Optimising mobility and charging services with shared electric autonomous vehicles

This study proposes a novel service platform in which operators hire privately owned electric autonomous vehicles to provide mobility services to passengers and charging services to these vehicles. The model maximises operator profits by integrating endogenous optimisation of the hired e-SAV numbers, a private-vehicle sharing mechanism, and nonlinear battery charging and discharging. To solve the problem, we tailor an Adaptive Variable Neighbourhood Search (Adaptive VNS) algorithm that integrates problem-specific operators, a charging-station insertion procedure, and a Service Start Time Optimisation (SSTO) procedure. We compare the Adaptive VNS algorithm with GUROBI, a traditional VNS algorithm, and an ALNS baseline on both small and large instances. The proposed algorithm produces competitive solutions in a much shorter time. Sensitivity analysis reveals that the effectiveness of operational strategies is scale-dependent: homogeneous fleets perform better under the tested vehicle-composition setting, while the benefits of mixed charging requirements vary with order scale.