Research
Research Overview
My research focuses on developing mathematical and computational frameworks for understanding complex traffic systems under heterogeneous and disordered traffic conditions. My work integrates traffic flow theory, continuum modeling, numerical methods, machine learning, and empirical traffic data to investigate traffic dynamics in developing countries.
Research Areas
Two-Dimensional Traffic Flow Modeling
Traditional traffic flow models primarily focus on longitudinal vehicle interactions. My research extends these models to two-dimensional traffic streams by incorporating both longitudinal and lateral dynamics. The goal is to better represent lane-free traffic conditions commonly observed in urban environments.
- Second-order macroscopic traffic flow models
- Longitudinal and lateral vehicle interactions
- Traffic instability and wave propagation
- Numerical solutions of conservation laws
Lane-Free and Disordered Traffic Dynamics
Many traffic systems in developing countries operate without strict lane discipline. My research investigates the collective behavior emerging from such heterogeneous traffic streams.
- Multi-class traffic modeling
- Lateral movement behavior
- Congestion formation mechanisms
- Mixed traffic interactions
Drone-Based Traffic Data Collection
I have developed a large-scale aerial traffic dataset using drone-based observations of urban traffic environments in India. The dataset contains approximately 59 hours of trajectory data and supports both microscopic and macroscopic traffic analyses.
- Vehicle trajectory extraction
- Traffic state estimation
- Model calibration and validation
- Traffic behavior analysis
Physics-Informed Machine Learning
My recent work explores the integration of traffic flow theory and machine learning through Physics-Informed Neural Networks (PINNs).
- Traffic state reconstruction
- Physics-informed learning
- Data assimilation techniques
- Hybrid traffic modeling frameworks
Current Projects
Physics-Informed Neural Networks for Traffic Modeling
Development of PINN-based frameworks for estimating traffic density, velocity, and flow while satisfying governing conservation and momentum equations.
Research Methods
- Traffic Flow Theory
- Partial Differential Equations
- Finite Difference Methods
- Numerical Simulation
- Physics-Informed Neural Networks
- Machine Learning
- Computer Vision
- Data Analytics
Future Research Directions
My future research aims to integrate traffic flow theory, machine learning, and large-scale trajectory datasets to develop next-generation digital twins for urban transportation systems and intelligent mobility applications.