Ride-sharing services like Uber and Lyft have been in the news a lot lately, but mostly because of their run ins with authorities and regulators. Few have probably given deep thought, not to mention devout an entire scientific paper, on their impact at large, for good or for ill. But that is exactly what researchers at the MIT Computer Science and Artificial Intelligence Laboratory (CSAIL) accomplished under the charge of Professor Daniela Rus. They developed an algorithm that showed that carpooling service can potentially reduce road congestion up to three times and, in the long run, even save our environment.
They used New York City as the target of their experiment, both famous and notorious for its cabs. NYC has a fleet of 14,000 taxis, contributing to both traffic jams and pollution. According to the algorithms, 95% of the demand for taxis can be adequately met by a fleet of 2,000 ten-person vehicles. But even more impressive, 98% of that demand can also be met by 3,000 four-passenger cars, exactly the type used by Uber and Lyft.
It’s not a simple case of replacing taxis with Uber and Lyft cars, however. The algorithm actually computes the most efficient routes for cars and even re-routes cars that are idle into higher traffic areas. It’s like one big brain telling drivers where to go. And it also has to take into account how passengers need to be on the same route as others in order to be completely efficient.
Current ride-sharing and carpooling systems unfortunately don’t work that efficiently yet. Getting a proper carpooled ride sometimes involves a bit of planning ahead of time and incurs some waiting time. The algorithm works best with fully autonomous cars, an argument that goes in favor of Uber’s and Lyft’s self-driving fleet plans.
SOURCE: MIT CSAIL