A Network-Effect-Aware Recommendation Engine for Multimodal Shared Mobility

Vehicle-sharing platforms are constantly gaining ground in smart cities around the world, reducing the number of traditional fuel-based vehicles on the roads in busy areas and thus contributing to the development of a sustainable environment.

Onthe other hand, the availability of a plethora of shared vehicles of different types across a city increases the need for their seamless combination, so that they are considered part of a unified transportation system within a smart city rather than independent solutions. In this work, we present a system that enables authorized users to gain access to shared vehicles of different transport modalities, allowing them to reach their destination without relying on a private car or public transport.

For this purpose, we have used existing systems and techniques from different fields, such as recommendation systems, machine learning, and route planning, which provide appropriate multimodal routes while taking into consideration several parameters, including user demographics, vehicle status, environmental conditions, and road traffic congestion. The evaluation of the system using simulated data showed that it enables users to identify suitable multimodal routes, either through explicit preferences or by inferring them from historical data, and revealed limitations to be addressed in future work.