Machine Learning-based Wait-time Prediction for Autonomous Mobility-on-Demand Systems

The development of more sophisticated autonomous course-determination mechanisms for Autonomous Mobility-on-Demand systems is an active area of research and development. In the case of traditional ridesharing systems, there are various factors to be considered such as efficient use of vehicular assets, minimizing passenger wait times and selecting course of travel. In this paper we consider the use of machine learning to aid the selection of a destination that is predicted to result in a lower wait time until the next rideshare ride request will occur in the vicinity of a trip's destination. We draw upon a real-world ridesharing dataset to build and evaluate predictive machine learning models to provide an exploratory analysis of the utility of this approach to destination selection and demonstrate promising performance.