Localização e mapemento simultâneo (SLAM) com visão monocular aplicado em robô aéreo

Abstract

Mobile robotics is an area that has aroused the interest of the community academic. With the emergence and popularization of technologies such as MEMS ( electromechanical devices), digital cameras, RGB-D sensors, etc., it was possible to design robots Miniature Stable Aerials (MAVs) and low cost, enabling the safe realization of indoor flights. In this scenario, the quadrotor has become the platform of choice for academic community, emerging several commercial solutions. For proper control of mobile robots it is necessary to estimate the pose of the robot. An estimation technique discussed in much work is SLAM (Simultaneous Localization and Mapping). In this work, it is carried out a study on SLAM techniques for estimating the pose of mobile robots multirotor aerials in indoor flight, using as sensors an IMU (measurement unit inertial) low-cost strapdown configuration and a monocular camera. Specifically, the problem of estimation and control of a quadrotor was considered. AR.Drone from Parrot, without the need to modify the factory firmware. THE The SLAM technique developed in this work uses a Kalman filter developed by author the detection of printed patterns using the open source software ARToolkit, thus, the SLAM problem is simplified. The method was implemented in ROS (Robot Operating System) and takes advantage of AR.Drone's factory firmware features. Per Finally, the applicability of the technique as a source of feedback for a position controller.