CPS: Frontier: Collaborative Research: Data-Driven Cyberphysical Systems

CPS：前沿：协作研究：数据驱动的网络物理系统

• 批准号: 1645964
• 负责人: Alberto Sangiovanni-Vincentelli
• 金额: $ 30万
• 依托单位: University of California-Berkeley
• 依托单位国家: 美国
• 项目类别: Continuing Grant
• 财政年份: 2017
• 资助国家: 美国
• 起止时间: 2017-10-01 至 2020-09-30
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1645964&HistoricalAwards=false
• 关键词: CPS; Frontier; Collaborative; Research; Data

Data-driven cyber-physical systems are ubiquitous in many sectors including manufacturing, automotive, transportation, utilities and health care. This project develops the theory, methods and tools necessary to answer the central question "how can we, in a data-rich world, design and operate cyber-physical systems differently?" The resulting data-driven techniques will transform the design and operation process into one in which data and models - and human designers and operators - continuously and fluently interact. This integrated view promises capabilities beyond its parts. Explicitly integrating data will lead to more efficient decision-making and help reduce the gap from model-based design to system deployment. Furthermore, it will blend design- and run-time tasks, and help develop cyber-physical systems not only for their initial deployment but also for their lifetime.While proposed theory, methods and tools will cut across the spectrum of cyber-physical systems, the project focuses on their implications in the emerging application of additive manufacturing. Even though a substantial amount of engineering time is spent, additive manufacturing processes often fail to produce acceptable geometric, material or electro-mechanical properties. Currently, there is no mechanism for predicting and correcting these systematic, repetitive errors nor to adapt the design process to encompass the peculiarities of this manufacturing style. A data-driven cyber-physical systems perspective has the potential to overcome these challenges in additive manufacturing. The project's education plan focuses on the already much needed transformation of the undergraduate and graduate curricula to train engineers and computer scientists who will create the next-generation of cyber-physical with a data-driven mindset. The team will reach out to K-12 students and educators through a range of activities, and to undergraduate students from underrepresented groups through year-long research projects. All educational material generated by the project will be shared publicly.

数据驱动的网络物理系统在许多行业无处不在，包括制造业、汽车、交通运输、公用事业和医疗保健。这个项目开发了必要的理论、方法和工具来回答核心问题：在一个数据丰富的世界里，我们如何才能以不同的方式设计和操作网络物理系统？由此产生的数据驱动技术将把设计和操作过程转变为数据和模型--以及人类设计者和操作员--持续流畅地交互的过程。这种集成的观点承诺了超越其组成部分的功能。明确集成数据将导致更高效的决策，并有助于缩小从基于模型的设计到系统部署的差距。此外，它将融合设计和运行时任务，并帮助开发不仅用于初始部署，也用于其生命周期的网络物理系统。虽然提出的理论、方法和工具将跨越网络物理系统的范围，但该项目侧重于它们在添加剂制造的新兴应用中的影响。即使花费了大量的工程时间，添加的制造工艺通常也不能产生可接受的几何、材料或机电性能。目前，没有机制来预测和纠正这些系统的、重复的错误，也没有机制来调整设计过程以涵盖这种制造风格的特点。从数据驱动的网络物理系统的角度来看，有可能克服加法制造中的这些挑战。该项目的教育计划侧重于已经非常必要的本科生和研究生课程改革，以培养工程师和计算机科学家，他们将以数据驱动的心态创造下一代网络物理。该团队将通过一系列活动接触到K-12学生和教育工作者，并通过为期一年的研究项目接触到来自代表性不足群体的本科生。该项目产生的所有教育材料都将公开共享。

项目成果

A One-Class Support Vector Machine Calibration Method for Time Series Change Point Detection

• DOI: 10.1109/icphm.2019.8819385
• 发表时间: 2019-02
• 期刊: 2019 IEEE International Conference on Prognostics and Health Management (ICPHM)
• 作者: Baihong Jin;Yuxin Chen;Dan Li;K. Poolla;A. Sangiovanni-Vincentelli

Multi-source Domain Adaptation for Semantic Segmentation”, Advances in Neural Information Processing Systems

语义分割的多源域适应，神经信息处理系统的进展

• 发表时间: 2019
• 期刊: Advances in neural information processing systems
• 作者: Zhao, S.;Li, B.;Yue, X.;Gu, Y.;Xu, P.;Hu, R.;Chai, H.;Keutzer, K.
• 通讯作者: Keutzer, K.

Counterexample-Guided Data Augmentation

• DOI: 10.24963/ijcai.2018/286
• 发表时间: 2018-05
• 作者: T. Dreossi;Shromona Ghosh;Xiangyu Yue;K. Keutzer;A. Sangiovanni-Vincentelli;S. Seshia

A LiDAR Point Cloud Generator: from a Virtual World to Autonomous Driving

• DOI: 10.1145/3206025.3206080
• 发表时间: 2018-03
• 期刊: Proceedings of the 2018 ACM on International Conference on Multimedia Retrieval
• 作者: Xiangyu Yue;Bichen Wu;S. Seshia;K. Keutzer;A. Sangiovanni-Vincentelli