Selected Topics on Shared Mobility Systems
MetadataShow full item record
Autonomous vehicle technology is right around the corner. Millions of car miles have been tested by leading companies like Google, Uber, Tesla, General Motors and so on. An autonomous car-sharing system, or ACSS, is more plausible than ever. Compared to traditional taxi system, there are aspects of challenges and opportunities. One of them is that the self driving capability of the autonomous vehicles (AV) offers some unique features to the ACSS. For example, the vehicle is no longer limited by the location of drivers. An AV may serve a trip where the parking cost is high and park at suburbs by itself where the parking cost is low. Another aspect is the sheer volume of trips ACSS needs to deal with. With potentially much lower fare, an ACSS may finally become a compelling option to serve mainstream travel demand and replace a significant portion of private vehicle trips. It is not uncommon to have hundreds of thousands of demand or even more within a single day for metropolitan areas. Realizing new modeling and computational needs, this dissertation attempts to tackle some of the challengesThe first chapter provides a random utility based framework to estimate the parameters for a mathematical model called household activity pattern problem (HAPP). The second chapter tackles ACSS as a variant of vehicle scheduling problem (VSP). We propose two models to solve the same problem. One is called autonomous dial-a-ride problem (ADARP) which works with a continuous time line while the other is a time space flow model called ACSS which works with discretized time stamps and aggregated zone-to-zone flow demand. The goal is to analytically investigate the difference between the results from these two models and utilize the ACSS solution to provide upper and lower bounds for ADARP. The third chapter deals with the changing demand due to customer reneging. The potential customers may switch to alternative mode, private AV trips that are comparable to ACSS, provided that the service quality is degraded. The proposed model endogenously balances lost profit due to reneged customers and the increased operational cost to improve service level. An improving heuristic is proposed to generate a feasible solution to the problem.