The recent advancements in LiDAR-based technologies have significantly enhanced the precision and efficiency of 3D mapping and motion prediction in complex environments. Innovations in bundle adjustment methods, such as the introduction of graph optimality-aware stochastic optimization with progressive spatial smoothing, have demonstrated robust, efficient, and scalable solutions for large-scale LiDAR data processing. These methods leverage detailed LiDAR correspondences to refine sensor orientation and point cloud accuracy, overcoming limitations of traditional pose-graph-based approaches. Additionally, the integration of LiDAR with multi-object tracking frameworks has expanded the applicability of these systems, particularly in autonomous driving scenarios, by eliminating the need for overlapping camera fields of view. The development of multimodal feature integration for motion prediction, incorporating detailed local LiDAR features, has also shown promising results, improving accuracy and performance metrics. Furthermore, real-time photo-realistic mapping frameworks, utilizing Gaussian splatting and Gaussian Process Regression, have achieved state-of-the-art performance in large-scale outdoor environments, emphasizing both efficiency and rendering quality. These advancements collectively underscore a shift towards more integrated, real-time, and high-fidelity solutions in LiDAR-based research, addressing critical needs in autonomous systems and photogrammetry.