CVAIMay 8

One Token Per Frame: Reconsidering Visual Bandwidth in World Models for VLA Policy

arXiv:2605.0793180.71 citations
Predicted impact top 29% in CV · last 90 daysOriginality Incremental advance
AI Analysis

For VLA policy learning, this work shows that extreme visual bandwidth reduction to one token per frame is feasible and beneficial under a constrained adaptation budget.

OneWM-VLA compresses each visual frame into a single semantic token via Adaptive Attention Pooling and uses a unified flow-matching objective for latent stream and action trajectory, improving average success rate from 47.9% to 61.3% on MetaWorld MT50, reaching 95.6% on LIBERO-Long, and 60.0% on a real deformable task.

Vision-language-action (VLA) models increasingly rely on auxiliary world modules to plan over long horizons, yet how such modules should be parameterized on top of a pretrained VLA remains an open design question. Existing world-model-augmented VLAs typically pass the per-frame visual stream into the world module at high visual bandwidth and treat its rollout as a side product of action prediction; under a constrained adaptation budget on a frozen backbone, this leaves both the per-frame representation and the latent action coupling under-examined. We introduce OneWM-VLA, which compresses each view into a single semantic token per frame through an Adaptive Attention Pooling, and produces the resulting latent stream and the action trajectory under a single flow-matching objective rather than connecting them through a separate decoder. Empirically, we find that per-frame visual bandwidth can be reduced to a single token without compromising long-horizon performance under our setup. Trained with 14.71M LoRA parameters on a $π_0$ (2B) backbone, OneWM-VLA improves the average success rate from 47.9% to 61.3% on MetaWorld~MT50, reaches 95.6% on LIBERO-Long (vs.85.2% for $π_0$), and reaches 60.0% on the long-horizon deformable task Fold Cloth on a real Piper arm (vs.20.0% for $π_0$).

Foundations

The foundational work for this paper's niche, ranked by how specifically the neighbourhood builds on it — not by global fame.

Your Notes