Pedestrian Models for Autonomous Driving Part II: High-Level Models of Human Behavior

Autonomous vehicles (AVs) must share space with pedestrians, both in carriageway cases such as cars at pedestrian crossings and off-carriageway delivery navigating through crowds on pedestrianized high-streets. Unlike static obstacles, pedestrians are active agents complex, interactive motions. Planning AV actions the presence of thus requires modelling their probable future behaviour well detecting tracking them. This narrative review article is Part II a pair, together surveying current technology stack involved this process, organising recent research into hierarchical taxonomy ranging from low-level image detection to high-level psychological models, perspective an designer. self-contained covers higher levels stack, consisting models behaviour, prediction individual pedestrians' likely destinations paths, game-theoretic interactions between autonomous vehicles. survey clearly shows that, although there good for optimal walking social still remains open question that many conceptual issues be clarified. Early work has been done descriptive qualitative but much needed translate them quantitative algorithms practical control.