Future Demand Aware Vehicle Dispatching Systems

In an era marked by the widespread adoption of GPS-enabled devices and the emergence of vehicle dispatching platforms, on-demand ride-hailing services have become increasingly popular in urban areas. Improving the served-passenger ratio stands as the primary objective for vehicle dispatching systems, as it directly impacts their profitability. Although extensive research has been conducted in pursuit of this goal, three major challenges persist in dispatching systems: (1) the need to reduce high operational costs arising from balancing supply and demand, (2) the requirement to consider dynamic traffic and demand data when partitioning road networks for vehicle dispatching, and (3) the development of a profit-oriented vehicle dispatching system tailored to driverless fleets. This thesis undertakes an extensive investigation to address these three challenges, with a specific focus on cost-effective supply-demand balancing, the development of a dynamic-information-aware road network partitioning, and the creation of a profit-centric dispatching system. Primarily, the thesis introduces the Future-Demand-Aware Vehicle Dispatching system (FDA-VeD), which enhances the served-passenger ratio by intelligently relocating idle vehicles based on potential future demands. By considering both pickup and drop-off requests, FDA- VeD effectively reallocates idle vehicles at a low frequency, resulting in a significant improvement in the served-passenger ratio at a lower cost. Subsequently, the research proposes the Dynamic Future-Demand-Aware Vehicle Dispatching system (dFDA-VeD), which dynamically determines relocation centres, taking into account changing travel demand and traffic conditions. In contrast to traditional relocation centres based on static information such as point-to-point distances, dFDA-VeD adapts to real-time conditions. The system is evaluated using real-world datasets, demonstrating significant improvements in the served-passenger ratio over existing state-of-the-art methods, while maintaining operational efficiency. Furthermore, the thesis delves into the Profit-Oriented Future-Demand- Aware Driverless Vehicle Dispatching system (pFDA-VeD). In the context of driverless vehicles, where human drivers are absent, pFDA-VeD takes ownership of autonomous cars and seeks to maximise net profit. The system accomplishes this by factoring in the potential future profitability of trips, guiding proactive vehicle relocation to balance supply and demand across regions. Comparative assessments against multiple state-of-the-art dispatching systems confirm that pFDA- VeD achieves the highest levels of profitability. In conclusion, this thesis introduces innovative strategies aimed at cost-effectively enhancing service performance, efficiently integrating dynamic traffic and demand data into the partitioning of road graphs, and establishing a profit-oriented framework for the dispatch of driver-less vehicles. Additionally, this research ventures into exploring prospective pathways for the future development and study of vehicle dispatching systems.<p></p>