February 12, 2019
As cities continue to grow, the role of public transportation shows no sign of lessening. Whether it’s moving people within cities on buses or subways, or between cities via trains, engineers and technologists are looking for ways to make transit infrastructure more efficient for everyone.
It’s worth noting off the bat that a key benefit of public transit is its efficiency. In an opinion piece in January’s issue of IEEE Spectrum, author Vaclav Smil breaks down the energy intensity of different travel options. “When I’m the only passenger in my Civic,” he says, “it requires about 2 megajoules per passenger-kilometer in city driving. Add another passenger and that figure drops to 1 MJ/pkm, comparable to a half-empty bus.” Smil continues: “Of course, public-transit trains are far superior: At high passenger loads, the best subways need less than 0.1 MJ/pkm.”
The numbers make it clear that mass transit options have less environmental impact per passenger than a car. However, there’s still a lot of room to improve efficiency.
For buses, researchers in China and the U.K. have been using machine learning to help drivers reduce fuel consumption. Since driving styles can have a huge impact on efficiency (up to 30% in an urban environment, according to the journal article), providing the driver with real-time assistance can encourage them to optimize for fuel efficiency.
Displaying prompts for shifting gears, changing the gradient of the accelerator pedal and changing the pedal’s average depth can help the driver make positive changes. And machine learning can be used to display the option closest to the operator’s driving style, increasing the likelihood that they adhere to the suggestions. In total, the researchers say, “the LPP algorithm can guide the driver to increase the fuel economy by 6.25% under the premise that the driving task is basically unchanged.”
When it comes to subways, safety is consistently one of the biggest concerns. While subway systems have been incorporating shallow neural networks to reduce accidents, there’s significant potential to use deep learning to provide superior results. In a paper submitted to the 2017 IEEE/ACIS International Conference on Computer and Information Science (ICIS), researchers introduced user satisfaction and rare-event probability into their safety prediction model, greatly increasing the accuracy.
High-speed rail projects capture the public imagination, but can be time intensive to roll out. According to IEEE Spectrum, this year, East Japan Railways plans to test a new generation of bullet trains, the Alfa-X, capable of topping 400 km/h. While the company may limit the operating speed to 360 km/h, that’s more than 10% faster than today’s trains. Meanwhile, they say, “Morocco is finally on track to put its first high-speed train into operation by early 2019, an effort that has been plagued by delays. Tests in 2018 pushed that train’s speed to 357 km/h, a record for Africa.”
With such extreme speeds, there’s the question of how to reliably relay wireless data to these machines. An article published in IEEE Access explains how high-speed railway communication (HSRC), which “delivers information not only for special applications such as train scheduling control and safety monitoring but also for public applications such as internet services for passengers,” can be improved through different methods of power allocation.
If you want to learn more or get involved, we have both the IEEE Intelligent Transportation Systems Society (ITSS) and the IEEE Transportation Electrification Community (TEC) focused on making transportation more efficient. The ITSS works on transportation systems of all kinds, while the TEC deals with the electrification of cars, ships, aircraft and more.