Performance evaluation of path planning algorithms for autonomous UAV deployment using two urban scenarios

In this work, three path planning algorithms for autonomous unmanned aerial vehicles (UAVs) are proposed to be implemented in two urban scenarios. The first scenario is a simple environment without obstacles and the second scenario is a realistic environment in which an urban environment with its infrastructures such as bridge is modeled. In the first step, the test environments are modeled using Gazebo 3D with Robot Operating System. Then, 3D occupancy grid mapping using the Octomap library. Lastly, the path planning simulation is conducted for three sampling-based algorithms, namely, Probabilistic Roadmap (PRM), Rapidly Exploring Random Tree-star (RRT*), and Expansive Space Tree (EST). Three performance objectives from each algorithm are evaluated, i.e., planning time, path length, and number of subpoints. The results show that in a simpler environment with fewer disturbances, PRM or EST is more suited as they take less computational time. In complex environments with more obstacles, EST is more suitable to be implemented as it generates shorter paths with reasonable time but when the shortest path distance is mandatory, selecting RRT* is preferable. Overall, the selected methods can be easily applied in an actual environment by programming the instructions in a microcontroller of a UAV equipped with similar depth camera sensors.