3d Road Surface Meshing with LIDAR

摘要

The primary goal of this research is to create 3d maps of George Mason Fairfax roads. These maps will be used in the 3d driving simulator. Towards this goal I have decided to use Google Cartographer, which uses Lidar and Inertial Measurement Unit (IMU) data. The Cartographer has been developed and applied for slam inside buildings. In the original application Lidar and IMU units were worn in a backpack. Because of that Cartographer is adjusted for abundance of vertical features and relatively slow moving speeds. I have collected Lidar, IMU, HD video, and GPS data for the major GMU campus roads and several parking structures and lots over four days. I have used two different Lidars (VLP-16 and VLP-64) mounted on two different vehicles. I have written the needed drivers to integrate GPS data collection with VLP-16 data. I have fabricated mounts for VLP-16 and reconfigured VLP-64 mounts for these collections. I have processed the data using Cartographer in Robot Operating System (ROS). The collected data was archived on GMU servers for future teaching and research uses. In our application the Lidar and units were placed on a moving vehicle which travels at several times the speed of a walking person. In addition, the vertical features were more struct accurate point clouds for some roads and parking lots and structures, while in several cases more parameter tuning may be needed to make it work.