Driving · 2026-10-14
1Step-WAM: Your World Action Model but Faster using 1-Step Action Generation
Rundong Su, Pengtao Chen, Haohan Geng, Sha Zhang, Zhihang Yuan, Yicheng Rong, Chong Yu, Hao Huang, Tao Chen
AI-assisted summary based on the paper. Consult the linked source for methods and limitations. Benchmark numbers appear after editor verification.
TL;DR
1Step-WAM speeds up World Action Models by converting multi-step action denoising into a single forward pass. It uses a lightweight calibrator to correct the first-step velocity, plus bandwidth-constrained (low-frequency DCT) action decoding. Because one-step action generation can run in parallel with visual prefill, the authors report large inference speedups on RoboTwin, LIBERO, and real-robot tasks while largely preserving multi-step success.
Why it matters
WAMs’ real-world use is limited by iterative action denoising and sequential “imagination-then-action” execution. The paper shows a geometric property (approximately straight denoising trajectories) and exploits it to produce one-step action outputs that can overlap with visual prefilling, aiming to remove this sequential bottleneck for high-frequency robotic control.
Method
- Analyze multi-step action denoising trajectories and argue the first-step velocity is a strong estimate of the final action direction (approximately straight denoising trajectory).
- Introduce 1Step-WAM: freeze the action DiT backbone and train a lightweight velocity calibration head to correct the initial predicted velocity toward the multi-step teacher endpoint.
- Add bandwidth-constrained action decoding by projecting the one-step action onto the first K temporal DCT modes, then overlap one-step action generation with video prefill using layer-wise cached visual K/V features and CUDA stream pipelining.
Abstract (from arXiv)
World Action Models (WAMs) unify visual world modeling and action generation, yet their real-world deployment is heavily bottlenecked by iterative action denoising and the sequential execution of visual prefilling and action prediction. We demonstrate that the action-denoising trajectories in WAMs are approximately straight, rendering the first-step velocity a strong estimate of the final action direction. Building on this observation, we introduce 1Step-WAM, a framework that transforms multi-step action denoising into one single forward. Specifically, we employ a lightweight calibrator to directly correct the first step velocity, and use bandwidth-constrained action decoding to further reduce errors in the calibrated one-step actions. Compressing action generation into a single step enables it to execute in parallel with visual prefilling, completely breaking the sequential bottleneck of the traditional imagination-then-action paradigm. Evaluations on RoboTwin and LIBERO show that 1Step-WAM largely preserves multi-step task success with only about 13.4M trainable parameters, while achieving inference speedups of $7.44\times$, $5.98\times$, and $24.84\times$ for FastWAM, ImageWAM and Kairos, respectively. Furthermore, real-world robotic deployments using the FastWAM backbone confirm a consistent 7.44$\times$ speedup while maintaining close success rate to the multi-step baseline. Code is avaliable at https://github.com/changdong1687/1Step-WAM.
Related papers
- Sparse-WAM: Accelerating World Action Models via Action-Guided Sparse Imagination
- RealtimeWAM: How Fast Can I Run My World Action Model?
- RealtimeWAM: One-Step Asynchronous World Action Models
- World Action Modeling with Progressive Visual Planning
- MARS: Systematic Multi-Level Acceleration for World Action Models
Publication record
Bibliographic evidence describes where and when a record appeared. It does not validate scientific claims.
First public release: 2026-10-11 · Indexed by Pub-AI: 2026-10-14 UTC
1Step-WAM: Your World Action Model but Faster using 1-Step Action Generation
Preprint posted: 2026-10-11 · Original record ↗
- arxiv
- 2610.15625
Explicit source versions (1)
- v1 · 2026-10-11
- Preprint available
Source date: 2026-10-11 · First detected: 2026-10-25 UTC · Historical baseline
Source evidence (1)
- Official evidence ↗ · atom:published (official metadata)
Abstract · arxiv
World Action Models (WAMs) unify visual world modeling and action generation, yet their real-world deployment is heavily bottlenecked by iterative action denoising and the sequential execution of visual prefilling and action prediction. We demonstrate that the action-denoising trajectories in WAMs are approximately straight, rendering the first-step velocity a strong estimate of the final action direction. Building on this observation, we introduce 1Step-WAM, a framework that transforms multi-step action denoising into one single forward. Specifically, we employ a lightweight calibrator to directly correct the first step velocity, and use bandwidth-constrained action decoding to further reduce errors in the calibrated one-step actions. Compressing action generation into a single step enables it to execute in parallel with visual prefilling, completely breaking the sequential bottleneck of the traditional imagination-then-action paradigm. Evaluations on RoboTwin and LIBERO show that 1Step-WAM largely preserves multi-step task success with only about 13.4M trainable parameters, while achieving inference speedups of $7.44\times$, $5.98\times$, and $24.84\times$ for FastWAM, ImageWAM and Kairos, respectively. Furthermore, real-world robotic deployments using the FastWAM backbone confirm a consistent 7.44$\times$ speedup while maintaining close success rate to the multi-step baseline. Code is avaliable at https://github.com/changdong1687/1Step-WAM.