Grasp and Manipulation Planning in the Presence of Dynamics and Uncertainty

该项目是伦斯勒理工学院（RPI）和宾夕法尼亚大学（U Penn）的计算机科学家、数学家和工程师的多学科合作成果。研究的重点将是发展新的规划技术，以应付没有间歇性接触就无法成功完成的任务。这样的任务被称为操作任务。操作任务，特别是那些动态影响系统性能和结果的操作任务，在机器人技术中是非常重要的。机器人执行这些任务的能力完全改变了它的性格，从被动的观察者转变为主动的参与者，能够执行有用的工作，无论是在建筑物倒塌的现场协助第一响应者，还是在微型设备中组装零件。这项研究需要同时开发新的算法和数学模型，以分析和预测具有摩擦、柔顺和间歇接触的多体系统的运动。这些发展是在关注实际工程问题（通过实验验证）的情况下完成的，这些问题驱动了数学模型的形式，进而驱动了算法和数学。更广泛的影响这个项目的结果直接适用于运动和操纵的一大类问题，这是正在进行的人形机器人研究的核心，以及在制造环境中的固定和装配。在教育方面，研究团队正在宾夕法尼亚大学和RPI之间开发一门新的研究生水平的操纵计划分析课程，该课程将由合作pi共同教授。

This project is a multidisciplinary effort of computer scientist, mathematicians, and engineers at Rensselaer Polytechnic Institute (RPI) and University of Pennsylvania (U Penn). The focus of the research will be on the development of new planning techniques for tasks that cannot be performed successfully without intermittent contacts. Such tasks are referred to as manipulation tasks. Manipulations tasks, particularly those in which the effects of dynamics influence the system performance and outcome, are exceedingly important in robotics. The ability of a robot to perform such tasks completely changes its character from passive observer to active participant capable of performing useful work, whether it be assisting a first- responder on the scene of a building collapse or assembling parts in micro-scale devices.Intellectual MeritThis research requires the simultaneous development of new algorithms and mathematical models for analyzing and predicting the motion of multibody systems with frictional, compliant, intermittent contacts. These developments are being done with attention paid to practical engineering concerns (verified by experiments) that drive the form of mathematical models, and in turn, the algorithms and mathematics.Broader ImpactsThe results of this project are directly applicable to a large class of problems in locomotion and manipulation that is central to ongoing research on humanoid robots, and on fixturing and assembly in manufacturing settings. On the educational front, the research team is developing a new graduate-level course on Manipulation Planning Analysis, across UPenn and RPI, that will be co-taught by the co-PIs.