Determining Typology of Signalized Intersections Based on Pedestrian Actuation Through Use of Traffic Signal Controller Data

Class

Article

College

College of Engineering

Department

Civil and Environmental Engineering Department

Presentation Type

Oral Presentation

Abstract

Pedestrian volume data is essential in traffic planning, determining critical pedestrian locations, improving pedestrian facilities, and a vital use is to feed pedestrian volume models. Pedestrian counts, either manually or through the use of technologies like RADAR, and video image processing, are the existing ways to determine pedestrian activity at intersections. Owing to high cost and labor intensive methods for pedestrian counts, there is a need for alternative methods to determine pedestrian activity. Data from traffic controller logs, which contains information about pedestrian actuations, phasing, and cycle length can be a useful data source. Utah Department of Transportation (UDOT) uses data from traffic signal logs in their innovative data analysis and visualization platform called Automated Traffic Signal Performance Measures (ATSPM). UDOT collects data from over 2,000 intersections in the state of Utah and then analyses and calculates performance measures to produce the visual data in web server of UDOT ATSPM webpage. Real time pedestrian activity can be measured as pedestrian actuations (from pedestrian push-buttons) for each hour in a day. Other research involving pedestrian activity at signalized intersections tackles the issues of vehicle-pedestrian collisions, pedestrian safety, models of pedestrian activity, and the effects of other factors on pedestrian activity. This research emphasizes a broader view by preparing a categorization/classification of signalized intersections based on pedestrian activity. Time-series clustering is applied to ATSPM data to group similar intersections using conventional k-means and k-mediods algorithms. Clusters are then analyzed according to land use, residential density, and built-in environment factors. Findings of this study can provide a novel approach in exploring pedestrian activity.

Location

Room 155

Start Date

4-10-2019 10:30 AM

End Date

4-10-2019 11:45 AM

This document is currently not available here.

Share

COinS
 
Apr 10th, 10:30 AM Apr 10th, 11:45 AM

Determining Typology of Signalized Intersections Based on Pedestrian Actuation Through Use of Traffic Signal Controller Data

Room 155

Pedestrian volume data is essential in traffic planning, determining critical pedestrian locations, improving pedestrian facilities, and a vital use is to feed pedestrian volume models. Pedestrian counts, either manually or through the use of technologies like RADAR, and video image processing, are the existing ways to determine pedestrian activity at intersections. Owing to high cost and labor intensive methods for pedestrian counts, there is a need for alternative methods to determine pedestrian activity. Data from traffic controller logs, which contains information about pedestrian actuations, phasing, and cycle length can be a useful data source. Utah Department of Transportation (UDOT) uses data from traffic signal logs in their innovative data analysis and visualization platform called Automated Traffic Signal Performance Measures (ATSPM). UDOT collects data from over 2,000 intersections in the state of Utah and then analyses and calculates performance measures to produce the visual data in web server of UDOT ATSPM webpage. Real time pedestrian activity can be measured as pedestrian actuations (from pedestrian push-buttons) for each hour in a day. Other research involving pedestrian activity at signalized intersections tackles the issues of vehicle-pedestrian collisions, pedestrian safety, models of pedestrian activity, and the effects of other factors on pedestrian activity. This research emphasizes a broader view by preparing a categorization/classification of signalized intersections based on pedestrian activity. Time-series clustering is applied to ATSPM data to group similar intersections using conventional k-means and k-mediods algorithms. Clusters are then analyzed according to land use, residential density, and built-in environment factors. Findings of this study can provide a novel approach in exploring pedestrian activity.