ManiVID-3D

Abstract

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