MIT researchers develop advanced, self-learning algorithm to revolutionize carpooling
After Uber and Lyft explored the potential of carpooling using mobile data, researchers have also started to prepare themselves technically to adapt to this trend. Carpooling, if used efficiently with the help of huge amount of data extracted from millions of taxi rides and maps of road network, can reduce the traffic congestion, save fuel and time, and reduce vehicular pollution along with road accidents in the New York city, claims the researchers at the MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL).
The findings of the study were published in this week’s issue of the Proceeding of the National Academy of the Sciences (PNAS).
The lab has achieved a milestone by developing an “anytime optimal algorithm” to facilitate carpooling that keep learning with the number of rides, making it better and more efficient. The researchers claimed that a fleet of 3,000 carpooling cabs (4 passenger capacity) can cut down load of 13,000 taxis running on NYC roads.
It works in real-time to reroute cars based on incoming requests. It can send idle cars to other locations after dispatching vehicles to high-demand areas. This would increase the speed of service upto 20 percent, claims the team.
In another scenario, if larger vehicle is used, the team’s findings revealed that 95 percent of total taxi demand in NYC can be met with 2,000 vehicles (10 passenger capacity). Currently, this demand requires about 14,000 taxis in the city.
The algorithm can process thousands of requests instantly and can calculate the best route and best ride (type of vehicle) based on the analysis. The system is capable of identifying high-demand areas and dispatch nearest and most suitable vehicles to these locations.
Professor Danilea Rus of MIT’s CSAIL has authored a related paper with former CSAIL postdoc Javier Alonso-Mora, assistant professor Samitha Samaranayake of Cornell University, PhD student Alex Wallar and MIT professor Emilio Frazzoli.
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,
Rus claims that 3,000 carpooling vehicles can fulfill 98 percent of demand with an average wait-time of just 2.7 minutes.
Instead of transporting people one at a time, drivers could transport two to four people at once, results in fewer trips, in less time, to make the same amount of money. A system like this could allow drivers to work shorter shifts, while also creating less traffic, cleaner air and shorter,
less stressful commutes, said Rus.
Further, Rus explained that the system is best suited for autonomous taxis, which are the future of transportation.
What’s more, the system is particularly suited to autonomous cars, since it can continuously reroute vehicles based on real-time requests,
There is no doubt that globally, cities are getting congested and traffic jams are getting longer. Ride-sharing services can help out as the concept has enormous potential for positive societal impact with respect to problems of the modern world. Considering these developments, the future of transportation is self-evident.