The field of embodied intelligence and navigation is rapidly advancing, with a focus on developing more efficient and robust methods for navigating complex environments. Recent research has explored the use of hierarchical structures, semantic-aware approaches, and adaptive curriculum learning to improve navigation performance. Notably, the use of bigraphs and open scene graphs has shown promise in organizing and maintaining spatial information effectively at scale. Additionally, the integration of curiosity-driven exploration and intrinsic rewards has led to more robust and diverse exploration strategies. The development of benchmarks such as UAV-ON has also facilitated the evaluation and comparison of navigation methods in open-world environments. Some noteworthy papers include: GeoExplorer, which proposes a curiosity-driven exploration approach for active geo-localization, and UAV-ON, which introduces a benchmark for open-world object goal navigation with aerial agents. SA-GCS, which presents a semantic-aware Gaussian curriculum scheduling framework for UAV vision-language navigation, and GACL, which proposes a grounded adaptive curriculum learning approach for robotics tasks. CogniPlan, which leverages conditional generative layout prediction for uncertainty-guided path planning, and SkeNa, which introduces a sketch map-based visual navigation task and a large-scale dataset for research. $NavA^3$, which proposes a hierarchical framework for understanding high-level human instructions and performing spatial-aware object navigation, and Open Scene Graphs, which introduces a modular system for open-world object-goal navigation using foundation models and open scene graphs. HDDPG, which proposes a hierarchical deep deterministic policy gradient algorithm for autonomous maze navigation of mobile robots.