Autonomous Navigation in GPS-Denied Environments Using Visual SLAM

This paper presents the real-time implementation of a visual Simultaneous Localization and Mapping (SLAM) for indoor quadcopter navigation. The research aims to autonomously navigate in GPS-denied spaces by localization and mapping in real time. The proposed system integrates an Intel RealSense D435 depth camera, a Raspberry Pi 4B for onboard computation, a laser rangefinder for altitude measurement, and a Pixhawk 4 flight controller. The RTAB-Map SLAM is first tested in a simulation using ROS and Gazebo to ensure its robustness and further deployed in real-world experiments. The experiments carried out successfully demonstrated autonomous indoor UAV navigation with a quadcopter operating at a maximum speed of 0.5 m/s while maintaining a set altitude of 0.5 m with a mean path error and mean altitude deviation of 0.094 ± 0.031 m and 0.027 ± 0.005 m respectively, across different indoor environments, while producing 2D grid maps and 3D point cloud (PCL) maps. This work demonstrates that indoor autonomous UAV navigation can be realized using low-cost, off-the-shelf components and open-source SLAM techniques.