Cellular Automata (CA) has emerged as a powerful computational model that has
been widely applied in the field of traffic flow simulation, effectively
capturing the complex dynamic behaviors of traffic systems and variable
environmental conditions. With the rapid advancements in autonomous driving
technology, traditional CA traffic flow simulation models for human-driving
condition are updating, especially adapting to the Artificial Intelligence (AI)
integrated driving behavior of autonomous vehicle (AV). This paper conducts an
analysis on the existing explorations of CA-based traffic flow modelling for
AVs. First, this paper utilizes the knowledge graph analysis tool “VOSviewer” to
visually represent the relations among the state of art studies. The keyword
clustering helps to reveal current research hotspots and developmental
trajectories. Subsequently, the paper classifies how CA models are improved to
adapt the AVs, from the view of the car-following, lane-changing, AV platoon,
and AV dedicated lane. Furthermore, this paper unravels how AI technologies can
be integrated with CA models to enhance the accuracy and practicality of mixed
traffic flow models. Finally, the paper summarizes the reviews as well as the
research trends, including current research difficulties, challenges, and
potential development directions, offering valuable references and insights for
researchers and engineers in related fields.