Implementation of informed rapidly-exploring random tree on a pre-mapped Octomap generated by real time appearance based map

Simultaneous Localization and Mapping (SLAM) is a solution in robotics in which Global Positioning System (GPS) positioning is impossible. One particular example is indoor navigation which is used in poor signal or distance of objects is negligible relative to GPS precision. In this research, visual SLAM is implemented using a stereo depth camera simulated in Gazebo which will generate Point Cloud which later converted into Octomap, a spatially efficient data structure. Based on the information gathered a path planning algorithm can be implemented for the desired UAV, in this case a quadcopter because of its versatility and 3D navigation. The path planning is implemented using Informed Rapidly-exploring Random Tree (Informed RRT*) which uses an ellipsoid boundary to optimize the stochastic path planning. All these are implemented in the Robot Operating System (ROS) framework. In this research it is shown that Informed RRT* algorithm gives a solution for a quadcopter.