CEDex：基于类人接触表征的大规模跨本体灵巧抓取生成
CEDex: Cross-Embodiment Dexterous Grasp Generation at Scale from Human-like Contact Representations
查看机构详情
- Department of Engineering, King’s College London（伦敦国王学院工程系）
- Imperial College London（帝国理工学院）
- College of Software, Nankai University（南开大学软件学院）
机构由 AI 辅助整理，请以论文原文为准。
浏览论文内容
中文总结 AI 辅助
本文提出 CEDex，通过类人接触表征对齐、拓扑合并与 SDF 物理约束优化，实现跨机械手形态的大规模灵巧抓取生成，并构建含 50 万物体、2000 万抓取的数据集。
中文摘要 AI 辅助
跨本体灵巧抓取合成是指为具有不同形态的各种机械手自适应地生成和优化抓取。这种能力对于在多样环境中实现通用机器人操作至关重要，并且需要大量可靠且多样的抓取数据，才能有效训练模型并实现鲁棒泛化。然而，现有方法要么依赖缺乏类人运动学理解的基于物理的优化，要么需要大量人工数据采集流程，且仅限于拟人结构。本文提出 CEDex，这是一种新颖的大规模跨本体灵巧抓取合成方法，通过将机器人运动学模型与生成的类人接触表征对齐，在人类抓取运动学和机器人运动学之间建立桥梁。给定物体点云和任意机械手模型，CEDex 首先使用在人类接触数据上预训练的条件变分自编码器生成类人接触表征。随后，它通过拓扑合并执行运动学人体接触对齐，将多个人手部分整合为统一的机器人部件，接着进行带有物理感知约束、基于符号距离场的抓取优化。利用 CEDex，我们构建了迄今为止最大的跨本体抓取数据集，涵盖四种夹爪类型上的 50 万个物体，共包含 2000 万次抓取。大量实验表明，CEDex 优于最先进方法，并且我们的数据集以高质量、多样化抓取促进了跨本体抓取学习。
英文摘要
Cross-embodiment dexterous grasp synthesis refers to adaptively generating and optimizing grasps for various robotic hands with different morphologies. This capability is crucial for achieving versatile robotic manipulation in diverse environments and requires substantial amounts of reliable and diverse grasp data for effective model training and robust generalization. However, existing approaches either rely on physics-based optimization that lacks human-like kinematic understanding or require extensive manual data collection processes that are limited to anthropomorphic structures. In this paper, we propose CEDex, a novel cross-embodiment dexterous grasp synthesis method at scale that bridges human grasping kinematics and robot kinematics by aligning robot kinematic models with generated human-like contact representations. Given an object's point cloud and an arbitrary robotic hand model, CEDex first generates human-like contact representations using a Conditional Variational Auto-encoder pretrained on human contact data. It then performs kinematic human contact alignment through topological merging to consolidate multiple human hand parts into unified robot components, followed by a signed distance field-based grasp optimization with physics-aware constraints. Using CEDex, we construct the largest cross-embodiment grasp dataset to date, comprising 500K objects across four gripper types with 20M total grasps. Extensive experiments show that CEDex outperforms state-of-the-art approaches and our dataset benefits cross-embodiment grasp learning with high-quality diverse grasps.
