A Hierarchical Pedestrian Behaviour Model to Reproduce Realistic Human Behaviour in a Traffic Environment

Abstract

Understanding pedestrian behaviour in traffic environments is a crucial step in the development and testing of autonomous vehicles. As the environment's most vulnerable road users, pedestrians introduce an element of unpredictability that can lead to dangerous scenarios if their behaviours are unfamiliar to or misinterpreted by vehicles. In this thesis, we present a hierarchical pedestrian behaviour model that interprets high-level decisions through the use of behaviour trees to produce maneuvers that are executed by the low-level motion planner using an adapted Social Force Model. The presented hierarchical model is evaluated on two real-world data sets collected at separate locations with different road structures. The first data set provides a busy four-way intersection with signalized crosswalks, while the second location provides an unsignalized crosswalk across a two-way road at a Canadian university. Our model was shown to replicate the real-world pedestrians' trajectories and decision-making processes with a high degree of accuracy given only high-level routing information (start point, end point, and average walking speed) for each pedestrian. The model is integrated into GeoScenario Server, extending its vehicle simulation capabilities with pedestrian simulation. The extended environment allows simulating test scenarios involving both vehicles and pedestrians to assist in the scenario-based testing process of autonomous vehicles.