A comprehensive review of deep learning techniques for interaction-aware trajectory prediction in urban autonomous driving

Abstract

Autonomous vehicles can improve urban transport by using multiple components that accurately represent their surroundings and improve decision-making processes. One essential component is trajectory prediction, which estimates the future states of traffic participants and anticipates hazardous scenarios. There are different approaches for trajectory prediction, in which Intention-aware and Interaction-aware approaches represent the state-of-the-art since they involve better representation of the surroundings. This paper reviews the literature on Interaction-Aware Trajectory Prediction for autonomous vehicles. It explores how incorporating maneuver intentions and interactions can improve prediction accuracy, and it examines the techniques and datasets employed in this field.