About

Andrew Smyth is a faculty member at Columbia GSAPP. The page does not provide specific details about his research focus, background, or key contributions. Therefore, no further biographical information is available from the provided content.

Research topics

• Artificial Intelligence
• Computer Science
• Data Mining
• Mathematics
• Algorithm
• Applied mathematics
• Engineering
• Mathematical analysis
• Physics

Selected publications

• A Vision-Based Analysis of Congestion Pricing in New York City

  Open MIND · 2026-02-03

  preprintSenior author

  We examine the impact of New York City's congestion pricing program through automated analysis of traffic camera data. Our computer vision pipeline processes footage from over 900 cameras distributed throughout Manhattan and New York, comparing traffic patterns from November 2024 through the program's implementation in January 2025 until January 2026. We establish baseline traffic patterns and identify systematic changes in vehicle density across the monitored region.

  DOI

• A Vision-Based Analysis of Congestion Pricing in New York City

  ArXiv.org · 2026-02-03

  articleOpen accessSenior author

  We examine the impact of New York City's congestion pricing program through automated analysis of traffic camera data. Our computer vision pipeline processes footage from over 900 cameras distributed throughout Manhattan and New York, comparing traffic patterns from November 2024 through the program's implementation in January 2025 until January 2026. We establish baseline traffic patterns and identify systematic changes in vehicle density across the monitored region.

  PublisherOA PDF

• Effectiveness of the actuator line method in capturing unsteady gust interactions with tidal rotors

  2025-01-01

  articleOpen access

  Tidal turbines operate in unsteady flow environments that prompt complex flow interactions and challenge the validity of existing computational modelling methods.In this study, four numerical models of varying fidelity will be used to simulate a horizontal-axis tidal turbine under small-amplitude axial flow oscillations across a range of frequencies.The objective of the study is to assess the effectiveness of the different numerical models in capturing unsteady flow effects, as well as gaining insights regarding the strengths and shortcomings of each method.

  PublisherOA PDFDOI

• High Efficiency Swimming using Hydrodynamic Eigenmode Decomposition

  2025-01-01

  articleOpen access

  Underwater swimming can exhibit remarkable efficiency, as evidenced by adept aquatic animals.This study introduces a novel approach for systematically uncovering high-efficiency motion patterns in flexible three-dimensional rectangular bodies via hydrodynamic eigenmode decomposition.

  PublisherOA PDFDOI

• Impact assessment of construction methods and support structures on vibration levels of ancient reliefs

  npj Heritage Science · 2025-06-16

  articleOpen access

  The preservation of ancient stone reliefs in museums is a substantial concern in the field of heritage conservation, especially for art that has been altered and removed from its original configuration. This case study was conducted in advance of active construction work, through a series of pilot (exploratory) tests, to assess vibration to which the art objects—ancient Assyrian reliefs and lamassu, dating to 883–859 BCE—were subjected. The study further explores the role of the gallery’s structural system and mounting details in transmitting vibrations to the art, supplemented by a numerical simulation of the relief dynamics. The present work serves as a useful reference for construction project managers and museum conservation specialists to help gauge the potential hazards caused by various construction means and methods. The study highlights the complex interplay between the art and the supporting structures that are used to stabilize, protect, and display the objects.

  PublisherOA PDFDOI

• Spectrum and Physics-Informed Neural Networks (SaPINNs) for Input–State–Parameter Estimation in Dynamic Systems Subjected to Natural Hazards-Induced Excitation

  Mechanical Systems and Signal Processing · 2025-01-01

  preprintOpen accessSenior author
  PublisherOA PDFDOI

• Analysis of Wind Turbine Wake Meandering Using the Actuator Disc Model

  2025-01-01

  articleOpen accessSenior author
  PublisherOA PDFDOI

• Dynamic penalty-based dispatching decision-making for improved emergency response in urban environments: a heuristic approach

  Frontiers in Future Transportation · 2025-05-20

  articleOpen accessSenior authorCorresponding

  Emergency medical services (EMS) are a crucial component of urban safety and responsiveness, and optimizing their operations aligns with the broader goals of creating safe, resilient cities. This study focuses on improving the EMS dispatching process by leveraging urban mobility data collected by connected vehicles and simulation. EMS dispatching is inherently sequential and dynamic, where each decision impacts future resource availability. Traditional greedy approaches, which dispatch the nearest available unit without considering supply-demand dynamics in the surrounding area, can lead to suboptimal outcomes. This study introduces a penalty metric that quantifies supply-demand levels within each ambulance’s catchment zone—defined by isochrones that delineate the area the ambulance can reach within an allowable time—prior to dispatch. This metric forms the foundation of a dynamic penalty-based dispatching strategy that penalizes dispatches from high-demand, low-coverage areas for low-priority calls, ultimately conserving resources for high-priority emergencies. The heuristic method was tested simulating EMS operations in Manhattan, New York. Simulation results showed that 90% of episodes with the heuristic policy had a mean response time of less than 6 min for high-priority calls, compared to only 75% with the conventional greedy approach. This paper presents a proof-of-concept study that introduces a novel ambulance dispatching policy and contributes to the optimization of emergency response systems in urban environments. Additionally, this study demonstrates how smart technologies and large-scale mobility data can enhance decision-making support tools, improving EMS efficiency and resource utilization and aligning with sustainability goals.

  PublisherOA PDFDOI

• AutoDrive-QA: A Multiple-Choice Benchmark for Vision-Language Evaluation in Urban Autonomous Driving

  ArXiv.org · 2025-03-20

  preprintOpen accessSenior author

  Evaluating vision-language models (VLMs) in urban driving contexts remains challenging, as existing benchmarks rely on open-ended responses that are ambiguous, annotation-intensive, and inconsistent to score. This lack of standardized evaluation slows progress toward safe and reliable AI for urban mobility. We introduce AutoDrive-QA, the first benchmark that systematically converts open-ended driving QA datasets (DriveLM, NuScenes-QA, LingoQA) into structured multiple-choice questions (MCQs) with distractors grounded in five realistic error categories: Driving Domain Misconceptions, Logical Inconsistencies, Misinterpreted Sensor Inputs, Computational Oversights, and Question Ambiguity. This framework enables reproducible and interpretable evaluation of VLMs across perception, prediction, and planning tasks in complex urban scenes. Experiments show that fine-tuning LLaVA-1.5-7B improves accuracy by about six percentage points across tasks, GPT-4V achieves the strongest zero-shot performance with up to 69.8% accuracy, and Qwen2-VL models also perform competitively, particularly in multi-view settings. Moreover, traditional metrics such as BLEU and CIDEr fail to distinguish strong from weak models. By providing an objective, domain-grounded evaluation protocol, AutoDrive-QA contributes to more transparent benchmarking of urban AI systems, supporting the development of safer and more trustworthy autonomous driving technologies for smart cities.

  PublisherOA PDFDOI

• SOD-YOLOv8—Enhancing YOLOv8 for Small Object Detection in Aerial Imagery and Traffic Scenes

  Sensors · 2024-09-25 · 84 citations

  articleOpen accessSenior author

  Object detection, as a crucial aspect of computer vision, plays a vital role in traffic management, emergency response, autonomous vehicles, and smart cities. Despite the significant advancements in object detection, detecting small objects in images captured by high-altitude cameras remains challenging, due to factors such as object size, distance from the camera, varied shapes, and cluttered backgrounds. To address these challenges, we propose small object detection YOLOv8 (SOD-YOLOv8), a novel model specifically designed for scenarios involving numerous small objects. Inspired by efficient generalized feature pyramid networks (GFPNs), we enhance multi-path fusion within YOLOv8 to integrate features across different levels, preserving details from shallower layers and improving small object detection accuracy. Additionally, we introduce a fourth detection layer to effectively utilize high-resolution spatial information. The efficient multi-scale attention module (EMA) in the C2f-EMA module further enhances feature extraction by redistributing weights and prioritizing relevant features. We introduce powerful-IoU (PIoU) as a replacement for CIoU, focusing on moderate quality anchor boxes and adding a penalty based on differences between predicted and ground truth bounding box corners. This approach simplifies calculations, speeds up convergence, and enhances detection accuracy. SOD-YOLOv8 significantly improves small object detection, surpassing widely used models across various metrics, without substantially increasing the computational cost or latency compared to YOLOv8s. Specifically, it increased recall from 40.1% to 43.9%, precision from 51.2% to 53.9%, mAP0.5 from 40.6% to 45.1%, and mAP0.5:0.95 from 24% to 26.6%. Furthermore, experiments conducted in dynamic real-world traffic scenes illustrated SOD-YOLOv8’s significant enhancements across diverse environmental conditions, highlighting its reliability and effective object detection capabilities in challenging scenarios.

  PublisherOA PDFDOI

Recent grants

• Data Fusion of Heterogeneous Sensor Measurements for Enhanced Structural Modeling

  NSF · $287k · 2011–2015

• CAREER: Development of Nonlinear Modeling Tools for Analysis, Simulation, and Structural Health Monitoring

  NSF · $389k · 2002–2009

• High Fidelity Probabilistic Structural Health Monitoring

  NSF · $343k · 2016–2020

• Enhanced Modeling of the Rocking and Overturning of Objects on a Moving Base

  NSF · $313k · 2012–2016

• NSF Engineering Research Center for Smart Streetscapes (CS3)

  NSF · $25.2M · 2022–2027

Frequent coauthors

• Sami F. Masri

  San Jose State University

  93 shared

• A. G. Chassiakos

  California State University, Long Beach

  77 shared

• T. K. Caughey

  71 shared

• Norman Hunter

  26 shared

• Masatoshi Nakamura

  Kanzaki (Japan)

  25 shared

• Jin‐Song Pei

  University of Oklahoma

  24 shared

• George Deodatis

  18 shared

• Guillermo Franco

  18 shared

Labs

• Technology LabPI

Education

• M.A.

  Columbia University Graduate School of Architecture, Planning and Preservation