We move towards an era of smart cities, where shared autonomous vehicles will provide efficient and safe on-demand transportation for all. In this talk, and building upon his seminal work on on-demand high-capacity ridesharing via dynamic trip-vehicle assignment, Dr. Alonso-Mora will give an overview of his recent work in this field. He will first analyze some of the sources of uncertainty present in ridesharing systems, followed by a discussion of methods for predictive routing that utilize a model of future demand. Secondly, he will discuss a multi-objective optimization that enables to design the required vehicle fleet accounting for the trade-off between quality of service and operation cost.