ORCID Identifier(s)

ORCID 0000-0002-0395-2707

Graduation Semester and Year

Summer 2024

Language

English

Document Type

Thesis

Degree Name

Master of Science in Aerospace Engineering

Department

Mechanical and Aerospace Engineering

First Advisor

Dr. Kamesh Subbarao

Abstract

In this thesis, an application of RTAB-Map’s visual SLAM (Simultaneous Localization and Mapping) is used to explore autonomous cooperative navigation in GPS-denied environment using a rover and a quadrotor. The primary objective is to enhance cooperative navigation by allowing the rover and the drone to guide each other using the maps constructed or generated by them. The map is generated using the RTAB-Map SLAM technique using stereo camera only. The thesis addresses three major issues: quality of map created by UAV and UGV, the visual odometry drift, and the robustness of the cooperative framework in the static and dynamic environment. The cooperative navigation is done using the ROS navigation package. The static environment would be the 2D grid map with already available obstacle information. The dynamic environment for the navigation is ensured by spawning the virtual obstacle in the trajectory of both rover and quadrotor. The simulation is performed in the virtual world (parking lot) with obstacles like traffic cones, street lights, a dumpster, and a ditch. The work is conducted using software such as Gazebo for the simulation environment, RViz for sensor data and point-cloud visualization, Blender for creating realistic textures and appearances of the parking lot.

The findings of this thesis indicate that the rover, being stable on the ground, provides better quality of map. On the contrary, the 3D point cloud map generated by the drone had some blurred and distorted locations due to the vibrations during flight. The results show that the navigation using both agent’s grid map yields the same result. Hence, cooperative mapping of a large environment can be done effectively with high quality in less time. The cooperative navigation, where both agents guide each other was very robust in reaching up to the goal. Also, the cooperative navigation was found out to be robust in avoiding the obstacle spawned in the static environment. With these results, the thesis highlights the effectiveness of the RTAB-Map visual SLAM for enhancing the cooperative navigation.

Keywords

Cooperative, Navigation, Visual SLAM, ROS, RTAB-Map, Gazebo

Disciplines

Navigation, Guidance, Control and Dynamics

Available for download on Sunday, August 10, 2025

Share

COinS