EAGER: Dexterous Robotic Cutting

这项研究将形成机器人切割和操纵柔软和不规则形状物体的理论。这些成果的应用正在导致更多熟练灵巧的机器人，面向一系列工业应用，改善安全风险，并提高可靠性和一致性。该项目的技术目标是深入了解如何操作精细、灵活和易滑的物品，有技能地处理工具，机械臂和手之间的协调，以及基于多模式传感和变形建模的运动规划和控制。该项目包括两个阶段。第一阶段开发切割设备的基本操作策略，解决包括视觉引导的手部放置、实时大变形建模、拾取和稳定以及灵巧抓取和重新抓取等问题。第二阶段研究两个机械臂-手对的协调进行切割，并综合轨迹和控制来实现不同的切割技能。这一阶段要解决的技术问题包括材料粘弹性、接触检测、动态切割规划和执行、手臂和手轨迹规划以及视觉和力数据的融合。

This research will form a theory of robotic cutting and manipulation of soft and irregularly shaped objects. Applications of the results are leading to more skillful dexterous robots geared for a range of industrial applications, improving safety risks, and increasing reliability and consistency. The technical goal of this project is to understand in depth about manipulation of delicate, flexible, and slippery items, handling of tools with skills, coordination among robotic arms and hands, and motion planning and control based on multi-modality sensing and deformable modeling. The project consists of two phases. The first phase develops strategies for basic maneuvers of cutting devices, tackling issues that include vision-guided hand placements, large deformable modeling in real time, pickup and stabilization, and dexterous grasping and re-grasping. The second phase investigates coordination of two robotic arm-hand pairs to carry out cutting, and synthesizes trajectories and controls to implement different cutting skills. Technical issues to address in this phase include material viscoelasticity, contact detection, cut planning and execution on the fly, arm and hand trajectory planning, and fusion of visual and force data.