Due to its capability for low latency data processing in close proximity to the data source, edge computing represents a significant transformation in modern network architectures. Furthermore, Artificial Intelligence (AI) is increasingly utilized at the edge to enhance user interactions, optimize resource distribution, and facilitate decision-making as the volume of data continues to rise and latency-sensitive applications become more common. In this context, Federated Learning (FL), an AI architecture that safeguards privacy while minimizing the flow of data to centralized servers, emerges as particularly noteworthy. Nevertheless, deploying AI at the edge presents numerous challenges. The necessity to execute complex models on devices with limited resources raises concerns about energy consumption and computational demands. Additionally, threats from adversarial attacks and potential data breaches could undermine the security and efficacy of distributed AI models. This research offers valuable insights for autonomous systems, smart urban environments, and industrial Internet of Things (IoT) applications by integrating theoretical AI concepts with practical edge implementations. Future efforts will aim to ensure dependable, scalable edge intelligence by merging AI with the forthcoming 6G networks and advancing hardware-software collaboration.