Show simple item record

dc.contributor.authorRigby, Michael James
dc.date.accessioned2016-04-14T00:52:58Z
dc.date.available2016-04-14T00:52:58Z
dc.date.issued2015
dc.identifier.urihttp://hdl.handle.net/11343/91283
dc.description© 2015 Dr. Michael James Rigby
dc.description.abstractExisting ridesharing systems are too rigid for on-demand travel. Built from the perspective of classic, vehicle assignment algorithms in operations research and optimisation fields, their designs are highly prescriptive and rely on the full disclosure of trip information from both vehicle and client to perform ride matching. For a client user this approach becomes problematic, particularly during poor or uncertain service conditions, when they may not be able to find any ride matching their request. Critically, existing systems do not provide the client with adequate feedback to assist their travel planning, resulting in a significant knowledge gap that hinders the overall usability of the system. New players such as Uber and Lyft have tried to bridge this gap using visual map-based client user interfaces. Yet in retaining the classic, vehicle perspective they too encounter the same usability issue when they are unable to respond to requests by optimising or allocating vehicles due to real-time constraints such as demand or congestion. To account for these aspects this research leverages location-aware mobile devices and spatial information systems to develop a new approach for ridesharing. It inverts the classic vehicle perspective into a service representation to create a novel way to think about and use ridesharing systems, one which allows a client to respond to the system's current state however uncertain or heterogeneous it may be. It hypothesises that it is possible to design a representation of ridesharing service potential matching a client's request, which can be understood and used for travel planning purposes. This hypothesis is tested using an incremental approach, which builds from theory through design towards validation. First, it adapts and exploits concepts from quantitative time geography to create a discrete representation of ridesharing opportunity called launch pads. Second, it extends this model to create a continuous representation that fully communicates the system's potential. Third, it enhances both representations to incorporate additional ridesharing dimensions for decision making. Fourth, usability testing is performed in various simulated scenarios, providing early feedback on the proposed approach and identifies a set of future work towards validating the concept. This thesis presents novel contributions to improve the design of software applications for ridesharing. Its major results are the identification of a knowledge gap in existing systems and the design of an opportunistic approach that overcomes it. The key contributions include the launch pad concept based on time geography, methods to extend and enhance this representation with accurate ride attributes and the design of a novel 2-step negotiation for improved design making. Importantly the new approach facilitates client flexibility in both space and time for ridesharing and contributes towards improving the usability of intelligent shared ride systems using autonomous vehicles.en_US
dc.subjectgeographic information scienceen_US
dc.subjectintelligent transportation systemsen_US
dc.subjecthuman-computer interactionen_US
dc.subjectridesharingen_US
dc.titleShared intentions in virtual space for on-demand transportationen_US
dc.typePhD thesisen_US
melbourne.affiliation.departmentInfrastructure Engineering
melbourne.affiliation.facultyEngineering
melbourne.contributor.authorRigby, Michael James
melbourne.accessrightsOpen Access


Files in this item

Thumbnail

This item appears in the following Collection(s)

Show simple item record