aiqbench analysis
This research addresses a critical latency bottleneck in Vision-Language-Action (VLA) models by introducing AHEAD, a wrapper that uses a latent world model to predict object motion before execution. By forecasting future scene tokens based on optical flow and uncertainty thresholds, the method enables robotic arms to successfully intercept moving objects like conveyor items and projectiles where standard baselines fail entirely. The approach demonstrates significant performance gains in both simulation and physical hardware without retraining the core 7B parameter VLA backbone.
Intercepting the Future: Latent-Space Predictive World Model for Dynamic VLA Manipulation
arXiv:2606.02486v2 Announce Type: replace Abstract: Vision-Language-Action (VLA) models generalize across static manipulation but fail when objects move during task execution. They map the current observation to an action and assume the scene is stationary between observation and execution, so at any non-trivial object speed the resulting latency exceeds the time available to grasp. We close this gap with AHEAD (Anticipatory Horizon Extrapolation with Adaptive Dynamics), a predict-then-act wrapper that augments a frozen VLA with a motion-aware latent world model. A small world model trained on manipulation video forecasts future patch tokens in the VLA's feature space, conditioned on per-token velocity and acceleration from optical flow. A language-and-motion saliency mask concentrates prediction on task-relevant patches, and the model rolls forward for an adaptive horizon, halting when prediction uncertainty crosses a threshold. The frozen action decoder then receives the predicted future tokens in place of the current ones. AHEAD adds 4.9M parameters to a frozen 7B OpenVLA and reaches 79 to 97% success across 20 dynamic simulation scenarios where the strongest baseline reaches 31 to 58%. On a physical UFactory xArm 7, AHEAD succeeds on 29/30 to 30/30 on three conveyor and rolling-ball tasks, 23/30 on paddle interception, and 19/30 on projectile catching where every baseline scores 0/30.
Abstract:Vision-Language-Action (VLA) models generalize across static manipulation but fail when objects move during task execution. They map the current observation to an action and assume the scene is stationary between observation and execution, so at any non-trivial object speed the resulting latency exceeds the time available to grasp. We close this gap with AHEAD (Anticipatory Horizon Extrapolation with Adaptive Dynamics), a predict-then-act wrapper that augments a frozen VLA with a motion-aware latent world model. A small world model trained on manipulation video forecasts future patch tokens in the VLA's feature space, conditioned on per-token velocity and acceleration from optical flow. A language-and-motion saliency mask concentrates prediction on task-relevant patches, and the model rolls forward for an adaptive horizon, halting when prediction uncertainty crosses a threshold. The frozen action decoder then receives the predicted future tokens in place of the current ones. AHEAD adds 4.9M parameters to a frozen 7B OpenVLA and reaches 79 to 97% success across 20 dynamic simulation scenarios where the strongest baseline reaches 31 to 58%. On a physical UFactory xArm 7, AHEAD succeeds on 29/30 to 30/30 on three conveyor and rolling-ball tasks, 23/30 on paddle interception, and 19/30 on projectile catching where every baseline scores 0/30.
Accepted at the 10th Conference on Robot Learning (CoRL 2026). 31 pages, 7 figures, 19 tables
Robotics (cs.RO)
I.2.9; I.2.6; I.2.10
arXiv:2606.02486 [cs.RO]
(or arXiv:2606.02486v2 [cs.RO] for this version)
https://doi.org/10.48550/arXiv.2606.02486 Focus to learn more arXiv-issued DOI via DataCite
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 311 words
Related coverage
Runway’s WorldPrompt and the Engineering of Real-Time Worlds
Latent Space · Sep 25, 2026
LaWAM: Latent World Action Models for Efficient Dynamics-Aware Robot Policies
arXiv · Jun 16, 2026
SG-WAM: Self-Guided World Modeling in Geometry-Aware Policy Space
arXiv · Aug 4, 2026
IDOL: Inverse-Dynamics-Guided Future Prediction for End-to-End Autonomous Driving
arXiv · Sep 29, 2026