Optimal Routing for Parking Enforcement Vehicles
The City of Raleigh has begun using trucks outfitted with License Plate Recognition (LPR) equipment to streamline parking enforcement. Finding optimal routes for the trucks to evenly and efficiently traverse three sections of city streets that have curbside parking has been a challenge, so the city partnered with an NC State Industrial Engineering class to work on solutions as a Senior class project. When IE student Matt McMillan started looking at software options for solving vehicle routing problems, he contacted the Libraries' Data and Visualization Services for help.
Overview
Esri's ArcGIS Network Analyst software provides a robust set of routing tools for finding optimal vehicle routes, given any number of settings and constraints that are entered into the solver configuration. The "Find Route" tool depends on having street data inputs that are specially designed for network analysis and include one-way designations, turn restrictions, speed limits (for time optimization), and lane counts. However, the basic "street centerlines" data provided by the City for the students' project was not network-enabled. So the students' first hurdle was finding usable input data, or else they would need to do a considerable amount of fieldwork to create their own.
The second hurdle was that they needed to learn how to use the ArcGIS software and its Network Analyst extension. The configuration options for Network Analyst are fairly complex, and knowing how to achieve optimal results takes experience and a lot of trial and error.
Additionally, the requirements for the License Plate Recognition routing are more unique than most routing scenarios. The License Plate Recognition trucks need to go down each street with curbside parking and be in a lane adjacent to the parking spaces. This means that one-way streets with parking on both sides need to be traveled twice in the same direction—a scenario that isn't optioned in Network Analyst.
Other challenges arose as well, such as the need to greatly reduce the number of turns and u-turns that the model was initially prescribing, and ways to output clear directions and route animations to better visualize and comprehend the final routes.
How We Did It
Fortunately, Esri provides a national streets dataset for use with their Business Analyst Desktop product that does contain network parameters. GIS and Data Librarian Jeff Essic extracted a subset of this database for the areas within Raleigh needed by the student team.
For help with understanding how to correctly configure the Network Analyst route solver, Jeff suggested turning to staff at NC State's Institute for Transportation Research and Education (ITRE). Jeremy Scott and Chase Nicholas with ITRE were able to quickly meet and advise Matt on ways to modify the input data and model settings in order to achieve correct results. Jeff then helped Matt with the needed data modifications using editing tools in ArcGIS Pro, and we were able to force travel down one-way streets twice by creating duplicate parallel street segments.
Our Data Science Consultant, Ph.D. graduate student Vishnu Mahesh Vivek Nanda, was instrumental in helping Matt create travel animations of the final routing plans, so City staff could more easily see how the enforcement trucks should navigate through the streets.
