Abstract

Path Planning is one of the important problems to be explored in unmanned aerial vehicle (UAV) to find the optimal path between starting position and destination. The aim of path planning technique is not only to find the shortest path but also to provide the collision-free path for the UAV in unknown environment. Although there have been significant advances on the methods of path planning where the map of environment is known in advance, there are still some challenges to be addressed for dynamic autonomous navigation for the UAV in unknown environment.

This thesis research proposes a new path planning method named Fuzzy Kinodynamic RRT for unmanned aerial vehicle flying in the unknown environment. This method generates a global path based on RRT [1] (Rapidly-exploring random tree) and utilizes fuzzy logic system to avoid obstacles in real time. A set of heuristics fuzzy rules are designed to lead the UAV away from unmodeled
obstacles and to guide the UAV towards the goal. The rules are also tested in different scenarios, and they are all working efficiently both in simple and complicated cases. The UAV starts to fly along the path generated by RRT, and the fuzzy logic system is then activated when it comes across the obstacle. When the sensor detects no collision within a specific distance, the fuzzy system is turned off and the UAV flies back to the previous path towards the final destination. The simulations of the developed algorithm have been carried out in various scenarios, with the sensor to detect the obstacles. The numerical simulations show the satisfactory results in various scenarios for path planning that considerably reduces the risk of colliding with other stationary and moving obstacles. A more robust and efficient fuzzy logic controller which embeds the path planning is finally proposed and the simulation shows the satisfactory results in complicated environments.