Preparation of LIDAR Point Cloud Data for Use in a ROS/Gazebo Simulation Environment

Open Access
Author:
Sirera, Nicholas Robert
Area of Honors:
Mechanical Engineering
Degree:
Bachelor of Science
Document Type:
Thesis
Thesis Supervisors:
  • Sean N Brennan, Thesis Supervisor
  • Sean N Brennan, Honors Advisor
  • Eric Russell Marsh, Faculty Reader
Keywords:
  • ROS
  • Gazebo
  • Driving Simulator
  • LIDAR
Abstract:
The driving simulator at Penn State University is currently in a state of disarray. Several attempts have been made in the past year to make bandage style fixes, each time raising more questions than they have answered. This thesis examines the prospects of designing an entirely new software architecture from scratch based on open-source freely available software. This decision was made not only to keep costs down, but to also take advantage of the large user bases and potential help available in the open-source community. The previous architecture, based on CarSim, was deemed too expensive and too prone to failure to pursue again. This research will examine an environment based on the Robotic Operating System (ROS) and the Gazebo Simulator. In addition to determining if this type of architecture is possible, this research will also investigate methods for converting LIDAR roadway scans into simulation models, and will attempt to make these models lifelike by applying physical characteristics. In addition, general conclusions as to the potential success of this new driving simulator architecture will be explored. Specifically, it was concluded that the method used in this paper to convert LIDAR point cloud data into a Gazebo simulation should not be recommended for further study. However, the underlying architecture of ROS and Gazebo simulator is strongly recommended for continued application in Penn State’s driving simulator.