GORDON: Graph-based Object-centric Rewards for Decomposition of Long-Horizon Manipulation

作者：Andrea Protopapa, Davide Buoso, Francesca Pistilli, Georgia Chalvatzaki, Giuseppe Averta · 单位：Interactive Robot Perception & Learning (PEARL) Lab, Computer Science Dept., TU Darmstadt, Germany., Robotics Institute Germany (RIG). · 会议/期刊：arXiv preprint · 方向：cs.RO · 发布日期：2026-08-05