AI-Powered Taxi Service Aims to Improve Mobility in Tokyo

The evolution of mobility-based artificial intelligence technology could help reduce congestion in the world’s most traffic-plagued cities by predicting demand for ride-hailing services and taxis.

This includes car-choked Tokyo, where a partnership between Japanese auto-giant Toyota, JapanTaxi, Japanese telecommunications operator KDDI and global consulting firm Accenture have begun trials on a taxi dispatch support system using location-based big data gathered from smartphones.

The idea is to use AI to more accurately predict when an influx of taxi requests is likely, due to special events (the upcoming Summer 2020 Olympics in the city, for example), public transport service availability or inclement weather.

Toyota’s role is to collect data affecting taxi demand, process and analyze that data on the company’s proprietary Mobility Services Platform (MSPF), and then provide JapanTaxi with its prediction data.

Accenture will develop the AI analytics algorithm powering the taxi demand prediction engine, while KDDI will develop the platform analyzing population movement, which it will gathered from the company’s location-based smartphone data.

With this data, JapanTaxi hopes to create a more efficient mobility model through its domestic taxi app, Zenkoku Taxi, which boasts more than 4 million downloads.

The Zenkoku Taxi network covers 60,000 taxis across 47 prefectures, tantamount to 25% of domestic taxis.

The service, which is available in three languages — English, Chinese and Korean — can estimate taxi fares and cashless payment through credit cards and QR codes.

The four companies began piloting the system with select taxis a JapanTaxi affiliate in February of this year, with the help of taxis equipped with tablets running the system.

The AI algorithm can predict the number of occupied taxis in Tokyo using 1500-foot, mesh-based parameters, as well as taxi service log data and demographic predictions.

According to a press statement from Toyota, a recent trial run boasted a 94% accuracy rate in predicting demand.

Over time, the AI system will be able to learn from previous data sets to even more accurately predict where taxis will be needed and when, while the tablets in the taxis will show drivers where future occupied and unoccupied taxis are located.

Taxi drivers with excellent records of finding passengers can also supply information to the system, providing insights as to where drivers are more likely to find passengers based on prior experiences.

Toyota plans to roll out dozens more taxis equipped with the system this year as it moves towards a full-fledged deployment of the technology, the company noted in the statement.

In a sign that the platform is already working, Toyota said drivers using the system recorded an average sales increase of 20.4% in February 2018, compared to an overall average increase of nearly 10%.

Meanwhile, rival Japanese automaker Nissan is underway with its fully automated ride service trial, part of an effort to put those vehicles on the road in Japan by the early 2020s.

In addition, Japanese tech giant SoftBank Group and All Nippon Airways have recently began trials of an autonomous Hino Motors minibus, which travels from Tokyo’s Haneda Airport into the city.

Japanese Prime Minister Shinzo Abe has made putting a fleet of self-driving taxis, buses, trucks and private vehicles on Tokyo roads a key goal ahead of the 2020 Summer Olympics.

— Nathan Eddy is a filmmaker and freelance journalist based in Berlin. Follow him on Twitter @dropdeaded209_LR.