2022)
- [Wan: Open and Advanced Large-Scale Video Generative Models](https://mlsanity.com/paper/2503.20314/) (2025)
- [World Action Models are Zero-shot Policies](https://mlsanity.com/paper/2602.15922/) (2026)
- [UniVLA: Learning to Act Anywhere with Task-centric Latent Actions](https://mlsanity.com/paper/2505.06111/) (2025)
- [RoboCasa: Large-Scale Simulation of Everyday Tasks for Generalist Robots](https://mlsanity.com/paper/2406.02523/) (2024)
- [Fast-WAM: Do World Action Models Need Test-time Future Imagination?](https://mlsanity.com/paper/2603.16666/) (2026)
- [Latent Action Pretraining from Videos](https://mlsanity.com/paper/2410.11758/) (2024)
- [Causal World Modeling for Robot Control](https://mlsanity.com/paper/2601.21998/) (2026)
- [LIBERO-Plus: In-depth Robustness Analysis of Vision-Language-Action Models](https://mlsanity.com/paper/2510.13626/) (2025)
- [mimic-video: Video-Action Models for Generalizable Robot Control Beyond VLAs](https://mlsanity.com/paper/2512.15692/) (2025)
- [GigaWorld-Policy: An Efficient Action-Centered World--Action Model](https://mlsanity.com/paper/2603.17240/) (2026)
- [Unified 4D World Action Modeling from Video Priors with Asynchronous Denoising](https://mlsanity.com/paper/2604.26694/) (2026)
- [villa-X: Enhancing Latent Action Modeling in Vision-Language-Action Models](https://mlsanity.com/paper/2507.23682/) (2025)
- [AHA-WAM:Asynchronous Horizon-Adaptive World-Action Modeling with Observation-Guided Context Routing](https://mlsanity.com/paper/2606.09811/) (2026)
- [Native Video-Action Pretraining for Generalizable Robot Control](https://mlsanity.com/paper/2607.08639/) (2026)
- [BEAST: Efficient Tokenization of B-Splines Encoded Action Sequences for Imitation Learning](https://mlsanity.com/paper/2506.06072/) (2025)
- [τ0-WM: A Unified Video-Action World Model for Robotic Manipulation](https://mlsanity.com/paper/2606.01027/) (2026)
- [Latent Action Pretraining Through World Modeling](https://mlsanity.com/paper/2509.18428/) (2025)
- [FASTer: Toward Efficient Autoregressive Vision Language Action Modeling via Neural Action Tokenization](https://mlsanity.com/paper/2512.04952/) (2025)
- [ActionCodec: What Makes for Good Action Tokenizers](https://mlsanity.com/paper/2602.15397/) (2026)
- [Action Chunking and Exploratory Data Collection Yield Exponential Improvements in Behavior Cloning for Continuous Control](https://mlsanity.com/paper/2507.09061/) (2025)
- [OAT: Ordered Action Tokenization](https://mlsanity.com/paper/2602.04215/) (2026)
- [VLA-Corrector: Lightweight Detect-and-Correct Inference for Adaptive Action Horizon](https://mlsanity.com/paper/2607.01804/) (2026)
- [OmniSAT: Compact Action Token, Faster Auto Regression](https://mlsanity.com/paper/2510.09667/) (2025)
- [B-spline Policy: Accelerating Manipulation Policies via B-spline Action Representations](https://mlsanity.com/paper/2607.09648/) (2026)

## Abstract (arXiv)

World action models (WAMs) are large embodied policies that jointly predict future video and the actions to execute, emitting a fixed-length action chunk per inference call. Such a policy allocates its computational budget uniformly in time, unable to execute for longer over free-space motion or to spend more inference on contact-rich manipulation, which limits the throughput a WAM can reach when served in the cloud. We present SplineWAM, which adaptively compresses the action trajectory into a fixed-size window of cubic B-spline parameters, fitting the knot times to the characteristics of the motion. One parameter budget then decodes into chunks of varying temporal resolution and duration, and both the executed span and the interval until the next policy call follow from the prediction itself. Aligning the video supervision to the fitted knot times of the demonstration rather than to a uniform grid concentrates the supervised frames where the action trajectory is complex. For asynchronous deployment we introduce Jacobian-Pullback Real-Time Chunking (JP-RTC), which imposes chunk continuity on the decoded raw actions the robot executes rather than on the spline parameters, and corrects the parameters through the decoder so that the executed prefix agrees with the actions already committed. On LIBERO-Plus and RoboCasa, SplineWAM improves success rate over an action chunking WAM by 8.2 and 4.4 points while cutting policy calls per episode by 22% and 26%. On three bimanual real-robot tasks under asynchronous execution, it leads or matches the baseline while decoding 1.2 to 1.6 times as much executed motion per call.