Current Research
Geometry-Informed CV for Calibration-Free Distance Estimation
Perspective-geometry–informed distance estimation from monocular images with benchmarking and ablation studies. Achieved MAE 12.6 ± 2.9 cm in cross-validation; includes uncertainty and error analyses and reproducible pipelines.
Dissertation Projects
Geometric Overtaking Detection with Early Warning
Geometry-informed overtaking detection and tracking (YOLOv5 + ByteTrack) with 98.7% precision, 98.1% recall, and 2.89 s early warning on 319 events (41.5k frames). Supports cyclist–vehicle safety analyses with robust event validation.
Aggressive Driving: Acceleration/Deceleration Profiles vs. Fuel Economy Cycles
Comparative analysis of real-world aggressive driving styles vs. EPA fuel economy cycles. Results inform sustainable driving practices and policy; presented at SAE WCX 2025.
Aggressive Driving Prediction from Longitudinal Jerk
Machine learning pipeline for classifying aggressive driving using speed-adjusted jerk features. Achieved 93.8% accuracy (Random Forest) across 556 trips with rigorous cross-validation.
Master's Thesis
Computational Human Performance Modeling using Queuing Network in an Open-Source Platform
This research implements an open-source version of the QN-MHP cognitive architecture using Python and SimPy for computational UI testing. The model simulates human cognition as a queuing network of information processors, providing designers a cost-effective tool to evaluate UI designs without extensive human subject testing. Model validation using visual-manual tasks demonstrated alignment with empirical data, making it a valuable resource for the computational human modeling community.
Research Portfolio: Completed Projects
Intelligent Vehicle Lane Change System
This research develops an intelligent lane change prediction system for autonomous vehicles using Temporal Convolutional Networks (TCN). Using real-world vehicle trajectory data from the NGSIM dataset, I created a robust deep learning model that achieved 98.66% accuracy in predicting lane change behaviors. The system's high performance and quick computational time make it particularly valuable for improving autonomous vehicle decision-making and safety in complex road scenarios.
Mixed Methods Study on Data Privacy in Messenger Applications: A Comparison between WhatsApp and Signal
This research explores user perceptions of data privacy in messaging applications, focusing on the balance between privacy, trust, and usability. Through a mixed-methods study comparing WhatsApp and Signal, we analyzed user attitudes toward data collection and privacy policies. Our findings revealed users' strong preference for transparency in data collection and aggregate data usage over personal data collection. The study highlights the growing importance of data privacy in user trust and messaging app selection.
An Autonomous Driving System - Dedicated Vehicle for People with ASD and their Caregivers
This project focuses on designing autonomous vehicles specifically tailored for individuals with Autism Spectrum Disorder (ASD). Through a human-centered design approach and extensive collaboration with caregivers, we developed a specialized ADS-DV (Automated Driving System - Dedicated Vehicle) prototype with customized interior design and a companion mobile application. The design addresses key needs including safety, monitoring, and individual preferences, aiming to enhance independent mobility for people with ASD.