Skip navigation
August 15, 2018

New engineering center CRISP makes three seed grant awards

The recently-established College of Engineering CRISP center has made three seed grant awards in the first year of its seed grant competition.

A recently-established College of Engineering center has made three seed grant awards in the first year of its seed grant competition. Researchers with the Center for Resilient Infrastructures, Systems, and Processes (CRISP) develop solutions to questions such as: What causes some systems - computing, cyber physical, or large-scale engineered systems - to be resilient to disruptions of various kinds? And what causes some systems to “bounce back” from a failure quickly? The projects chosen for seed funding will address different aspects of these broad questions.

  1. Shreyas Sundaram (ECE) and Shaoshuai Mou (AAE), "Resilient Distributed Information Fusion in Hostile and Dynamic Environments," Domain area: Cyber physical systems.
    “Development of resilient strategies in a fully distributed scenario has been a very challenging problem. This seed funding from CRISP will provide us with a critical boost to pursue our high-risk high-reward research idea at just the right time – we are very excited to leverage this opportunity to explore this important topic!”
  1. Ilias Bilionis (ME) and Shirley Dyke (CE/ME), "Automating Exposure and Probabilistic Vulnerability Quantification for Assets in the Built Environment using Street-View Images,” Domain area: Built  environment.
    “Accelerating the evaluation of residential structures after disasters is critical for allowing homeowners to more rapidly recover and resume normal activities. Through this project we will exploit the vast amount of data collected in recent hurricanes to automate these procedures using powerful machine learning methods.”
  1. Dengfeng Sun (AAE) and Xiao Wang (Statistics), "Resilient Operations of Unmanned Aerial Vehicle Systems," Domain area: Cyber physical systems.
    “The last decade has witnessed the ever-increasing activities of unmanned aerial vehicle (UAVs), or commonly known as drones, in the civil aviation domain. There is a lack of research on resilient operations of such UAV systems with the inevitable interaction of traditional manned aircraft. This seed fund from CRISP will provide an opportunity to address this important issue by bringing together expertise from statisticians and aerospace engineers. Both the PI and Co-PI are aviation enthusiasts and are very excited to explore this critical domain in order for a safe and resilient aviation system.”
from l to r: Xiao Wang (Statistics), Dengfeng Sun (AAE), Shreyas Sundaram (ECE), Saurabh Bagchi (CRISP Director and ECE faculty), Shirley Dyke (CE/ME), Shaoshuai Mou (AAE) (not pictured, Ilias Bilionis (ME)

The seed grant competition was held in July and the winners were notified in early August. Proposals were judged by a panel of reviewers from the CRISP leadership team. Each award supports a half-time Graduate Research Assistant for AY 2018-19.

The center involves faculty members in leadership roles from multiple engineering departments: Saurabh Bagchi as Director from Electrical and Computer Engineering, Jitesh Panchal and Milind Kulkarni as Associate Directors, from Mechanical Engineering and Electrical and Computer Engineering respectively. Gesualdo Scutari from Industrial Engineering serves as the thrust lead on optimization and Felix Lin from Electrical and Computer Engineering on cyber-physical systems. There are 20 affiliate faculty members in the 2017-19 cohort.

Abstracts and viewgraphs for the winning proposals are listed below:

Resilient Distributed Information Fusion in Hostile and Dynamic Environments
Shreyas Sundaram (ECE) and Shaoshuai Mou (AAE) (Download viewgraph)

The proposed research aims to establish the foundations for a fundamentally new method for swarms of autonomous agents to perform distributed fusion of observations in hostile and dynamic environments.  The novelty of the research stems from bridging the gap between information-theoretic and control-theoretic approaches for dealing with resilient information fusion problems. In the short term, the project aims to create a novel scalable approach for combining high-dimensional information from multiple agents in the network in a way that mitigates the influence of malicious agents, without needing to explicitly prevent, detect, or identify malicious behavior. This approach will form the building block for scalable and distributed information fusion algorithms that are resilient against Byzantine attacks in highly time-varying networks. Over the long term, the project will integrate information- and control-theoretic approaches into a formal framework to enable attack-resilient fusion by large heterogeneous teams. In addition, the project will validate the proposed algorithms on a platform consisting of 3DR Parrot Drones.

Automating Exposure and Probabilistic Vulnerability Quantification for Assets in the Built Environment Using Street-View Images
Ilias Bilionis (ME) and Shirley Dyke (CE/ME) (Download viewgraph)

2017 was the costliest year on record regarding natural disasters. 16 disaster events surpassed $1B in damages, and the total cost of damages in all events exceeded $300B (so far). The degree and breadth of damage in 2017 hurricanes show that cities need more rigorous decision-making approaches to improve their resilience and mitigate the impact of such events. Pre-event information related to the built environment, empirically-calibrated vulnerability models, and rapidly available post-event information are critical data to formulate resilience-oriented decision-making problems. The objective of this collaboration between the Predictive Science Lab and the Intelligent Infrastructure Systems Lab is to improve the resilience of the built environment to hurricanes through an automated vision-based system that generates actionable information in the form of probabilistic pre-event prediction and post-event assessment of damages. The system will depend on automatically extracted pre-event street-view images and will leverage real-world image recognition methods to automatically identify critical visual features needed to associate buildings with pre-determined probabilistic vulnerability functions. These vulnerability models, calibrated using detailed post-disaster data collected from recent hurricanes in the U.S., will then be used to predict the post-event state of infrastructure across the city. With this automated capability in place, government agencies, insurance companies, and our communities will be able to rapidly and consistently assess vulnerability and quantify exposure (pre-event), and assess damage and process claims (post-event). 

Resilient Operations of Unmanned Aerial Vehicle Systems
Dengfeng Sun (AAE) and Xiao Wang (Statistics) (Download viewgraph)

This project will bring together expertise on aerospace engineering and statistical science to design a system-theoretic framework for modeling, analysis, and resilient control strategies for unmanned aerial vehicle systems (UASs). The principal investigator, Dr. Dengfeng Sun has the experience in large-scale infrastructure system management and resilient control for aerospace systems, while co-investigator Dr. Xiao Wang is an expert in stochastic systems and queuing theory. This research will hopefully take advantage of the complimentary expertise from Dr. Sun and Dr. Wang in order to achieve resilient operations of the UASs, including conflict detection and resolution and the management of congested airspace sections for both manned and unmanned aerial vehicles (UAVs) with limited communication capabilities. The project will focus on a class of stochastic perturbations resulting from uncertain capacity of individual airspace sectors and airports. The modeling framework captures the UAV traffic characteristics under these perturbations, and new analysis and resilient control design tools to improve the system performance will be developed. Aiming for the resilient operations for large-scale infrastructure systems, the investigators and their student expect to contribute to the theory of stochastic switching systems, control of continuous-time Markov processes, and game-theoretic models of queuing systems.