Aerial Vehicle Rapid 3D Map Generation for Safe Landing Area Detection

This project assists in building a robust control LiDAR scanning unit that helps aerial vehicles land autonomously. The LiDAR scans the terrain, detects static or dynamic obstacles, and helps target safe landing areas. The whole terrain scanning system uses a robust control system to control the LiDAR position to collect point clouds around the environment and use ICP (Iterative Closest Point) to create a 3D point cloud map. In the robust control system, IMU (Inertial Measurement Unit) is used to control the servo to ensure the control system had an accurate target angle for the LiDAR scanning. MSAC algorithm is used in the 3D mapping to help the system detect flat areas for an aerial vehicle to land. This project also shows the different voxelization levels (1 × 1 and 5 × 5) of the flat surfaces and the maximum point to plane distance changed (0.05, 0.25, 1, and 10). Also, an exaggerated voxelization level (10 × 10) was investigated to get a rapid first-level estimation of flat surfaces. The results of the different values and comparisons are also noted and shared. This study is the variable resolution mapping and perception portion of our full-scale heterogeneous localization and mapping research project of Autonomous Mobile Robotics Research Group.