The recent advancements in autonomous driving research have significantly focused on enhancing sensor fusion techniques and improving the efficiency of data processing. A notable trend is the integration of radar and camera data for 3D object detection, addressing the limitations of each sensor by leveraging their complementary strengths. Innovations in transformer-based models and query-based frameworks have shown promising results in accurately detecting objects in complex environments. Additionally, there is a growing emphasis on developing efficient LiDAR data alignment methods for precise mapping and localization, with continuous-time trajectory optimization emerging as a key technique. Furthermore, the field is witnessing the development of high-precision on-device depth perception systems, which combine multiple data sources to achieve superior depth mapping accuracy. These developments collectively push the boundaries of autonomous driving technology, aiming for safer and more reliable systems.