The emergence of advanced 3D sensing technologies, such as LiDAR, has significantly increased the availability of point cloud data, driving the need for robust analytics through deep learning. Point clouds, with their detailed spatiotemporal structures, are vital across numerous applications, requiring innovative approaches for effective interpretation and utilization. This chapter delves into the intersection of deep learning and point cloud analytics, covering essential tasks like point classification and semantic segmentation. It then explores place recognition, object retrieval, and registration, emphasizing their importance in interpreting dynamic environments. This chapter concludes with an examination of multimodal analysis, showcasing the synergistic potential of integrating point cloud data with other data modalities. Each section systematically unpacks the problems, general solution strategies, seminal contributions, and emerging trends, encapsulating the state-of-the-art in deep-learning-based point cloud analytics and paving the way for future advancements in the field.