World Model for Robot Learning: From Simulation to Reality

摘要

This work presents a unified world model framework for robot learning that bridges simulation and reality through domain-adaptive prediction, enabling efficient sim-to-real transfer for manipulation tasks.