Photo/IllutrationA Tsubame Taxi Group cab is installed with a dedicated display device during an experiment. (Provided by NTT Docomo Inc.)

Taxi dispatching just put the pedal to the metal, with NTT Docomo Inc. developing a system that utilizes artificial intelligence to find areas where people are milling about but cabs are in short supply.

Based on base-station data that connects cellphones, the system finds locations where people are gathered and analyzes the past boarding records of taxis in the area. It then estimates the needed number of cabs.

Tokyo Musen Kyodo Kumiai, a group of taxi companies in Tokyo, and Nagoya-based Tsubame Taxi Group began using the system on Feb. 15. They expect that taxis with “vacant” signs will converge more readily around areas where many customers are waiting, which will lead to an improvement in service.

According to NTT Docomo, a map displayed on a device installed at the driver’s seat is divided into 500-meter square grids. In each area, the number of taxis needed in the next 30 minutes is shown.

The data is updated every 10 minutes. Even en route to a destination, drivers can know which lanes have more demand.

They can also ascertain areas where the number of people suddenly increased other than train stations where people are constantly flocking about, such as when events are held or trains are delayed.

The AI system will gradually be installed in a total of 1,350 cabs at Tokyo Musen group and 1,150 at Tsubame Taxi.

At an experiment conducted for four months from December 2016, taxis equipped with the system earned about 1,400 yen ($13.10) more on average per day.