Research

My research integrates Neurosymbolic AI, Advanced Air Mobility (AAM), and optimization to support interpretable, safety-aware decision systems.

I focus on methods that are technically strong, explainable to stakeholders, and useful in real transportation planning environments where demand, infrastructure, policy, and operational constraints interact.

Core Research Areas

Advanced Air Mobility Demand Modeling

I study regional and urban demand forecasting for Advanced Air Mobility systems, including trip-generation models, airport-connected urban air mobility, regional air mobility, and mobility-energy coordination. This work supports infrastructure planning questions such as where portals or vertiports should be located, how demand varies over time, and how AAM services could connect with existing transportation networks.

Neurosymbolic and Trustworthy AI

My neurosymbolic AI research combines neural learning with symbolic knowledge, rules, and constraints to improve interpretability, robustness, and controllability. I am especially interested in methods that make machine learning models more useful for safety-aware decision support, including rule-guided prediction, symbolic knowledge distillation, and hybrid reasoning workflows.

Forecasting and Optimization for Operations

I develop forecasting and optimization methods for planning under operational constraints, including temporal demand prediction, scenario analysis, resource allocation, scheduling, and disaster-response mobility planning. A recurring theme is using AI to accelerate search and improve decision quality while preserving transparency for planners and domain experts.

Research Themes

Advanced Air Mobility Forecasting and Infrastructure Planning

Advanced Air Mobility introduces new planning challenges because future demand depends on travel behavior, airport access, regional connectivity, energy availability, weather, policy, and public adoption. My work develops data-driven models that estimate potential AAM demand across urban and regional settings, with an emphasis on models that can support practical infrastructure decisions.

This includes airport-focused Urban Air Mobility demand modeling for New York City, regional air mobility demand modeling in Tennessee, and broader studies of opportunities and open challenges in AAM demand forecasting. The goal is to connect predictive modeling with decisions about where AAM services may be useful, how systems should be evaluated, and what constraints must be included before deployment.

Related outputs: AAM publications and conference presentations.

Neurosymbolic AI for Interpretable Decision Support

Many high-impact AI systems need more than accurate predictions. They also need explanations, constraints, uncertainty awareness, and the ability to incorporate domain knowledge. My neurosymbolic AI research explores how symbolic reasoning, rule structures, and knowledge-guided modeling can be integrated with neural networks and modern machine learning pipelines.

In transportation and AAM applications, this direction supports interpretable demand prediction, safety-aware planning, and model behavior that can be inspected by domain experts. More broadly, my work examines symbolic knowledge distillation, neurosymbolic reinforcement learning and planning, and the role of neurosymbolic methods in robust and trustworthy AI systems.

Transportation Forecasting, Optimization, and Resilience

Transportation systems require decisions under uncertainty, especially when planning for disruptions, emergencies, and evolving mobility demand. I work on forecasting and optimization pipelines that combine predictive models with search, scheduling, and constraint-aware planning methods for operational decision support.

This includes neural-accelerated genetic algorithms for pre-disaster mobility planning, multi-objective planning under constraints, and scenario-driven evaluation for resilient transportation operations. The broader objective is to make AI-assisted optimization more efficient, interpretable, and useful for real planning workflows.

Active Projects

NASA ULI: AAM Demand Modeling

Developing demand forecasting pipelines for mobility-energy coordinated AAM systems using data-driven and neurosymbolic approaches.

Rolling-horizon trip request generation for scalable simulation.
Temporal demand estimation for regional and urban scenarios.
Decision support for infrastructure and policy planning.

Neurosymbolic Methods for Transportation AI

Designing interpretable prediction models by combining symbolic structure with neural learning.

Knowledge-guided model design and rule-aware prediction.
Symbolic distillation and reasoning-integrated workflows.
Applications in demand prediction and planning support.

AI-Assisted Disaster Response Planning

Building optimization-driven frameworks to improve mobility response under emergency constraints.

Neural-accelerated genetic optimization strategies.
Multi-objective scheduling and constrained resource planning.
Scenario-driven analysis for resilient operations.

Methodology

Modeling: Time-series forecasting, predictive analytics, uncertainty-aware learning.
AI Techniques: Deep learning, reinforcement learning, neurosymbolic architectures.
Optimization: Multi-objective and constraint-aware optimization pipelines.
Validation: Scenario-based evaluation with practical transportation constraints.

Impact and Collaboration

I collaborate with academic and applied partners on interpretable AI for transportation and critical infrastructure contexts. I am open to joint projects involving:

Advanced Air Mobility modeling and simulation.
Neurosymbolic and trustworthy AI methods.
Optimization for planning, resilience, and emergency response.

For publication details and outputs, see Publications. For profile and contact: Home and Contact.