Dynamic Environment Aware Autonomous Mobile Robot Navigation

Abstract

Robotic technologies are getting popular in today’s world. Robots are used in a wide range of areas like defense industry, military, agriculture, research and rescue, logistics, healthcare etc. Tasks that are monotone, hard or dangerous for humans can be done by robots more efficiently and robustly. Different types of robots can be used depend on complexity of a task and technical knowledge of involved engineers. There are lots of unique problems in many of application areas, each of which requires a specific solution in terms of robot platforms and software. Today, mobile ground robots are widely used in both academic and commercial projects. However, some applications such as exploration can be implemented more efficiently due to the mobility of unmanned aerial vehicles (UAVs). Although there are different types of UAVs in the market, multi-propeller UAVs are preferred due to their low costs, better mobility and easier dynamics compared to other UAVs. When multiple UAVs are used together, a task can be accomplished even much more quickly and effectively. However, especially when working with more than one multirotor UAV in indoor environments, it can cause problems even if the environment is static. The problem is that UAVs moving indoors may crash into each other. In the path planning phase of navigation, possible collisions can be eliminated between UAVs, which are dynamic obstacles in this case. In this context, a system that creates a common 3D map using sensor data from multiple UAVs and navigates these UAVs using 3D costmaps has been developed. In xi order to prevent UAVs from hitting each other while navigating, a method that is based on costmaps has been implemented using Robot Operation System (ROS). For multi mapping, a customized version of OctoMap is used. Path planning is done by using A* and Lee Position Controller is used for path tracking. Experiments have been done in different simulation environments in Gazebo using RotorS simulator.