EAGER: Toward Magnetic Manipulation of Nonmagnetic Objects

在过去的十年里，研究人员在使用磁铁精确移动物体而不需要任何物理接触方面取得了重大进展。然而，这些物体是特别设计的，主要由铁磁性材料制成，这样它们就可以被磁铁移动。但是，如果有可能使用磁铁来操纵根本不含磁性材料的金属物体，这在许多工程设备中很常见呢？众所周知，移动的磁铁会产生电场，而电场反过来又会通过产生涡流在金属物体中产生力。该项目探索如何利用这些力来控制这些物体的运动，使微重力和海底操作成为可能，并为科学研究提供新工具。该项目系统地描述了非磁性但导电球体（固体和薄壁）的磁性操纵的物理特性。对于用于在球体中产生涡流的时变磁场，使用旋转磁偶极子场，这是由首席研究员在NSF事先支持下开发的独特电磁和永磁体硬件平台产生的。该项目将基于涡流的操作描述为机器人操作问题，以确定偶极驱动源的数量和位置如何影响可操作性。该项目测试了一个假设，即球体可以通过学习模型在有限的先验知识下被操纵。S参数，同时使用防护动作，以避免失去控制权限。这种联合学习和控制问题被表述为一个信息论的、主动学习的问题，以寻找有信息的样本进行学习，并受到系统电磁物理所施加的未知操纵约束。最后，对该现象在实际应用中的可行性进行了研究。该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

Over the past decade, researchers have made significant advances on using magnets to precisely move objects without any physical contact. However, the objects are specially designed to be made mostly of ferromagnetic material so that they can be moved by magnets. But what if it were possible to use magnets to manipulate metal objects that contain no magnetic material at all, which is common in many engineered devices? It is well known that moving magnets create electrical fields that, in turn, can generate forces in metal objects through the generation of eddy currents. The project explores how these forces can be used to control the movement of such objects, enabling manipulation in microgravity and undersea, and enabling new tools for scientific research.This project methodically characterizes the physics of magnetic manipulation of nonmagnetic-but-conductive spheres (solid and thin walled). For the time-varying magnetic fields used to generate eddy currents in the spheres, rotating magnetic dipole fields are used, which are generated by unique electromagnetic and permanent-magnet hardware platforms developed by the Principal Investigator with prior NSF support. The project characterizes eddy-current-based manipulation as a robotic-manipulation problem, in order to determine how the number and placement of the dipole actuation sources affects manipulability. The project tests the hypothesis that a sphere can be manipulated with limited prior knowledge by learning the model?s parameters while using guarded actions to avoid loss of control authority. This joint learning and control problem is formulated as an information-theoretic, active-learning problem to find informative samples for learning, subject to unknown manipulation constraints imposed by the electromagnetic physics of the system. Finally, the project includes a feasibility study in the practical application of this phenomenon.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.