| Resumo: | Over the past few years the use of autonomous mobile robots that operate in indoor environments has grown significantly. They are used not only in industry settings but are creeping in on our homes and office environments. They are commonly used for application such as transportation, telepresence and cleaning. The robot’s perception for navigation purposes usually depends on laser range finders, infra red depth sensors or the typical cameras. However recent advancements on Frequency Modulated Continuous Wave radar technology permits a new solution for this navigation task. The main focus of this dissertation is the evaluation of the FMCW radar as a sensor for indoor navigation. To do that we will compare it with the 2D-LiDAR, that is the very common when it comes to robotic sensors. First, a study between each technology will be presented including the operating principle of each one, applications and what limitations they have. After that we will uncover how we can use the Robot Operating System (ROS) framework and state of the art algorithms to address the autonomous navigation problem. Afterwords we will describe the robot’s hardware and software components and how they are interconnected to produce a suitable robotic platform that will be used to do navigation tasks. Several experiments were devised and implemented to evaluate the usage of the radar in navigation. The results obtained proved the validity of using the FMCW radar as an obstacle detector sensor device. They showed some type of objects not or poorly detected by a LiDAR can be better detected by the radar. They also showed that the less dense data produced by the radar does not have a significant impact in keeping the navigation costmaps updated. Finally an exploratory work will be conducted that tries to use the doppler channel readings of the radar to aid in more safe pathing trajectories for social indoor environments.
|
|---|