A Dual-System Vision-Language-Action Architecture for Whole-Upper-Body Humanoid Control
Noch nicht auf Deutsch verfügbar: Anzeige im englischen Original.
In 30 Sekunden
A slow VLM planner (7–9 Hz) and a fast visuomotor policy (200 Hz) control a full humanoid upper body from language.
Forschungsfrage
How to get language understanding and fast reactive control in one deployable stack?
Problem
Large VLMs are too slow for 200 Hz torque control.
Bisheriger Ansatz
Scripted skills or single-rate policies.
Neuer Ansatz
Two systems communicating via a latent vector; trained on ~500 h of teleop.
Ergebnisse
Multi-robot collaboration and picking of thousands of novel household items.
Grenzen
No public benchmark; single-embodiment; teleop data cost.
Bedeutung für die Branche
The architecture pattern is now widely copied; watch actuator bandwidth as the next bottleneck.