Federated learning for vehicular networks is an emerging research area that focuses on enabling decentralized, privacy-preserving model training across connected vehicles and roadside infrastructure without sharing raw data. This approach addresses challenges in data privacy, bandwidth constraints, heterogeneous vehicle data, and dynamic network topology. Research explores integrating federated learning with deep learning architectures such as convolutional neural networks (CNNs), recurrent neural networks (RNNs), graph neural networks (GNNs), and transformer-based models to support tasks like traffic prediction, autonomous driving, vehicle-to-everything (V2X) communication, and anomaly detection. Studies also investigate communication-efficient aggregation, personalization for vehicle-specific conditions, secure and privacy-preserving techniques (e.g., differential privacy, secure multiparty computation), and robustness against adversarial attacks and data poisoning. Applications include intelligent transportation systems, real-time traffic management, cooperative autonomous driving, predictive maintenance, and vehicular edge computing. Recent research highlights federated learning as a promising paradigm for scalable, secure, and adaptive vehicular networks in the era of connected and autonomous vehicles.