The use of artificial intelligence for public transportation has been explored in places in America and developed in places around the world including the U.K. and China. The problems addressed in some implementations, as discussed in Artificial Intelligence and Bus Transportation, include lack of drivers, high costs and emissions, safety, and inefficient routes and methods of direction. Implementations in large U.S. cities may focus on manufacturing vehicles which are robust and environmentally sustainable, designing safe and traffic-oriented self-driving software, and developing timely algorithms for efficient routing.
An article from the World Economic Forum titled This US City put an algorithm in charge of its school bus routes and saved $5 million highlights the MIT Operations Research Center solution to some of the issues that the school bus system faces, including the limited number of drivers and the costs of fuel and maintenance. As participants of the Boston Public School Transportation Challenge, the Quantum Team from MIT developed an algorithm which analyzes data from Google Maps and student locations to “identify the most efficient and cost-effective routes” for Boston’s school buses (Fleming 2019). After one year of implementing MIT’s algorithm in the Boston school bus system, annual costs have dropped $5 million and total bus miles decreased by 1.6 million kilometers.
Faculty is a leading tech company in the U.K. which is using advanced machine learning models to improve bus routes in London, as outlined in Significantly improving bus scheduling for a London bus operator. A significant private-sector bus operator in London consists of a team of route controllers who guide drivers all day using GPS and radio technology. Faculty worked on a variety of complex computer models to calculate the best routes by analyzing patterns in historical data. The Support Vector Regression model (SVR) was the most complex of those tested and provided the greatest improvement to the benchmark at 38% (Faculty 2021).
As part of a series of phases of AI public transportation development, the first step which may be implemented immediately is the use of artificial intelligence optimization (AIO) for bus routing. Based on previous uses of AI modeling, algorithms could direct bus drivers while balancing factors such as bus riders, stop locations, and fuel efficiency at a rate exceeding that of human communication.
A second stage to supplement bus routing algorithms would be the use of mobile and web applications to not only log bus riders, but track numbers of people requesting public transport and waiting at various bus stops in order to provide the real-time data necessary for efficient routing. Though the prevalence of Internet access continues to rise in the United States, as shown in Internet Access data from OECD, this web technology would be optional for riders in order to ensure that the target audience of lower-income communities is uninhibited by Internet and mobile device barriers.
All while maintaining and updating bus routing algorithms, the later stages would involve small-scale trials of AI-assisted public transportation prototype vehicles in large cities with sufficient income to support the project. These prototypes would be developed with the goal of operating without a present driver and under the guidance of bus routing algorithms. The current state of autonomous vehicles in the United States is making slow progress. In the CBS article What’s the status of self-driving cars?, personal vehicles such as the 2022 Honda Civic have made vast improvements in steering, accelerating, and braking, but have not yet reached the status of “hands-free technology.” However, driverless buses are being developed in other countries such as the 5G autopilot buses in Zhengzhou, China which have the capability of being monitored and assisted by a remote driver. The trial period would model this feature to promote security and reliability to participants who may hold skepticism toward a fully autonomous vehicle. Subsequent stages of AI public transportation would involve updating the self-driving technology and vehicle models.
