OPTIMIZATION-BASED CONTROL ALGORITHM: DEVELOPMENT AND TESTING FOR DYNAMIC ON-DEMAND SAV OPERATION

Abstract

Motivated by the growth of ride-sharing services and the technological evolution in autonomous vehicles (AV), this thesis seeks to develop and assess an operational platform for an on-demand autonomous vehicle hybrid sharing system (AVHS). The AVHS system is comprised of a fleet of AVs controlled by a Central Operation Manager (COM) that provides three levels of service to travelers ranging from a taxi-like service, to a flexible-route, flexible-schedule transit-like system. The proposed AVHS system operational platform is built as a dynamic, sequential, and time-dependent stochastic control problem whose objective is to simultaneously minimize the costs associated with the operator and the traveler. The system is developed and tested using an agent-based modeling and a simulation framework where the different agents and layers of the system interact with each other and react to the surrounding environment. The multi-layered system consists of three main components: i) the traffic network layer, ii) the vehicles fleet and the COM layer, and iii) the travelers layer. This study develops a complex dynamic optimization-based control algorithm for a dynamic on-demand shared autonomous vehicles operation and tests the system’s flexibility and resilience, through a sensitivity analysis on different testing cases defined by the AV fleet size, travel demand rate, and demand composition by levels of service. The operational performance is assessed against different KPIs from both, operation and system user perspectives. The system showed its ability to entertain multiple levels of service where, to the best of the authors’ knowledge, this has not been achieved before. Results showed that the developed system was able to maintain the quality of service among different levels of services by reducing the travelers’ waiting time, increasing vehicle occupancy, and reducing the empty vehicles miles traveled.