VLAct: representation-centric VLA pre-training
Loading video
Loading videoVLAct turns a fixed robot-data budget into transferable visual-action knowledge through multi-embodiment representation-centric continued pre-training, improving simulation, real-world, and unseen-embodiment transfer.
VLActVLACross-EmbodimentMulti-EmbodimentRepresentation LearningRobot LearningSimulationTransfer LearningHumanoidVision-language-action (VLA)
Category: research
Author: @yokoshaan
Date: 2026-09-05T00:00:00
Duration: 17.2s
Reference: https://arxiv.org/abs/2608.27550
