On-demand ridesharing operation: matching, pricing, and routing

Excessive private vehicles in densely populated cities, together with the increasing need for mobility, have been constantly challenging the existing transportation systems. Fortunately, mobility on-demand services, such as ride-hailing and ridesharing, are becoming a growing trend in megacities due to their convenience and cost-effectiveness. These services are envisioned as enablers of a shift from car ownership to vehicle usage. Nonetheless, the impact of mobility on-demand service on transport systems is complicated and largely depends on governance and operation strategies. Accordingly, this dissertation aims at developing novel management strategies, involving matching, pricing, and routing, to improve ridesharing system efficiency. Meanwhile, the impact of such methods on urban transportation systems is evaluated. Firstly, an innovative strategy is proposed to integrate vehicle assignment with the prediction of time-dependent link travel times. We unify the assignment and routing problem into a linear integer problem where k-shortest paths are provided to reduce congestion. The results indicate that the proposed strategy can significantly improve ridesharing system performance, such as reducing the passengers' waiting and travel times, by mitigating congestion effects arising from ridesharing fleets. Additionally, we account for traveller's modal choice and ridesharing pricing fairness. A novel discounting method is designed based on the proposed fairness principles. Moreover, computationally efficient optimisation models are constructed accounting for co-existing ride-hailing and ridesharing services. The real travel dataset is utilised to assess the proposed method. The results indicate that the proposed optimisation strategy, considering traveller behaviour and fairness, can significantly improve fleet performance while maintaining fair service quality. Lastly, we present a simulation-based service assessment framework to test online ridesharing strategies with shared autonomous vehicles. Individual socio-demographic features are considered in generating future demand for SAVs. Travellers' mode choices are explicitly modelled, and advanced ridesharing strategies, involving optimal matching and pricing, are tested in a mixed-traffic urban network, with both private cars and shared autonomous vehicles. Compared to rule-based methods, optimal matching and fairness pricing combined method can greatly improve both fleet performance and transportation efficiency. In summary, this dissertation reveals that dynamic ridesharing offers a promising pathway toward achieving more sustainable mobility, provided it is properly managed.