Balaji Prabhakar, a Stanford University computer scientist, was stuck in a massive traffic jam in Bangalore, when he began to think about the congestion plaguing rapidly growing cities across the world. The experience sparked an experiment he conducted with the Indian company Infosys and 20,000 of its employees in the city. Could you cut down on congestion if you gave people incentives to drive to and from work at “off-peak” hours?
It worked. By offering prizes on a lottery system, within six months in 2008, 17% of traffic had shifted from the peak travel times to slightly before or after, easing congestion and spurring a similarly successful project he conducted right on Stanford University’s campus, where he also directs the Center for Societal Networks
“Fifty percent of us humans live in cities already, and there’s going to be more than 5 billion of us by 2030,” says Prabhakar. “They’re bumping into each other more and more, and so the question is how do we get them to avoid each other, on buses, trains, and on the road?”
His answer to that question is Urban Engines, a software company he co-founded to help city transit agencies and planners around the world use their data to understand where the choke points on their rail, bus, and road networks are–and, importantly, give them the tools to ease them through rewards and prizes.
To take on this difficult challenge, Urban Engines, which has been operating quietly for two years until this week, leans on the computer science and modeling expertise of Prabhakar and his co-founder and CEO, former Google engineering vice president Shiva Shivakumar. Usually, it’s very difficult for cities to understand where the congestion “hotspots” of people are on a train or bus, for example. They either rely on inexact surveys or have to install expensive equipment systems, such as sensors, video cameras, or even big weighing scales inside trains.
What Urban Engines does is more elegant, a combination of “crowd sensing” and big data analysis and modeling, using existing data that cities already have–commuter card swipes. Its software creates a digital replica of the subway or bus system, and can estimate statistics about where the people in it are located in near real time, such as how long individuals waited for a bus or train; how many people were on them; and where systems are getting overloaded. The screen, in a demo the pair demonstrated, looks like a fun game of people moving in and out of cars and trains chugging along.
The secret to how they do it lies in being able to access the anonymized swipe data from commuter cards, which they receive from a city they are working with and store on their secure cloud servers. “Any one person’s ‘swipe in, swipe out data’ tells us, say, what normally should take 40 minutes, took 60 minutes, so you were stuck in the system somewhere. But where? And how long? For this we need all the other people’s data,” says Prabhakar. “If you have other people’s data, they shared some portions of the trip with you, and you can tell where people got stuck. It really is no one person’s data has got all the information, it’s all the data put together.”
Already, Urban Engines is working with three cities. In Washington, D.C., they’re helping the transit agency reconstruct 700,000 average daily trips to improve the agency’s planning. In Sao Paulo, Brazil, they’re working with the World Bank to help the city model a 15,000-bus system and provide an initial commuter engagement program for bus riders.
But it is Urban Engines’ work in Singapore that illustrates its full potential. Like in Prabhakar’s initial experiment in Bangalore, there, they are working with the transit authority to use its data to create a “travel rewards” program that eases peak congestion.
There, commuters sign-up by registering their smart cards, earning points for every trip–with one point per kilometer traveled. For each off-peak trip, however, they earn three points. The points can be redeemed for future trips on their card, or they can use the points to play a “micro-raffle game” in an app, getting the chance to win larger dollar amounts. So far, the company says there’s been a 7% to 13% reduction in peak hour traveled with 200,000 commuters participating.
Carlos Daganzo, an University of California, Berkeley, engineer who studies transportation networks and isn’t involved in the company, says Urban Engines’ data-enabled incentives to reduce congestion are compelling. Many people in the transportation world are just starting to think about how to think about how to provide “carrots,” not sticks, so people change their commuting behavior for the good of all. “If you can analyze the data day-by-day very cheaply, than you can design better policy incentives to direct traffic,” Daganzo says.
Now the company hopes to expand to other cities, announcing this week investors including Google Ventures, Andreessen-Horowitz, Samsung Ventures, and Google executive chairman Eric Schmidt. Shivakumar, Urban Engines’ CEO, says they plan to take advantage of the fact that more and more cities globally have smart card technology. “Many cities are already there. It’s just that data hadn’t been thought of being seriously crucial or relevant before,” he says.