AI: NYC Can Operate Smoothly with 75% Fewer Taxis
New York City has nearly 14,000 taxi medallion licenses, but MIT CSAIL researchers say 3,000 four-person taxis could serve 98 percent of demand in NYC.
Self-driving cars aren’t the most immediate threat to cab drivers. Optimized ride-sharing is.
New York City could function smoothly with just one-quarter the number of cabs if everybody used the carpooling option on ride-share services, according to a new algorithm developed at MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL).
New York City has nearly 14,000 taxi medallion licenses, but CSAIL researchers say 3,000 four-person taxis could serve 98 percent of demand in NYC, and the average waiting time would be only 2.7 minutes. The algorithm also found that 3,000 two-person cars could serve 94 percent of demand and only 2,000 ten-person vehicles could serve 95 percent of demand.
Using data from 3 million NYC taxi rides, the algorithm works in real-time to reroute cars based on incoming requests, and can also proactively send idle cars to areas with high demand, which CSAIL made service 20 percent faster.
Brad Templeton: The Economics of Self-Driving Taxis
“To our knowledge, this is the first time that scientists have been able to experimentally quantify the trade-off between fleet size, capacity, waiting time, travel delay, and operational costs for a range of vehicles, from taxis to vans and shuttles,” says MIT CSAIL Professor and project leader Daniela Rus. “What’s more, the system is particularly suited to autonomous cars, since it can continuously reroute vehicles based on real-time requests.”
CSAIL says services like Uber and Lyft have only been optimizing carpool routes for the last two years, and the existing approaches are still very inefficient. For example, some ride-sharing systems require that user B be on the way for user A, and need to have all the requests in before they can actually create a route.
The CSAIL algorithm allows requests to be re-matched to different vehicles. It can also analyze a range of different types of vehicles, determining where or when a 10-person van would be of the greatest benefit. Rus calls the approach an “anytime optimal algorithm,” meaning it gets better the more times you run it.
“Ride-sharing services have enormous potential for positive societal impact with respect to congestion, pollution and energy consumption,” says Rus. “I think it’s important that we as researchers do everything we can to explore ways to make these transportation systems as efficient and reliable as possible.”