DexGraspNet 2.0：在大规模合成杂乱场景中学习生成式灵巧抓取
DexGraspNet 2.0: Learning Generative Dexterous Grasping in Large-scale Synthetic Cluttered Scenes
- Peking University（北京大学）
- UC Berkeley（加州大学伯克利分校）
- Beijing Academy of Artificial Intelligence（北京智源人工智能研究院）
机构由 AI 辅助整理，请以论文原文为准。
AI总结：
针对杂乱场景下灵巧抓取数据稀缺的问题，研究提出大规模合成基准DexGraspNet 2.0及基于局部几何条件扩散模型的两阶段生成抓取方法，仿真表现优于基线，零样本 sim-to-real 迁移后现实抓取成功率达90.7%。
AI中文摘要：
由于数据稀缺，在杂乱场景中使用灵巧手抓取仍然极具挑战性。为解决这一问题，我们提出了一个大规模合成基准，包含1319个物体、8270个场景和4.27亿次抓取。除基准测试外，我们还提出了一种新颖的两阶段抓取方法，通过使用以局部几何为条件的扩散模型从数据中高效学习。我们提出的生成方法在仿真实验中优于所有基线。此外，借助测试时深度恢复，我们的方法实现了零样本仿真到现实的迁移，在杂乱场景中取得了90.7%的现实世界灵巧抓取成功率。
英文摘要：
Grasping in cluttered scenes remains highly challenging for dexterous hands due to the scarcity of data. To address this problem, we present a large-scale synthetic benchmark, encompassing 1319 objects, 8270 scenes, and 427 million grasps. Beyond benchmarking, we also propose a novel two-stage grasping method that learns efficiently from data by using a diffusion model that conditions on local geometry. Our proposed generative method outperforms all baselines in simulation experiments. Furthermore, with the aid of test-time-depth restoration, our method demonstrates zero-shot sim-to-real transfer, attaining 90.7% real-world dexterous grasping success rate in cluttered scenes.
