Anyone who has ever hopped on a Los Angeles-area freeway between 3 p.m. and 6 p.m. knows too well what gridlock feels like. Los Angelenos may soon be able to find some solace soon, thanks to a pilot program between Xerox and the Los Angeles County Metropolitan Transportation Authority that uses big data to keep traffic moving for drivers on the I-10 and I-110 freeways who are willing to pay
. That program, called ExpressLanes, is just one of many irons Xerox (via its Affiliated Computer Services subsidiary) has in the fire as it tries to use its considerable technology portfolio to understand and improve traffic on U.S. roadways.
Central to most of Xerox’s anti-congestion projects, including ExpressLanes,is the idea of dynamic pricing, which rises with demand in order to maintain some semblance of order. As Natesh Manikoth, Xerox’s chief technology officer for transportation solutions, explained to me, if a driver is paying to drive in the HOT (high-occupancy tolling) lane, he’s guaranteed a consistent speed of 45 miles per hour. If traffic starts backing up, prices for individual cars will rise to discourage them from entering, saving the lanes (which, before this program were high-occupancy-vehicle lanes) for high-occupany vehicles such as buses and those involved in carpools.
Xerox has another program in Los Angeles called ExpressPark, the goal of which is to let people know when they’re about to leave the house whether and where they might find parking, and how much it will cost. “It’s not enough to know how to set the price, you have to make sure that data gets to users in real time,” Manikoth said. Drivers also need to know parking spots will still be there when they arrive in 40 minutes. That’s a prediction problem.
The answers lie in big data, difficult data
The key to all of this, of course, is lots of data. ExpressLanes, Manikoth explained, works by sensing traffic flows in the HOT lanes as well as in the adjacent lanes and calculating travel times. Because a pre-defined algorithm won’t work, the model is designed to learn as it takes in more data about how any given set of conditions affect traffic flow. Xerox is just getting started with developing its model, Manikosh said, and he aknowledges it won’t be easy.
Traffic accidents, broken down cars and other unforeseen incidents can quickly make a mess of even the best models, especially because no one can predict how long an accident will take to clean up or how many lanes it will close down. And Los Angeles is a particularly unique beast among large cities because it lacks a strong city center, so traffic is relatively constant and in all directions. However, he said, “We stepped up, we’ll have to now prove it.”
View Larger Map
While the parking project depends heavily on prediction, those predictions rely heavily on history. Manikoth said his team can build a machine learning system that looks at historical data as it relates to the current price of parking to predict whether spots will remain available. But that’s easier said than done — or at least, done right — because parking behavior is affected by myriad hard-to-predict factors such as how long someone will be at a meter and how many drivers are willing to park illegally. The meters’ sensors and payment systems can track occupancy rates and what people pay, but not how individual people will act in any given situation.
Solve traffic, solve a lot more
Xerox isn’t alone in trying to help bring order to the chaos that is big-city driving, though. Companies such as IBM and Siemens are also working on the problem — and it’s all part of a larger effort to minimize the problems that cities — the economic engines of our society — experience as they grow. Drivers circling blocks looking for parking spots and commuters stuck in freeway gridlock contribute to pollution and generally lower the quality of life for everyone involved.
Manikoth said the real answer lies in combining information from other sources, such as mass-transit systems, toll highways, traffic sensors and weather data (all of which Xerox also collects) to paint a real-time picture of what traffic actually looks like. Armed with this type of information, city planners might be able to devise more-intelligent stoplights, bus routes and train schedules — maybe even dynamically — and commuters might be able to decide they’re better off just taking the subway today. By collecting diagnostic data from buses, he said, transportation authorities could spot potential issues that might otherwise result in a future breakdown that messes with schedules and people’s lives.
Of course, given how big and expansive a problem traffic management is, there’s monetary incentive for anyone who’s actually able to solve it. “We firmly believe that solving problems for cities is a good thing for society as a whole,” Manikoth said “but it’s also good business.”
Feature image courtesy of Shutterstock user Aaron Kohr.