Automation of 3D Building Model Generation Using Quadrotor

摘要

Abstract Recently, in course of the last development in information technology, control algorithms, models and simulators of buildings management, an essential role in given to 3D information modeling BIM in order to facilitate, improve and expand structure related services. A building model is generated combining images taken from different angles. The share of aerial photo acquisition is noticeably augmenting. Chronologically, this has started since the migration of the unmanned vehicle technologies to the civil market. As a result, a generation of sophisticated aerial machines equipped with high-resolution cameras can provide photographs of a surface taken from different angles allowing generating fairly accurate 3D models. The risk analysis shows that the human factor represents the biggest probable technical disturbance in the image acquisition process. Hence companies dealing with such projects try to minimize the role of the human-operator by automating the workflow. Automated aerial photo acquisition can be a solution. The quality of 3D building model, that we are trying to present in this paper, is directly related to the optimized usage, flight stability and the efficiency of the electronics. From a robotic point of view, reliable positioning of unmanned machine is an up-to-date technical task. In this paper, we will be identifying the quadrotor as an unmanned aerial vehicle performing photogrammetry. Such model was chosen because of its size, cost, maintenance-friendly and most important its maneuverability. On the other hand, the quadrotor does not offer long flight range and it is not stable. In order to improve the stability we suggest to identify and eliminate the noises acquired by sensors readings, and, on the other hand, to develop adaptive fuzzy logic position controllers to deal with the external wind factor. The effectiveness of the stabilization systems is endorsed with simulation and experimental results.