Dense 3D LiDAR SLAM in an Apartment Complex (Colorized)

Mapping a full apartment complex using my custom-built handheld LiDAR SLAM scanner. For this test, I walked through the apartment complex where my parents live and recorded the entire area using a Livox Mid-360 and a FLIR machine-vision camera. The LiDAR data was processed using my custom SLAM pipeline, while synchronized RGB images were projected onto the reconstructed map to generate a colorized 3D point cloud. This dataset contains more than 30 million points and includes apartment buildings, roads, trees, parking areas, sidewalks, and small surface details such as road markings. One of the interesting results is that even markings such as “소방차 전용” (“Fire Engine Only”) can be clearly recognized in the colorized point cloud. Unlike my previous videos, which focused mainly on the hardware design and development of the handheld scanner, this video focuses on a larger-scale real-world mapping test. The goal was to see how well the system performs when mapping an entire residential environment rather than a small indoor space or short outdoor sequence. What's shown in this video: Handheld LiDAR SLAM mapping of an apartment complex Large-scale outdoor 3D reconstruction Colorized LiDAR point cloud Apartment buildings and surrounding environment Roads, vegetation, sidewalks, and parking areas Detailed road markings visible in the reconstructed map Dense 3D point cloud visualization Mapping using a fully self-contained handheld scanner Software: SLAM: Custom LiDAR SLAM pipeline (C++) Loop Closure: ScanContext++ Colorization: LiDAR–camera projection using synchronized RGB images Visualization: Web-based 3D point cloud viewer Hardware: LiDAR: Livox Mid-360 Camera: FLIR Blackfly S GigE (IMX265, global shutter) Compute: NVIDIA Jetson Orin Nano Super Battery: UGREEN Nexode 200W / 25,000mAh Power: USB-C PD 20V + internal DC-DC conversion Frame: Custom-designed 3D-printed enclosure GUI: Wireless mobile Web GUI The entire scanner operates from a single integrated battery, allowing the system to be carried and used without an external display or separate power pack. There are still several areas I would like to improve, especially camera–LiDAR synchronization, calibration accuracy, color consistency, and point density. But this test shows that the system can already reconstruct a fairly large residential environment while preserving recognizable visual details through colorized LiDAR mapping. ────────── SJY Robotics SLAM, Navigation, Multi-Robot Systems