Smart Human-centric Automation Resilience (SHARE) Lab

Research Overview

The United Nations estimates that by 2050 more than half of the world's population will be residing in urban areas. Consequently, sustainable and efficient operation of these urban areas has necessitated the emergence of "SMART CITY". However, to meet the needs of the stakeholders of such smart cities, namely the urban citizens, human-centric urbanization must be a priority.

• Safety, security, and resilience
• Sustainability
• Efficiency
• Accessibility and quality of life
• Citizen autonomy

Motivated by these goals, my research objective is to unify systems and control theory with machine learning to tighten safety and security, improve system resilience, and provide autonomy to citizens.

Resilient Human-centric Transportation Systems

Human-centric mobility is an integral part of a smart city. With increasing connectivity between smart infrastructures in a smart city, the human-centric transportation can be regarded as a Cyber-Physical-Social System (CPSS) or a Socio-Technical System (STS). However, such connectivity and inter-dependencies make smart cities increasingly vulnerable to cyberattacks and physical faults.

This has motivated my research in safety and security for smart mobility with an added focus on merging human-centric sensing and actuation with standard traffic measurements and management techniques. Our research goal is to ensure resilient operation of transportation systems in smart cities against cyberattacks using human-centric strategies.

Challenges in this area include: fusing social data with technical data, availability and integrity of social data, preprocessing of such data, vehicle-level vs traffic-level analysis, and the impact of human behavior on intelligent traffic flow.

Safeguarding Battery Energy Storage Systems

Our second research interest deals with challenges in ensuring safety in Battery Energy Storage Systems (BESS). These systems can facilitate greater adoption of sustainable energy technologies through penetration of reliable electric vehicles into the market or large-scale grid integration of renewables.

Designing intelligent management for these devices using both physics-based and data-driven approaches can potentially lead to safer and more reliable BESS. My research goal is to improve safe and efficient operation of energy storage systems.

Anomaly Characterization

The anomalies in dynamical systems can be categorized into three types:

• Faults: System malfunctions or defects leading to failures (e.g., CAV controller failure)
• Cyberattacks: Intelligent actions by adversaries to cause system disruption (e.g., hijacking CAVs)
• Outliers: Rare data/events presented to systems (e.g., camera jitters)

It is critical to detect, characterize, and classify these anomalies, as they can disrupt normal operations in systems like transportation or energy.

Furthermore, analysis of operational cyberattacks often raises questions about the source of the attack. Provenance is vital for prompt mitigation. Our research uses both model-based and model-free small data learning strategies to infer sources of attack injections such as sensor or actuation attacks.