A 3D Benchmark for UAV Path Planning Algorithms: Missions Complexity, Evaluation and Performance

Unnamed Aerial Vehicles can carry out several missions. It is essential to use path planning algorithms to obtain an efficient trajectory without collision to complete these missions. Several path planning algorithms in the literature are divided into four categories: exact classic, approximate classic, meta heuristic, and machine learning. However, it is not easy to define which is the best for each mission among all path planning algorithms. In this context, an analysis between these categories can facilitate research determining the techniques that obtain better results in each UAV mission. So, in this work, we perform a deep benchmarking of 3D path planning techniques in a simulated and real environment. As a result, classical techniques demonstrate better capacities in dynamic path planning. On the other hand, meta heuristic and machine learning techniques performed the best results for static path planning.