Pedestrian volume Estimates for Maine Towns

The state of Maine has produced a model of average annual daily traffic (AADT) for vehicular volumes on roadways, which has been used for countless applications, such as safety network screening, project design decision-making, preliminary intersection control evaluations, application of engineering instructions and design guidance, etc. An analogous model for pedestrian volumes remains lacking, making it difficult to make informed location decisions about pedestrian infrastructure and safety investments. Models for pedestrian volume estimates have recently been developed for urban districts (Sevtsuk et al. 2021; Sevtsuk et al. 2024) and even large cities (Alhassan & Sevtsuk, 2024). Caltrans and the UC Berkeley Safe Transportation Research and Education Center have co-produced partial pedestrian volume estimates for selected state highway intersections . Such estimates can enable important insights about the distribution non-motorized road users, suggesting, for instance, where pedestrian infrastructure improvements and investments could impact most constituents. Pedestrian volume estimates can also form a critical denominator for various pedestrian hazard data (e.g. crashes, noise, air pollution). For example, state traffic crash location records can illustrate total crashes involving pedestrians at intersections, which may be used to justify Vision Zero intersection improvements. However, normalizing such crash records with pedestrian volume estimates can illustrate what percent of non-motorized road users experience crashes—suggesting potentially different policy-relevant interventions and helping identify not only locations with most crashes, but also locations with highest crash probabilities. Pedestrian volume estimates could assist project decision-making, help identify locations for RRFB’s and Pedestrian Hybrid Beacons, identify High Priority Active Transportation routes, determine locations for continual pedestrian volume collection, etc. A pedestrian volume model for cities and towns in Maine would enable better decision-making around non-motorized street user needs and help further the state’s goals to support all street users equally, and to decarbonize the transportation sector in socially just ways. The research team proposes a collaboration between Maine Department of Transportation (MaineDOT) and the Massachusetts Institute of Technology (MIT) City Form Lab to develop a first pedestrian volume model for the 140 largest cities and towns in Maine, with populations over 2,500 residents. These towns constitute approximately 77 per cent of the state’s population. The model will estimate AADFT—average annual daily foot-traffic for all for all road and street centerlines in the selected towns. The model will be developed using detailed land use, demographic, transportation and point-of-interest data, which function as address-level origins or destinations for pedestrian movement. The research team will use the open-source Madina Python package (recently developed at the MIT City Form Lab) to estimate the spatial distribution of pedestrian trips during typical weekday and weekend periods between expected land use pairs and calibrate the model estimates on observed pedestrian counts from hundreds of available intersections throughout Maine, where pedestrian flows have been counted using computer vision detection from camera feeds (such as MioVision). Most existing MaineDOT pedestrian counts are from intersection turning movement counts which are represented in a geographic information system (GIS) interface on the Drakewell platform. Additional continuous pedestrian volume data will be available at new traffic signals. . The longitudinal nature of camera counts will enable us to use average weekday and weekend counts (instead of specific day counts) during different seasons, allowing the model to also estimate segment-level pedestrian flows for different seasons. The model will be similar in structure to the ones developed in New York City (Alhassan & Sevtsuk, 2024) and Melbourne Australia (Sevtsuk et al. 2021), but implemented at a state-wide scale for significantly larger lower-density areas for the first time. For one case-study town (TBD), the research team will also develop a detailed pedestrian sidewalk network-- comprising sidewalks, crosswalks, and footpaths using Tile2Net—another open-source Python package for mapping pedestrian infrastructure from aerial imagery tiles, developed at the MIT City Form Lab. Using these data, the research team will estimate a similar pedestrian volume model for these more detailed network segments (instead of road centerlines), providing a higher resolution overview of pedestrian activity, which can substantially differ between opposing sides of the same street segment, or different crossing segments at the same intersection. If this research is successfully implemented on most of the road segments in Maine where pedestrians are expected, there could be major safety benefits. This could help guide decision-making for construction of Complete Streets infrastructure, traffic calming elements, pedestrian hybrid beacons, highway design methods, etc. This could also reduce costs by reducing needs for data collection and reducing “missed opportunities” to construct infrastructure which requires additional programmed work. Infrastructure constructed where people are walking and where generators are present can improve access for residents and improve economic opportunity for the local businesses. It is MaineDOT Research and Innovation’s understanding that this project would provide the state of Maine with the most comprehensive “statewide” pedestrian volume model in the country. The research should result in at least an accessible and sharable GIS dashboard with the pedestrian volume estimates. It would be the primary research objective to get the results implemented as a layer in the MaineDOT Public Map Viewer. MaineDOT GIS would be engaged in this project from the beginning to increase the likelihood of successful data and GIS migration. The research team anticipates being able to accomplish this and they are going through a similar technology transfer process with the city of New York and other entities.

Language

• English

Project

• Status: Active
• Funding: $300,000.00
• Sponsor Organizations:

  Maine Department of Transportation

  16 Statehouse Station
  Augusta, ME  United States  04333-0016
• Managing Organizations:

  Maine Department of Transportation

  16 Statehouse Station
  Augusta, ME  United States  04333-0016
• Project Managers:

  Pulver, Jeffrey

• Performing Organizations:

  Massachusetts Institute of Technology, Cambridge

  77 Massachusetts Avenue
  Cambridge, MA  United States  02139
• Principal Investigators:

  Sevtsuk, Andres

• Start Date: 20250408
• Expected Completion Date: 20270930
• Actual Completion Date: 0

Subject/Index Terms

• TRT Terms: Annual average daily traffic; Geographic information systems; Pedestrian counts; Pedestrian flow; Sidewalks
• Geographic Terms: Maine
• Subject Areas: Highways; Pedestrians and Bicyclists; Public Transportation;

Filing Info

• Accession Number: 01959904
• Record Type: Research project
• Source Agency: Maine Department of Transportation
• Files: RIP, STATEDOT
• Created Date: Jul 2 2025 9:52AM