Thursday, August 8, 2024, 10 AM Central
Please register to receive the webinar connection information via this form: https://forms.illinois.edu/sec/659324035
ABSTRACTS
Presentation by Fellow Veronica Diaz Pacheco, North Carolina State University
Title: Analysis of the Grid’s Vulnerability to Coordinated Cyberattacks
Abstract: As shown on December 2, 2022, in Moore County, North Carolina, a failure to adequately plan for unforeseen disruptions in power sources can have tremendous consequences on human life. The issue of protecting our energy sources has never been more relevant as the grid – primarily due to the race towards decarbonization – becomes more central to modern life (via electrification of transportation, building energy, and manufacturing), and as the grid becomes more reliant on technologies that may increase its attack surface. In this study, we examine how the vulnerability of the bulk power system to intentional attacks could change due to decarbonization, specifically a shift away from centralized electricity production at large fossil fuel power plants towards more “distributed” generation via wind and solar. We explore this question as a network interdiction problem, a two-player sequential game model where an adversary aims to maximally damage a cost-minimizing system operator by destroying grid components (in our case, electrical substations). We generate representative operating scenarios (or conditions) for a real 662-bus bulk power system by performing a clustering procedure on annual load and renewable generation data. We analyze the attacker’s optimal strategy in today’s grid and future grid configurations that rely significantly more on wind and solar. We account for differences in the attack feasibility/cost as well as impact across substation types (e.g., power plant size, urban vs. rural service areas). Our results shed light on how priorities for investing in protective measures may change as a function of the evolution of the grid.
Presentation by Fellow Vlas Zyrianov, University of Illinois
Title: Asset-free LiDAR Simulator to Train Robust and Safe Autonomy
Abstract: Vlas will present LidarDM, a novel LiDAR generative model capable of producing realistic, layout-aware, physically plausible, and temporally coherent LiDAR videos. LidarDM stands out with two unprecedented capabilities in LiDAR generative modeling: (i) LiDAR generation guided by driving scenarios, offering significant potential for autonomous driving simulations, and (ii) 4D LiDAR point cloud generation, enabling the creation of realistic and temporally coherent sequences. At the heart of our model is a novel integrated 4D world generation framework. Specifically, we employ latent diffusion models to generate the 3D scene, combine it with dynamic actors to form the underlying 4D world, and subsequently produce realistic sensory observations within this virtual environment. Our experiments indicate that our approach outperforms competing algorithms in realism, temporal coherency, and layout consistency. We additionally show that LidarDM can be used as a generative world model simulator for training and testing perception models.
________________________________________________________________________________
BIOGRAPHIES
Biography for fellow Veronica Diaz Pacheco, North Carolina State University
Veronica is a third-year doctoral student and researcher in the Operations Research program at North Carolina State University. She earned her bachelor’s degree in economics with a minor in statistics from the University of Puerto Rico at Rio Piedras and her master’s degree in industrial engineering from the University of Puerto Rico at Mayagüez. Veronica’s research interests include mathematical modeling and programming, emergency management, simulation, network modeling, and energy systems.
After experiencing Hurricane Maria firsthand during September 2017, Veronica became deeply interested in the field of decision-making under disruptions. The devastating impacts of the hurricane on her homeland Puerto Rico sparked an interest for finding ways to enhance the resilience of critical infrastructure. This experience, and her passion for learning and mathematics, led her to pursue advanced studies in operations research, where she could develop and apply mathematical models to solve real-world problems.
Under the supervision of her advisor, Dr. Jordan Kern, Veronica is now exploring how advanced optimization methods can be used to improve the resilience of power systems. She is particularly focused on bi-level programming and the integration of renewable energy sources to evaluate and enhance the resilience of bulk power grid operations.
In addition to her academic pursuits, Veronica is involved in several professional organizations. She is a member of the Institute for Operations Research and the Management Sciences (INFORMS), and currently serves as the President of the Operations Research Graduate Student Association at NC State. She regularly attends conferences and workshops to stay updated on the latest developments in her field, and she is actively working with students and faculty to create a sense of community and enhance diversity within her academic circle.
When she’s not working, you can find her reading, jewelry making or rock climbing. Veronica believes in maintaining a healthy work-life balance and enjoys engaging in activities that allow her to either unwind and recharge or learn something new.
Biography for fellow Vlas Zyrianov, University of Illinois
Vlas Zyrianov is a fourth-year PhD candidate at the University of Illinois Urbana-Champaign in Computer Science, where he is supervised by Dr. Shenlong Wang. His research focuses on developing generative machine learning models for applications in self-driving vehicles. Specifically, his work has investigated the utilization of generative models to produce synthetic LiDAR data for use in training and evaluating autonomous vehicle (AV) perception and planning models; the application of generative models in modeling LiDAR sensor readings for sampling and unsupervised probabilistic inference tasks (such as point-cloud densification); and how generative models can be leveraged to provide priors for perception models.
Vlas earned his bachelor’s degree in computer science from Kent State University, where he conducted research in software engineering, code analysis, and eye tracking for software engineering research under the supervision of Dr. Jonathan Maletic.
His research has been presented at numerous international conferences, including the European Conference on Computer Vision (ECCV), the International Conference on Computer Vision (ICCV), the IEEE International Conference on Software Maintenance and Evolution (ICSME), the ACM/IEEE International Workshop on Eye Movements in Programming (EMIP), the IEEE International Conference on Program Comprehension (ICPC), and the Society for Industrial and Organizational Psychology (SIOP). His work has been published in six conference proceedings and journal publications.
He has gained industry experience through three internships at Nvidia, working on OpenGL/SPIR-V graphics compiler development, and has also worked at the augmented reality startup AiR Everywhere (Kent, OH).
His work has been recognized with a ICSME 2020 IEEE TCSE Distinguished paper award, 2022 National Science Foundation Graduate Research Fellowship Program (NSF GRFP) Honorable Mention, and first place at the 2023 University of Illinois Urbana-Champaign Coordinated Science Laboratory student conference in the “Machine Learning and Signal Processing” category.
He is actively involved in academic service by serving as the co-chair of the Machine Learning and Signal Processing section of the 2024 CSL Student Conference, reviewing for international conferences (ECCV’24, CVPR’24, ICRA’24, IROS’23, ETRA’19, ICSME’18, ICSME’17), and conference volunteering (IEEE ICSME’19).
________________________________________________________________________________
Please register to receive the webinar connection information via this form: https://forms.illinois.edu/sec/659324035
This research is part of the Blue Waters sustained-petascale computing project, which is supported by the National Science Foundation (awards OCI-0725070 and ACI-1238993) the State of Illinois, and as of December, 2019, the National Geospatial-Intelligence Agency. Blue Waters is a joint effort of the University of Illinois at Urbana-Champaign and its National Center for Supercomputing Applications.