Use this URL to cite or link to this record in EThOS:
Title: The data market : policies for decentralised visual localisation
Author: Gadd, Matthew
ISNI:       0000 0004 7966 0350
Awarding Body: University of Oxford
Current Institution: University of Oxford
Date of Award: 2017
Availability of Full Text:
Access from EThOS:
Full text unavailable from EThOS. Please try the link below.
Access from Institution:
This thesis presents a Data Market for the dissemination of expertise in the form of locally accurate topometric maps which are agnostic to sensor modality and estimation strategies amongst arbitrarily large teams of autonomous vehicles with a common goal of localising well in a dynamic world. Cooperation is at first centralised and allows for rapid construction of considerable maps by the conjoining of work uploaded by asynchronous processes before sharing expertise amongst the fleet by a selective downloading of mission-specific map contents. Embedding the centralised case within formalisations supporting distributed version control systems imposes obligatory commutation properties which ensure consistent integrity and eventual convergence in the state of each agent's map, allowing us to consider how any number of decentralised agents operate in concert via data sharing policies that are germane to the shared task of lifelong localisation. Vendors in the market intermittently initiate exchanges based on either mission parameters or a demand for localisation capability and choose trading partners with discrimination based on an internally evolving set of beliefs in each other's value as providers over a catalogue of distinct places of interest - which are treated as products for appraisal and purchase. The system is implemented in more than 50 000 lines of C++ code and evaluated over hundreds of kilometres of monocular and stereo imagery from a comprehensive collection of warehouse, urban, and planetary analogue environments featuring diverse deviation in appearance due to textural, atmospheric, lighting, and structural dynamics and analysed over an exhaustive combination of the sensory records of the Oxford RobotCar Dataset.
Supervisor: Newman, Paul Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID:  DOI: Not available