Autonomous electric minibus scheduling with departure-time-shifting strategy under random conditions

Autonomous electric minibuses are being deployed in various cities worldwide, offering the advantage of serving as a platoon of buses with variable capacities to accommodate fluctuating passenger demand in terms of both time and space. However, this type of service introduces new challenges to the vehicle scheduling (VS) problem, particularly under random conditions. This study proposes a novel stochastic optimisation model for the VS problem, utilising a trip-extended modelling approach. To enhance the connections between trips and improve VS efficiency, a departure-time shift procedure is introduced. An efficient solution approach is employed to solve the stochastic model by integrating sample average approximation method with a genetic algorithm. A case study conducted on a real bus route in Dandong City, China, demonstrates that the proposed VS model with departure-time shifting results in a significant reduction, saving 6.78% in fleet size and 2,458.24 CNY compared to the non-shifting solution.