ManiVID-3D

摘要

Generalizable view-invariant reinforcement learning for robotic manipulation via disentangled 3D representations.