RI: Small: Dexterous Manipulation Using Predictive Thin-Shell Modeling

RI：小：使用预测薄壳建模进行灵巧操纵

• 批准号: 1217904
• 负责人: Peter Allen
• 金额: $ 49.89万
• 依托单位: Columbia University
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2012
• 资助国家: 美国
• 起止时间: 2012-09-01 至 2016-08-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1217904&HistoricalAwards=false
• 关键词: RI; Small; Dexterous; Manipulation; Using

Grasping and manipulation of deformable objects presents a host of new research challenges that are much more demanding than for rigid objects. A particular challenge is to fully understand the physics of deformation and to model deformable objects in a way that can be used by real robotic systems in the presence of noise and uncertainty and with real-time constraints. This project will use offline simulation to predict states of deformable objects modeled as thin-shells (i.e. cloth, fabric, clothing) that can then be recognized by a robotic vision/grasping system to pick up and manipulate these objects. The Intellectual Merit includes a significant step forward in the synthesis of state-of-the-art numerical computation of plasto-elastica with a database-driven manipulation approach to allow robots to manipulate deformable objects. This includes fabric simulation technology that provides the requisite level of accuracy at speeds amenable to online computation parallel to the robotic grasping task. The Broader Impacts of this research include 1) creation of an open source, extensible, 3-D database of deformable objects for dissemination to the robotics and graphics communities, 2) extending the range of working environments for the emerging field of personal robotic assistants, and 3) developing and educating a new class of scientists who bridge the fields of computer graphics, computational mechanics, and robotics.

可变形物体的抓取和操纵提出了一系列新的研究挑战，这些挑战比刚性物体的要求更高。一个特殊的挑战是充分理解变形的物理原理，并以一种可以在存在噪声和不确定性且具有实时约束的情况下由真实机器人系统使用的方式对可变形物体进行建模。该项目将使用离线模拟来预测建模为薄壳（即布料、织物、衣服）的可变形物体的状态，然后机器人视觉/抓取系统可以识别这些物体，以拾取和操纵这些物体。智力优势包括在将最先进的塑性弹性体数值计算与数据库驱动的操纵方法相结合方面向前迈出了重要一步，使机器人能够操纵可变形物体。 这包括织物模拟技术，该技术能够以适合与机器人抓取任务并行的在线计算的速度提供必要的精度水平。 这项研究的更广泛影响包括 1) 创建一个开源、可扩展的可变形物体 3D 数据库，用于向机器人和图形界传播；2) 扩大个人机器人助理新兴领域的工作环境范围；3) 培养和教育新一代科学家，他们在计算机图形学、计算力学和机器人学领域之间架起桥梁。