WPI researcher aims for a better, cheaper commute

WORCESTER — Imagine transportation as it could be — without having to make a choice between cost and travel time, without the 20 or more stops until reaching the destination or the difficulty of finding a local station.

Yanhua Li, assistant professor of computer science and data science at Worcester Polytechnic Institute, is working on it.

The researcher is developing an idea for a new kind of transit system called CityLines that is inspired by airlines and built for urban areas.

Public transportation such as buses, subways and shuttles offer shared transport, which lowers the cost but also increases travel time and complicates transfers due to the needs of each individual passenger.

Private transport such as taxis and ride-sharing services cut travel time with door-to-door service, but increase costs by dealing with one passenger at a time.

“Can we have a transit system that can offer the transport service to the passengers with as low travel costs as the current public transit and the quality of service and travel time as the currant private transit? That’s our goal,” Mr. Li said.

Americans spend increasing amounts of time commuting, and most do so alone. About 76 percent of U.S. residents surveyed by the U.S. Census Bureau in 2013 said they drove alone to work. Only 5 percent used public transit.

The average commute time of about 24 minutes also meant that workers could easily spend more than 100 hours a year just getting to and from work.

CityLines, which is still just a theory, imagines a transportation system based on what airlines use every day, a hub-and-spoke model.

A hub acts as a main airport, or station. Flights, or routes, go out from the hubs.

Like airlines, CityLines' trips would be designed to take fewer detours to guarantee travel time, but also to lower operation costs by aggregating the trip demands of big-city dwellers.

“The uniqueness of this project is the novel idea of adapting hub-and-spoke from the airline industry to the urban area,” Mr. Li said. “Other civil engineering folks and computer science folks, when they look at a transportation problem, they either look at one transit mode – pubic or private – to try and develop a better bus law, a timetable, or taxi patterns. None of their work talks about how to fundamentally borrow another transit mode.”

The National Science Foundation has awarded Mr. Li a two-year grant, which he and his research team are using to compile, manage and analyze a large quantity of data for simulation testing. After that, they’ll be able to build the ideal system model for CityLines. The project's website is at https://users.wpi.edu/~yli15/CityLines/.

With PhD students Guojun Wu, Menghai Pan and Guanxiong Liu, Mr. Li has been working on this project since 2015.

“We collected large amounts of public travel data from Shenzehn, China, and public U.S. data in Boston, Chicago, and Minneapolis,” Mr. Li said. “State records and people moving around their city to commute or for other purposes by taking taxis, buses, and subways. From that targeted data, we apply data mining and emotional learning techniques to predict future trip demands in a city from a small region to other regions. By knowing accurate future trip demands, we know where people want to start from."

From WPI’s Urban Computing Lab, the team uses data extracted from records of public and private transit use, the urban infrastructure network, public transit timetables and human trip-demand data to understand which places in cities would be best for hub stations and which to avoid.

CityLines would not replace current transit systems but would function alongside them and make use of the other’s routes and stations. There are plans to collaborate with the University of Massachusetts Transportation Center at UMass at Amherst to tackle the civil engineering aspects of the project.

Mr. Li said that to take CityLines into a real-world situation, he would need input from people from other disciplines, such as civil engineering and government.

“The perfect scenario?” he said. “Maybe in two years we could have a small scale test on a campus, which may run for a year or two before we really deploy it to a larger scale.”