The field of federated learning is witnessing significant developments, particularly in addressing the challenges of heterogeneity and non-IID data distributions. Researchers are exploring innovative strategies to improve the efficiency, accuracy, and robustness of federated learning models. Notable advancements include the use of clustering methods to mitigate the effects of heterogeneity, the development of novel aggregation strategies, and the application of techniques such as prototype contrastive learning and feature disentanglement to enhance model generalization. These advancements have the potential to enable more effective and efficient federated learning in diverse scenarios, including edge devices and AIoT applications. Noteworthy papers include: FedIFL, which proposes a federated cross-domain diagnostic framework for motor-driven systems with inconsistent fault modes. Enhancing Federated Learning with Kolmogorov-Arnold Networks, which demonstrates the superiority of Kolmogorov-Arnold Networks over traditional Multilayer Perceptrons in federated learning tasks.