FMSG: Rapidly Repurposing and Scaling Assembly Lines Through the Co-design of Tooling and Controls of Robotic Work Cells

该项目允许重新利用和重新调整生产生产线的规模，以快速和经济地完成。典型的制造装配线由一系列工作单元组成，每个工作单元需要不同的机制来执行不同的流程。这种框架下的自动化生产通常是不灵活的——生产中的变化需要对组件进行耗时的重新设计和昂贵的重新编程。因此，自动化装配线生产无法对市场需求的快速变化做出反应。这不仅限制了产品更新和对市场动态的反应速度，而且还妨碍制造商在国家紧急情况期间迅速作出反应，而这种灵活性对于国家反应是必不可少的。该项目改变了装配线在工厂车间快速更新的方式，重点关注两个具体但广泛使用的自动化任务，使装配线更加智能，并对不断变化的产品线的生产做出反应，并有助于实现“制造即服务”的目标。该项目开发了一种有效的方法来共同设计特定任务的工具和控制，适用于制造系统，将数字制造与数字装配结合起来。该项目侧重于两项装配任务，重新定位和插入，以及它们作为网络物理子系统的工具和控制的共同设计。该项目借鉴了机器人、控制器优化、算法设计、软件和硬件、材料科学、机械工程和制造科学与工程等学科和技术。该项目采用基于模型的数据驱动方法进行对象操作，利用人工智能和机器学习技术，开发了一系列用于工具和控制协同设计的算法。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。

This project allows re-purposing and re-scaling of manufacturing production lines to be done quickly and affordably. A typical manufacturing assembly line is composed of a sequence of work cells, each requiring different mechanisms to execute different processes. Automated production in this framework is often inflexible – a change in the production requires the time-consuming re-design and expensive reprogramming of components. As a result, an automated assembly line production is unable to respond to rapid changes in market needs. This not only limits the speed of product updates and response to market dynamics, but also impedes manufacturers from reacting quickly during a national emergency when that flexibility is essential for a national response. This project changes how assembly lines can be rapidly updated on the factory floor, focusing on two specific but widely-used automated tasks, making assembly lines more intelligent and responsive to the production of changing products lines, and contributing to the goal of "manufacturing as a service."This project develops an efficient approach to co-design of task-specific tooling and control, applicable in general for manufacturing systems, uniting digital fabrication with digital assembly. The project focuses two assembly tasks, re-orientation and insertion, with their co-design of tooling and controls as cyber-physical sub-systems. The project draws upon the disciplines and technologies for robotics, controller optimization, algorithmic design, software and hardware, material science, mechanical engineering, and manufacturing science and engineering. The project takes a model-based data-driven approach to object manipulation, developing families of algorithms for co-design of tooling and controls, using techniques from artificial intelligence and machine learning.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.