https://www.mdu.se/

Connected and Automated Vehicles (CAVs) rely on Vehicle-to-Everything (V2X) communication to enable cooperative perception, coordination, and safety-critical driving decisions. However, conventional V2X communication remains largely bit-centric and does not inherently account for the meaning, context, or task relevance of the transmitted information. Consequently, it is becoming increasingly inefficient at supporting the exchange of high-volume sensor and perception data required by cooperative CAV applications. Semantic and task-oriented communication offers a compelling alternative by shifting the design objective from exact bit-level reconstruction to delivering meaning and task-relevant information, substantially reducing communication overhead while preserving what matters for safe and efficient driving. This paper presents a comprehensive review of semantic and task-oriented V2X communication for CAVs. To highlight current trends, design choices, and open challenges, we critically analyze the datasets, AI/ML models, evaluation tools, and channel modeling assumptions adopted across the reviewed literature. The review further investigates how existing studies define and operationalize semantics, goals, and task relevance, alongside the design of semantic encoders, decoders, knowledge bases, and transmission pipelines. We systematically categorize key performance metrics, such as reconstruction quality, semantic fidelity, and task effectiveness, and synthesize the reported gains of semantic communication over conventional wireless systems. Beyond synthesis, the paper critically examines key open design challenges, including semantic communication in broadcast scenarios, mobility-aware design, emerging semantic-level security threats, and the coexistence of semantic and conventional communication. Finally, we highlight key limitations and future research directions, concluding that while semantic and task-oriented V2X communication exhibits strong potential in controlled environments, substantial challenges remain before its real-world deployment in future intelligent transportation systems.