Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

FT-AED: Benchmark Dataset for Early Freeway Traffic Anomalous Event Detection

Austin Coursey, Junyi Ji, Marcos Quinones-Grueiro, William Barbour, Yuhang Zhang, Tyler Derr, Gautam Biswas, Daniel B. Work

TL;DR

This paper introduces the first large-scale lane-level freeway traffic dataset for anomaly detection and finds that unsupervised graph neural network autoencoders are a promising solution for this problem and that ignoring spatial relationships leads to decreased performance.

Abstract

Early and accurate detection of anomalous events on the freeway, such as accidents, can improve emergency response and clearance. However, existing delays and errors in event identification and reporting make it a difficult problem to solve. Current large-scale freeway traffic datasets are not designed for anomaly detection and ignore these challenges. In this paper, we introduce the first large-scale lane-level freeway traffic dataset for anomaly detection. Our dataset consists of a month of weekday radar detection sensor data collected in 4 lanes along an 18-mile stretch of Interstate 24 heading toward Nashville, TN, comprising over 3.7 million sensor measurements. We also collect official crash reports from the Nashville Traffic Management Center and manually label all other potential anomalies in the dataset. To show the potential for our dataset to be used in future machine learning and traffic research, we benchmark numerous deep learning anomaly detection models on our dataset. We find that unsupervised graph neural network autoencoders are a promising solution for this problem and that ignoring spatial relationships leads to decreased performance. We demonstrate that our methods can reduce reporting delays by over 10 minutes on average while detecting 75% of crashes. Our dataset and all preprocessing code needed to get started are publicly released at https://vu.edu/ft-aed/ to facilitate future research.

FT-AED: Benchmark Dataset for Early Freeway Traffic Anomalous Event Detection

TL;DR

This paper introduces the first large-scale lane-level freeway traffic dataset for anomaly detection and finds that unsupervised graph neural network autoencoders are a promising solution for this problem and that ignoring spatial relationships leads to decreased performance.

Abstract

Early and accurate detection of anomalous events on the freeway, such as accidents, can improve emergency response and clearance. However, existing delays and errors in event identification and reporting make it a difficult problem to solve. Current large-scale freeway traffic datasets are not designed for anomaly detection and ignore these challenges. In this paper, we introduce the first large-scale lane-level freeway traffic dataset for anomaly detection. Our dataset consists of a month of weekday radar detection sensor data collected in 4 lanes along an 18-mile stretch of Interstate 24 heading toward Nashville, TN, comprising over 3.7 million sensor measurements. We also collect official crash reports from the Nashville Traffic Management Center and manually label all other potential anomalies in the dataset. To show the potential for our dataset to be used in future machine learning and traffic research, we benchmark numerous deep learning anomaly detection models on our dataset. We find that unsupervised graph neural network autoencoders are a promising solution for this problem and that ignoring spatial relationships leads to decreased performance. We demonstrate that our methods can reduce reporting delays by over 10 minutes on average while detecting 75% of crashes. Our dataset and all preprocessing code needed to get started are publicly released at https://vu.edu/ft-aed/ to facilitate future research.
Paper Structure (30 sections, 8 equations, 13 figures, 5 tables)

This paper contains 30 sections, 8 equations, 13 figures, 5 tables.

Table of Contents

  1. Introduction
  2. Background
  3. Existing Datasets
  4. Problem Formulation
  5. FT-AED Dataset
  6. Data Collection
  7. Radar Detection Systems
  8. Incident Labels
  9. Data Processing
  10. Dataset Usage
  11. Experiments
  12. Additional Preprocessing
  13. Model Training
  14. Anomaly Detection Results
  15. Case Study
  16. ...and 15 more sections

Figures (13)

  • Figure 1: Map of the Radar Detection Systems (RDS) sensor network deployed for data collection. 49 RDS sensors are deployed along Interstate 24 toward Nashville, Tennessee. Each sensor captures speed, occupancy, and volume data for each of the four lanes every 30 seconds.
  • Figure 2: Freeway traffic dynamics: October, 2023 weekday lane 1 (high-occupancy vehicle (HOV) lane, often refers to the leftmost lane) speed data visualization for the Westbound (the direction to downtown Nashville) of I-24 section from road reference marker mile 71 to 53, morning peak hours from 4AM and 12PM, sensors are deployed with about a 0.3 to 0.4 mile interval on freeway.
  • Figure 3: Relational spatiotemporal graph design, illustrated over an example freeway. The full graph is made by forming these connections for all nodes across the time horizon.
  • Figure 4: Case study. Crash detection and reporting for the morning of October 11, 2023, using STG-GAT AE. The threshold was chosen such that there was a $10\%$ validation FPR.
  • Figure 5: Lane-level data comparison: the left is Monday, a normal workday, the right is Thursday, a very clear crash can be seen by the triangle shape present around 9:30 AM in the right figures.
  • ...and 8 more figures