Download files
Abstract
Given recent advancements of autonomous systems, the integration of autonomous vehicles into transportation systems is an inevitable situation that has implications that are not yet fully understood. Without additional research regarding large-scale deployments of autonomous vehicles, these advanced transportation systems may be deployed without sufficient planning and management in a sub-optimal manner. Under the continuous guidance of A/Prof. Dr. Jayantha Katupitiya, this thesis is aimed at studying and optimizing the widespread use of autonomous vehicles in a bus-based autonomous public transportation application.
Difficulties in studying large-scale deployments of autonomous vehicles in experiments and simulation are discussed. As there are no adequately suitable simulation software for a large-scale study, a new high-performance software platform is developed in this thesis. To drive the simulation platform, a problem setting including a transit network, passenger demand model and an autonomous vehicle model with related controllers is presented. The design focus and architecture, as well as a number of key features, of the simulation platform are also covered.
The developed simulation platform and models are then simulated and improved using new autonomous vehicle-to-vehicle and vehicle-to-infrastructure systems designed for autonomous vehicle use. A high-level vehicle management problem relating to vehicle dispatching is considered, studied and optimized. Cooperative vehicle maneuver generation using an event-based framework is presented for handling both common and exceptional situations such as lane changes and hazard avoidance. A priority-sensitive variant of autonomous intersection management is developed.
Simulation results of various bus-based autonomous public transportation systems deployments in multiple cities of differing sizes are used to validate and demonstrate the effectiveness of the methodology presented for studying large-scale deployments of autonomous vehicles. The measured performance showed that 1) overall benefits from the widespread use of autonomous systems over their traditional counterparts was obtained, and that 2) each deployment was affected by local factors such as topography, road network design, population dispersion and city size. A sensitivity analysis is performed to demonstrate the viability of using the methodology for optimizing important variables such as the city’s ideal fleet-size, vehicle capacity, and service frequency.