Join us to participate in the upcoming 2019 City:One Challenge. 

UnPlanIt: Software allowing transit planners to design networks that make unplanned trips easy

Developing the software tools for designing transit routes that are competitive with car ownership and ride sharing services

Photo of Matthew Laquidara
5 0

Written by

Many proposals for extending mobility focus on creating  technologies for more flexible operation of transportation. This overlooks the transformative benefits of something more fundamental: improving the design and planning of public transit networks. Rather than introducing potentially expensive and unpredictable new operational technology, this solution improves mobility by introducing UnPlanIt, a guided and automatic route planning software that will give transit planners the ability to reimagine Pittsburgh's public transportation network in a way that will provide the flexibility that people demand.


People make journeys to fulfill a broad variety of needs. These needs may arise suddenly: an individual may need to immediately go to an urgent care clinic at any time of day or night. The origins and destinations associated with these trips may be difficult for those planning transit networks to anticipate: a family in one residential neighborhood may need to travel to another to drop off children at a babysitter. For a person who owns a car, or can afford to regularly use ride hailing services, these spontaneous trips are not burdensome. For those who do not use these means of transportation, whether due to financial constraint or a desire to reduce the congestion, accidents, and pollution inherent to them, making these trips on public transit can be quite difficult.


Many assume that fixed-route transit service cannot enable this model of transit use. This assumption fails to recognize that transit networks have merely not been measured, and thus planned, on this basis. This has had deleterious consequences; many transit agencies have seen losses in ridership since ride hailing apps have become pervasive. To fix this problem, planners must be given access to software tools that allow them to redesign their fixed-route transit system to meet the spontaneous needs of their present and future customers. Within the six month challenge period, UnPlanIt will be used to generate a cost-neutral plan to restructure the bus routes servicing Pittsburgh that are operated by the Port Authority of Allegheny County. It will achieve this by evaluating modifications made by both human planners and artificial intelligence systems, resulting in a transit network that better meets the needs and values of Pittsburgh's residents.


Typically, a transit agency's planners make changes to transit routes periodically. Often, historic, aggregated ridership data is used to justify these decisions. Routes that are overcrowded are candidates for increasing frequency; those rarely ridden may be eliminated. New service may be introduced if there is speculation that a route between two endpoints would be well-used.


Because this evaluation method focuses so heavily on areas of high demand and times of heavy commuting, it largely neglects the plight of an individual taking an unplanned, unanticipated trip. The popularity of car ownership and ride hailing services, however, indicates that people demand the ability to behave spontaneously. Thus, a transit network that will regain riders lost to these methods must be first measured and then redesigned to maximize its capability for enabling spontaneity. The methodology behind this proposal focuses on measuring precisely that.


This capability is calculated with a measurement called Spontaneous Accessibility. In this measurement, the service area of a transit agency is divided into a tightly-spaced grid of sectors. For every minute of a day, the software determines which sectors can be reached within a given time, using each of the same sectors as starting points. It does this using real transit schedules and accurate walking routes, entirely derived from public data. For each point, this results in a probability that it can be reached given the random selection of a starting point and time. The overall quality of the network is then measured by aggregating the probabilities into the final Spontaneous Accessibility score. Changes can be evaluated by instructing the program to modify the transit network, re-running the analysis, and observing the changes to the probabilities, both at the full network level and at every sector of the grid. In that way, planners and artificial intelligence systems can iteratively make alterations to transit routes and rapidly understand the Spontaneous Accessibility implications.


Spontaneous Accessibility is already been recognized as a promising new measurement for public transit planning. My paper introducing this methodology was published in the 2018 Transportation Research Record by the Transportation Research Board of the National Academies of Sciences, Engineering and Medicine, and presented at the organization's annual meeting. It was awarded the Committee on Transit Capacity and Quality of Service's best paper award as well as the Executive Committee's Fred Burggraf Award for excellence in transportation research by those 35 or younger. The work in the paper was supported by a software tool that generates Spontaneous Accessibility scores. Subsequent work has made considerable improvements to the precision and speed of the software.


Being selected by The City of Tomorrow Challenge would enable a considerable expansion of this research idea. It would support the full-time development work that would transform the current command-line research tool into a production-ready system. UnPlanIt would guide urban planners in the City of Pittsburgh through the restructure process by providing a web-based interface that they could use to propose and evaluate the Spontaneous Accessibility impact of transit network modifications.  Its automation component would use artificial intelligence to consider changes to transit routes, freeing planners from having to envision those changes. This would allow them more time to commit to maintaining awareness of the community's needs, not to mention the many challenges of operating a transit network outside of route planning. Continued development would also allow the melding of Spontaneous Accessibility measurements with demographic, civic, and commercial data. This would allow Spontaneous Accessibility improvements to be prioritized, based on areas that are significant to the community's needs.


More importantly, selection would make a connection between the motivated planners in the City of Pittsburgh and a transit planning technology ready to emerge from the research sphere. With continuous development and agile practices, planners could begin using the software while improvements are regularly introduced. Within the six month period, it would be possible to create a proposal that improves bus service in Pittsburgh and present it to the Port Authority for implementation. This would provide a powerful case study for proving the practicality of Spontaneous Accessibility evaluation as a transit planning strategy. A successful trial in Pittsburgh could mean the application of Spontaneous Accessibility measurement at transit agencies throughout the world. As a result of this challenge, a pay-per-use or subscription-based web service that planners could use to compute it would be immediately available. Not only would Pittsburgh realize the tremendous benefit of extending better mobility options to its residents and visitors, but out of it would emerge a compelling product capable of improving mobility throughout the world.

Describe who will use your solution (1,000 characters)

UnPlanIt would be used by urban planners in the City of Pittsburgh. Public transit service within Pittsburgh is provided at the county level, but successful mobility within the city depends on it. City planners can use UnPlanIt to author feasible proposals that are presented to the Port Authority. It would allow them to consider modifications to the network, and have them quickly evaluated without large time or data collection burdens. The automation component would allow proposed changes to be generated without the guidance of planners, allowing them more time to work on the many other parts of their job. Once implemented, all people in the city benefit from having access to a transit network that supports their unplanned, unanticipated trips.

Describe your solution's stage of development

  • Prototype - you have built a prototype and tested it with potential users

Tell us about your team or organization (500 characters)

I am a long-time software developer and, more recently, an author of highly-regarded public transit planning research. I am looking to partner with transit agencies to bring my planning methodology, Spontaneous Accessibility, into use. While attempting to do this as a consultant, I realized that a better way would be to build a software tool that planners can use to evaluate their network's Spontaneous Accessibility and be guided through how best to improve it. https://publictransitanalytics.com

Size of your team or organization

  • I am submitting as an individual

Funding Request

  • $100,000

Describe how you would pilot your idea (1000 characters)

The challenge funding would first be used to take three months to develop a minimum viable product version of UnPlanIt. I would introduce this software to the planning staff of the City of Pittsburgh and solicit feedback. I would ask them what features would be needed for them to use it to present a proposal for restructured bus service within Pittsburgh to the Port Authority of Allegheny County. Using that feedback, I would prioritize these features while also considering how AI techniques could automate common usage patterns. Government agencies have been willing to invest in software products. A 16 month license to the Remix transit planning software cost the Puget Sound Regional Council $114,000. There are 2323 transit agencies in the US that have the ability to apply for federal grants supporting capital expenditures like planning software. With Pittsburgh serving as a case study, many agencies would have a compelling reason to purchase a subscription to this software.

Describe how you would measure the success of your pilot (1000 characters)

The most important data point will be whether a bus restructure proposal comes to fruition within 6 months. After that, the feedback provided by the Port Authority will provide valuable qualitative data on the feasibility of the proposal. While planners use UnPlanIt, it would be useful to collect measurements on their engagement. This would include how many proposed modifications that they submit, the amount of time they spend making modifications, and the number of changes that they make per proposal. These may be proxies for their level of satisfaction. As the software will be a hosted, web-based service, it will also be imperative to measure the cost of computational resources so that future deployments can be priced appropriately.

Attachments (1)

10-01522 Laquidara Measuring Spontaneous Accessibility.pdf

Research paper describing Spontaneous Accessibility measurement.

5 comments

Join the conversation:

Comment
Photo of Diana Avart
Team

Hi Matthew,

Thanks for submitting your idea! Does the software work in a way that completely overhauls the bus system or just suggests the small tweaks in a more efficient way than the current process? I am also curious if the software would be customized to Pittsburgh or is it meant to be able to be used by other locations in the future without a similar process?

- Diana, Facilitator

Photo of Matthew Laquidara
Team

Hi Diana,
Thanks for your comment! The software can work in either way. Because of the fine granularity of the measurement (usually 80 meters), it can assess something as small as the movement of a single bus stop. It could certainly be used for larger overhauls as well. When assessing a change, it's possible to look at the Spontaneous Accessibility change versus the change in-service vehicle hours (and the agency's cost per in-service hour). That way changes of different magnitudes can be evaluated against each other for their relative efficacy.
So far, I don't have any Pittsburgh-specific customizations in mind, but I would expect to learn what those would need to be while collaborating with the city staff. One of the strengths of UnPlanIt is that it needs very little data to evaluate a network, just GTFS schedule data and OpenStreetMap road and path data. I've, so far run, analyses for Seattle, Boston, and Maui.

Photo of Diana Avart
Team

Thanks for the clarification and further detail!

View all comments