Path planning algorithms for mobile robots

Abstract

The aim of this project is the study and implementation of a path planning system for a multi-rotor within the framework of autonomous navigation. The theoretical study begins with a brief introduction to the key concepts to understand the planning algorithms, then is presented a brief state of the art about some of the most relevant algorithms and its features. Then is chosen RRT* as the most suitable for our purpose, so we proceed to the analysis of different variants exist to address the next step, creating our own version in Matlab. This finally was not suitable for our purpose due to some shortcomings and inefficiencies. In order to implement an operating tool was decided to use ROS framework and more specifically the MoveIt! software. This makes necessary an introduction to key concepts and operation of the software used. MoveIt! needs some changes to adapt it, since it was conceived as software manipulation in robotic arms not for UAV navigation. All decisions during this period has been strongly influenced by the expectation of a future real implementation. So given some of requirements previously presented, the changes made and the overall performance of the tool are explained. Finally are presented operation advices and a brief benchmarking of different algorithms available in MoveIt! after some tests. Thus we have obtained a fully operative tool for multi-rotor autonomous navigation within a simulated environment in ROS. Note its great adaptability, because it can operate at different simulated environments and with different types of multi-rotors, after just minor modifications.