unidoormanip.github.io - UniDoorManip: Learning Universal Door Manipulation Policy Over Large-scale and Diverse Door Manipulation Environments

Description: Deformable Neural Radiance Fields creates free-viewpoint portraits (nerfies) from casually captured videos.

nerf (140) d-nerf (48) nerfies (47)

Example domain paragraphs

Our Proposed Environment, Dataset and Universal Manipulation Policy . We build a novel door manipulation environment equipped with a large-scale door dataset covering 6 door categories with hundreds of door bodies and handles, and configure it with different realistic door manipulation mechanisms. To learn the universal door manipulation policy, we propose a novel framework which can generalize to unseen shapes and categories.

Learning a universal manipulation policy encompassing doors with diverse categories, geometries and mechanisms, is crucial for future embodied agents to effectively work in complex and broad real-world scenarios. Due to the limited datasets and unrealistic simulation environments, previous works fail to achieve good performance across various doors. In this work, we build a novel door manipulation environment reflecting different realistic door manipulation mechanisms, and further equip this environment wit