Shared Autonomous Vehicle Scheduling with Heterogeneous Ride-Sharing Preferences and Mixed-Capacity Fleet Sizing

With the rapid advancement of shared autonomous vehicle (SAV) technology and its great potential in optimizing urban transportation, scheduling models that rely solely on single-capacity vehicles struggle to effectively accommodate passengers’ heterogeneous ride-sharing preferences. This paper proposes a mixed-capacity SAV scheduling strategy that accounts for heterogeneous ride-sharing preferences, aimed at balancing diverse passengers’ ride-sharing willingness and maximizing the transportation benefits of shared mobility. By introducing the concept of “weighted edge contraction,” the weighted graph structure for solving the ride-sharing matching problem is simplified. First, by considering passengers’ heterogeneous ride-sharing preferences that encompass no ride-sharing, 2-passenger ride-sharing, 4-passenger ride-sharing scenarios, we analyzed system performance across different fleet configurations. The results demonstrated that deploying a mixed fleet with capacities of two and four passengers led to better performance across both system efficiency and service quality metrics. This strategy not only reduced the total number of vehicles in the network, decreased overall travel distance, and lowered fuel consumption, it also lowered passenger travel fares without compromising average waiting times. Next, by considering differences in passengers’ willingness to ride-share, we analyzed how system performance changed with different ride-sharing request ratios. The results revealed that higher ride-sharing acceptance rates facilitated travel request consolidation, leading to fewer deployed vehicles, shorter travel times and distances, and reduced energy consumption. Therefore, our proposed mixed-capacity vehicle scheduling strategy offers theoretical insights for optimizing SAV operations across diverse request scenarios.