Visual perception modelling for intelligent virtual driver agents in synthetic driving simulation

This thesis documents new research into the modelling of driver vision and the integration of a new vision model into a microscopic traffic simulation tool. It is proposed and demonstrated that modelling of driver vision enhances the realism of simulated driver decision-making and behaviour, in turn, leading to improved simulation of driver interactions and traffic flow. Driving and traffic related research has traditionally fallen into the three distinct areas of driver psychology, traffic and highway engineering and vehicle dynamics, with modelling or experimentation in any of these areas supported by significant approximation in the others. In contrast to this, the vision research discussed here has been carried out in a context that aims to integrate all of these areas equally. This has been realised through the implementation of a new modelling environment, Synthetic Driving SIMulation, SD-SIM. [Continues.]