Mobile-device data, non-motorized traffic monitoring, and estimation of annual average daily bicyclist and pedestrian flows

Principal Investigator(s):

Raphael Stern, Assistant Professor, Civil, Environmental and Geo-Engineering

Co-Investigators:

  • Michael Levin, Assistant Professor, Civil, Environmental and Geo-Engineering
  • Greg Lindsey, Professor, Humphrey School of Public Affairs

Project summary:

Understanding pedestrian and bicyclist flows is vital to distributing a limited construction budget to new infrastructure for improved safety on specific roads. Unfortunately, statewide data collection for non-motorized flows is challenging. MnDOT and Minnesota counties historically have lacked estimates of bicycle and pedestrian traffic on trunk highways (TH) and county state aid highways (CSAH). Since about 2016, MnDOT has begun monitoring bicycle and pedestrian flow at more than 25 locations across the state, but--given the small number of counters and the variability of flows in response to variations in weather across Minnesota--these monitoring data are insufficient for estimation of annual average daily bicyclists (AADB) and annual average daily pedestrians (AADP). One option for obtaining travel data without expensive infrastructure is relying on mobile data collection. However, the accuracy of mobile data is unknown and may vary in different areas (e.g. urban vs. rural). This project will build trip distribution and network route choice models to predict road flows of pedestrians and bicyclists, and calibrate them to mobile data (e.g. StreetLight). A filtering approach that combines model predictions with erroneous data will be used to integrate mobile data (which captures only a fraction of trips) with the model of user trips and route choices (which does not perfectly capture user behavior). The ultimate deliverable of this project is a software tool that MnDOT can use to update pedestrian and bicyclist flows periodically (e.g. annually) as new data about trip frequency and mobile data becomes available. This project will provide a map-based visualization of current predicted non-motorist flows, as well as instructions to use the software tool for later updates.

Project details: