Exploring Human and Autonomous Vehicle Interactions through Agent-based Simulation

In this research, I will develop agent-based models (ABM) of urban mobility which make novel contributions to the modelling of pedestrian movement and activity planning. A richer set of pedestrian behaviours will be modelled which will be affected by interactions with the built environment and other road users, namely vehicles. The models will be used to explore the possible impacts of autonomous vehicles.

The pedestrian movement behaviours of interest are those related to crossing the road. This project will combine existing methods for modelling pedestrian movement and develop novel algorithms in order to develop pedestrian agents which exhibit the diversity of road crossing behaviour observed in the literature. Modelling road crossing interactions between pedestrians and autonomous vehicles will require developing pedestrian agents with heterogeneous perceptions of right-of-way in different sections of the road and of risk posed by oncoming vehicles. To do this, I will combine social force models of pedestrian movement and routing algorithms such as the flood fill algorithm with novel algorithms which determine the outcome of pedestrian and vehicle agent interactions in road crossing events. These novel algorithms will explore ways in which these interactions are likely to differ in the presence of autonomous vehicles and between different agents. For example, timid pedestrians will strictly wait until their right of way is assured. Opportunistic pedestrians will seek to exploit gaps in traffic to crossroads. This will allow a more holistic appraisal of autonomous vehicle mobility than is currently present in the literature.

An additional contribution of this research will be the geographic scale over which these behaviours are modelled. The ABM will consist of thousands of agents moving across a town-sized geographic area. This scale will allow the effects of changes in behaviour of agents on macroscopic traffic patterns.

An extension to this work will develop novel methods for modelling the activities people plan and participate in as they move through urban areas. Social and leisure activities are underrepresented in models of urban mobility, biasing the role of urban roads to move people to places against their role as places themselves. I will develop novel algorithms for modelling planned and unplanned social and leisure activities practised by people in urban areas. By including these activities in an ABM of urban mobility, the role of the transport system in facilitating and hindering them will be able to be better evaluated.