<p>Cellular Automata (CA) modelling has emerged as a promising approach to address the computational challenges associated with microscopic traffic flow models, especially under mixed traffic conditions. CA models offer advantages in terms of flexible evolution rules and high computational efficiency. The inherent discreteness and localized cell computations provide a unique capability to connect micro-level dynamics to macro-level traffic behaviour. This paper presents a comprehensive review of existing CA models and identifies potential directions for advancing current practice to capture the features of heterogeneous and mixed-traffic environments. The paper outlines the fundamental principles of CA, emphasizing its relevance in mixed traffic flow modelling. It explores how CA models can be customized to address the complexity of mixed traffic by modifying parameters such as cell size, cell structure, and randomization rules, adopting ideas from physics and biology literature. It reviews lane-change modelling, focusing on the nuances of lane-changing behaviours in both homogeneous and heterogeneous scenarios. It also explores the representation of mixed traffic within CA models through various dimensions, including cell representation, vehicle representation, and driver behaviour representation. Each of these aspects is critically evaluated, highlighting the strengths and limitations of existing models and proposing potential enhancements to better replicate scenarios marked by a diverse range of vehicle types and driver behaviours. Thus, this comprehensive review highlights the state-of-the-art research and practice regarding the cellular automata models, their limitations, and underscores future directions that hold potential for more accurate replication of mixed traffic environments.</p>