Pretrained generative Diffusion Transformers (DiTs) capture rich pixel-level visual and language-conditioned structure through large-scale image and video generation training. A growing line of robot policies builds on this generative prior, but how it should be transferred to control remains unclear, and existing approaches commonly instantiate this transfer through future visual prediction. We ask a more basic question: what a pretrained generative DiT actually contributes to action learning, and how this prior should be adapted for control. We introduce NowWAM, a future-target-free co-training formulation that denoises the current observation and predicts robot actions from the same visual stream, directly coupling the native generative objective to the action-facing representation across the denoising trajectory. Under matched controlled settings, past and future visual targets perform comparably, while restricting training to the clean endpoint substantially reduces robustness, suggesting that a separate future target is not essential for generative adaptation, while the denoising trajectory remains an effective interface for control. On LIBERO-Plus, NowWAM reaches 87.7% with FLUX2-Klein, improving over the future-target co-training baseline by 6.1 points while halving training visual tokens (784 to 392) and reducing step time from 2.85 s to 1.63 s, a 1.8x speedup. With the pure text-to-image Z-Image backbone, NowWAM further reaches 87.8%, showing that strong control adaptation is not tied to video generation or image-editing backbones.
Figures & tables
Figure 1: From future prediction to current denoising. (a) Future prediction is used for action prediction. (b) Future targets are used for generative co-training. (c) NowWAM denoises the current visual stream and predicts actions directly. (d) NowWAM improves robustness while reducing training cost.
Figure 2: NowWAM. During training, the current visual stream is sampled along the pretrained denoising trajectory and jointly supports generative prediction and action learning. At inference, the same policy operates at the clean endpoint ( σ=0 ) in a single forward pass.
Method
Spatial
Object
Goal
Long
Avg.
π0.5
98.8
98.2
98.0
92.4
96.9
X-VLA
98.2
98.6
97.8
97.6
98.1
LingBot-VA
98.5
99.6
97.2
98.5
98.5
Motus
96.8
99.8
96.6
97.6
97.7
Fast-WAM
98.2
100.0
97.0
95.2
97.6
ImageWAM
97.2
99.2
98.8
98.4
98.4
Table 1: Standard LIBERO and RoboCasa GR1 manipulation.
Perturbation category
Overall
Method
Camera
Robot
Lang.
Light
Bkg.
Noise
Layout
SR (%)
π0
13.8
6.0
58.8
85.0
81.4
79.0
68.9
53.6
π0.5
78.4
73.6
80.8
96.2
94.1
89.0
84.5
84.4
StarVLA
52.5
49.8
88.5
95.7
95.7
73.0
76.9
74.1
OpenVLA-OFT
56.4
31.9
79.5
88.7
93.3
75.8
74.2
69.6
ABot-M0
60.4
67.9
86.4
96.2
91.6
86.4
82.6
80.5
Table 2: Success rates (%) on LIBERO-Plus across seven perturbation categories.
Figure 3: Qualitative rollouts under distribution shift. We show representative trajectories under RobotInit and camera-view perturbations in LIBERO-Plus. Fast-WAM and ImageWAM exhibit distinct grounding and execution failures, whereas NowWAM completes the corresponding tasks. These examples qualitatively illustrate the robustness gains observed in Table 2 .
Init.
Visual loss
SR
Random
Edit
58.14
Pretrained
Edit
83.05
Random
None
50.55
Pretrained
None
75.81
Table 3: All rows are strictly matched; only the stated intervention changes. Results use non-EMA checkpoints on the same fixed 1,923-episode subset.
Method
Training proxy
Vis. tokens ↓
Step time (s) ↓
Peak mem. (GiB) ↓
ImageWAM
Future image
784
2.85
78.9
NowWAM
Current denoising
392 (50% fewer)
1.63 (1.8 × faster)
74.3 (5.8% lower)
Table 4: Training efficiency. Same 2× H200 setup with global batch size 64. Both methods use FLUX2-Klein-4B as the backbone.
Figure 4: Task-conditioned attention across robustness and transfer settings. Each row starts with the RGB observation, followed by attention maps over the rollout.
Generative visual models offer a foundation for learning representations of physical dynamics, yet their extension to continuous control raises a fundamental question: do visual prediction and action generation require separate computational pathways? Existing approaches usually introduce trainable action heads or separate action experts to bridge low-dimensional states and high-dimensional visual representations. In this work, we explore whether the visual backbone's existing capacity can also support control when actions are expressed in a compatible representation. Thus, we introduce PatchWAM (Patch World-Action Model), which treats continuous actions as another type of patch through a fixed mapping called Action-as-Patch. This allows a single model to predict both how the robot should move and what the scene may look like afterward. Visual prediction and action generation become parts of the same generative process, without a dedicated action head or separate action expert. Experiments with subsampled training windows show gains over a matched dual-expert control, while benchmark evaluations reach 91.8% success rate on LIBERO-Plus and 96.12% on RoboTwin 2.0 in a full-data setting with additional augmented demonstrations. More broadly, the result suggests that capability need not be added where it can be inherited: the constraint on extending a generative backbone is the interface a new signal is written in, not the capacity to model it.
Tianheng Wang, Zhou Xie, Heng Jia +3
Westlake University · Lanzhou University · Zhejiang University +2
Fast-WAM shows that video-action co-training improves control without generating future video at inference, making the representation from a single video diffusion Transformer forward central to action generation. However, future-observation prediction does not explicitly prioritize the future dynamics and visual structure needed for control. We present MT-WAM, which retains the original training objectives and adds complementary supervision for future two-dimensional point trajectories and visual features. A lightweight dual-stream branch copied from the video backbone's final blocks provides target-specific processing, while a structured attention mask prevents cross-stream attention. Motion-stream tokens supply additional dynamics conditions to the action expert. Future visual-feature prediction provides supervision in a feature space that captures object and spatial structure. This supervision trains the video backbone to provide more informative visual context for action generation under changing visual conditions, without adding visual-feature-stream tokens to action conditioning. At inference, MT-WAM uses video and motion caches computed once per replan and skips future-video prediction. Without additional embodied policy pretraining, MT-WAM achieves 98.2% success on LIBERO and 73.7% on LIBERO-Plus, exceeding Fast-WAM by 23.8 percentage points on the latter. On RoboTwin 2.0 Clean2Rand, Random success increases from 6.30% to 19.40%; across four real-world tasks, average success increases from 67.0% to 77.8%.
Yiguang Yang, Jiankun Peng, Xiaoming Wang +2
Aerospace Information Research Institute, Chinese Academy of Sciences · University of Chinese Academy of Sciences
Robust robot control benefits from explicitly modeling state transitions, but video-generation world action models (WAMs) introduce substantial deployment cost. Existing latent WAMs avoid explicit future generation, but often compress predictive representations or separate predictive modeling from the representations used for action generation. We introduce JEPA-WAM, a latent WAM built in a pretrained V-JEPA space, which couples latent transition prediction with continuous action generation through a shared predictor. JEPA-WAM predicts a spatially structured joint current-future target that captures task-shared visual temporal structure between current and future observations, while preserving dense patch-level correspondence. Through the shared predictor, transition supervision directly shapes the backbone, from which dedicated representations are extracted for action prediction. The same design can also be instantiated in pretrained VLA policies while preserving their original perception and action pathways. On LIBERO-Plus, JEPA-WAM achieves 79.2%, the best result without large-scale robot-policy pretraining, while its pretrained π0.5 instantiation reaches 86.3%, achieving the best overall performance. Experiments on RoboTwin 2.0 and real-world bimanual manipulation further demonstrate strong generalization under visual and spatial shifts.
Yihan Lin, Jiawei He, Shifeng Bao +6
School of Information, Renmin University of China, Beijing, China · XYZ Embodied AI, Beijing, China · Key Laboratory of Data Engineering and Knowledge Engineering, Beijing, China +3