Skip to Content

Outdoors Structures Visual Inspection by Unmanned Aerial Vehicles (UAV)

Funded by the Spanish Ministry of Science MICYT #DPI2004-06624

The current project is a continuation of two previous projects in the area of computer vision systems applied to unmanned aerial vehicles (UAVs).  The previous projects have produced working systems for real-time tracking of external references (power line to be inspected) by both cameras of a stereoscopic vision system, while additionally estimating the relative position of the UAV.  All processing took place on a model helicopter with an on-board vision system. 

Currently, the research intent is to advance by building on the previous results.  More specifically, the objective is to control the attitude of the UAV (stationary flight) based on vision-system feedback.  To meet this objective, it is proposed that the scale of the helicopter be increased from that of previous models used, to a mini-helicopter scale.  This larger-scale model will be able to simultaneously accommodate both the vision, and control subsystems.  The visual control of the  UAV attitude will provide reliable and centered imagery of the inspection objectives relying on predefined 3D positions.  By acquiring multiple images of the same inspection subjects from different viewing angles, 3D features will be extracted, allowing the classification and automatic visual inspection of said subjects.  The acquisition of these images will occur from various stationary-flight positions controlled by the use of visual feedback.  The sequencing of UAV positions and image acquisitions will be determined and executed automatically, based on the positional data for each inspection subject.  A database of all inspection subjects and their respective positional data, will estimate the reliability each inspection vantage point. 

Starting date: 
Finishing date: