Fast Situational Awareness and Reliable Response with Heterogeneous Feedback and Number-Theoretic Control Primitives

点击翻译按钮获取中文摘要

This grant will fund research that enables engineered systems to effectively rely on and respond to real-time data collected from multiple diverse and asynchronous sources, with applications to medical device technology and aerospace manufacturing platforms, thereby promoting the progress of science, and advancing the national prosperity. Rapid and uniform timekeeping is the current dogma for building real-time systems. Such systems, however, increasingly operate in the presence of ubiquitous and diverse forms of data and asynchronous, often irregular workflows. Foundational knowledge about how to leverage such data streams in designing and controlling the behavior of the next generation of critical systems, including those meant to assist humans in data-intensive situations, is lacking. This project will overcome this knowledge gap by building a new theoretical and algorithmic framework that demonstrates how heterogeneous, often slow measurements can be reliably and quickly combined to achieve fast information retrieval and a robust feedback response. This framework will enable unsurpassed situational awareness and response agility beyond existing boundaries of process monitoring. A suite of related interactive demonstrations and dynamic visualizations will be developed and disseminated through web applications to foster a mindset of algorithmic thinking in students of all ages. Additional educational impact will result from integration of research results in advanced technical coursework and engagement of undergraduate and graduate students in research.This research aims to make fundamental contributions to a theory of how asynchronously sampled sensor data may be captured and combined to maximize real-time information gathering, increase process observability, and design rapid and robust feedback for closed-loop autonomy. It achieves this outcome by first developing a number-theoretic collaborative sampling approach for sensor management that applies to general dynamic systems and a broad range of process signals with mathematically expressible intra-sample correlations. Next, it derives conditions on the sampling that preserve observability and enable probabilistic optimal sampling subject to uncertainty. Finally, an outer loop controller based on the Youla-Kucera parameterization is tailored to achieve high-performance tracking beyond the Nyquist frequency of the individual sensors. The number-theoretic parameterizations and feedback control algorithms will be evaluated in experiments on laser-material interactions in powder bed fusion additive manufacturing and automated robotic inspection of complex metallic objects.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.