Smart Insertion Strategies for Sustainable Operation of Shared Autonomous Vehicles

Abstract

As shared autonomous vehicles (SAV) emerge as an economical and feasible mode of transportation in modern cities, effective optimization models are essential to simulate their service. Traditional optimization approaches, based on first-come-first-served principles, often result in sub-optimal outcomes and, more notably, can impact public transport (PT) operations by creating unnecessary competition. This study introduces four insertion strategies within the MATSim model of the Melbourne Metropolitan Area, addressing these challenges. Two strategies optimize SAV operations by considering overall network costs, and the other two make insertion decisions based on the available PT service in the network. The findings show that strategic insertions of the requests can significantly enhance SAV service quality by improving the vehicle load and decreasing vehicle and empty kilometers traveled per ride. The analysis indicates that these strategies are particularly effective for smaller fleet sizes, leading to an increased number of served rides and a more equitable distribution of wait times across the network, reflected in an improved Gini Index. The findings suggest that prioritization-based insertions significantly enhance service quality by prioritizing users with limited access to PT, ensuring that those with fewer PT options are served first, and encouraging a more integrated and sustainable urban transportation system.