Low-Cost Aerial 3D Reconstruction System in Real-Time for Outdoor Environments based on Markovian Robust Kalman Filte

Abstract

This research project proposes the development of a low-cost 3D reconstruction system in real-time using Micro Aerial Vehicles (MAVs) for remote sensing. For this, it will be developed a Simultaneous Localization and Mapping (SLAM) approach based on robust Kalman filter for Markov Jump Linear Systems (MJLS) subject to parametric uncertainties. This filter will perform the fusion of the measures of an IMU (Inertial Measurement Unit ) and of a GPS (Global Positioning System) receiver with the position of visual markers provided by a monocular camera. The Markov jumps are used to model the transition between different modes of the system operation in order to obtain acceptable behavior and achieve performance requirements even in presence of sudden changes in the system dynamics. And the robust Kalman filter algorithm will be implemented in his Array algorithm version that has numerical characteristics interesting for real-time implementations.The acquisition of aerial imagery will be conducted by the 3D Robotics SOLO quadrotor, which will be equipped with camera, IMU and GPS receiver. The communication interface with the aircraft will be made using ROS (Robotic Operating System). For this, it will be developed a ROS driver for the aircraft SOLO, which will allow direct access to all sensors and actuators of the MAV through a computer connected via a wireless network. And the 3D reconstruction system will be developed in C ++ language.