Presented at
DEF CON 17 (2009),
Aug. 1, 2009, 11 a.m.
(50 minutes)
Accessing centimetric georeferenced images is crucial for local military or civil intelligence, reconnaissance and surveillance applications. When classical satellite or aerial imagery is not available the Unmanned Aircraft Systems (UAS) are often a very interesting solution. But having an operational UAS for spying operations implies to solve the following practical problems: design and realize a reliable flying micro-UAS develop efficient tools for real-time navigation to ensure that the UAS will cover all target points merging all trajectory data for real-time georeferencing of high resolution imagery design appropriate software for optimal data exploitation We present in this article our practical solutions to these different points.
The first section starts by presenting our Vertical Take Off and Landing quadrotor solution. This quadrotor is highly maneuverable; practically any flying movement is possible by controlling the different motors rotation speeds. Like every small UAS, the quadrotor is very instable by nature, so it is very important to have an efficient embedded real time control. Then we comment the general functional scheme and emphasize of operational security. In particular, we explain why we decide to split inboard computations in two levels : the critical and the nominal levels. Then, we present some new model aircraft technologies that permits to realize an efficient quadrotor. Finally, we discuss about practical limitations like the dependence on the weather conditions and the endurance of the aircraft.
The following section presents our Head-Up Display solution which displays on the real-time embedded video the piloting data and navigation instructions like the direction of the following target point and the already covered zones. It also present how the UAS locate itself by merging numerous captor as a GPS, an INS, a barometer and a ultrasound system, and thus is able to navigate safely in its environment. Precise localization is vital for navigation and for efficient data exploitation.
In the third part we explain why digital cameras are the best compromise between weight, self-sufficiency in energy and data storage, optical quality and adequacy to imagery mission. Indeed, even if already integrated systems induce limited choice in term of image compression, optical parameters, acquisition rates or communication protocols, there are also very attractive ready-to-use systems with reasonable weight/size, good optical quality and internal data storage capacity (thus avoiding the image processing and transmission problem).
The last technological lock with on-the-shelf retail cameras is the ability to control them in real-time with limited computational resources. To achieve the reactivity and flexibility needed for professional imagery we need to do some reverse engineering on our camera to take full control on the power on/off, on the trigger and most importantly on the dating of the pictures. We will present the reasoning and the result we obtain in our case.
At last we will present how, with the trajectory and the precise dating of the picture, we are able to construct very quickly the georeferenced footprint database of the pictures and thus allow the user to navigate in the data few minutes after the data retrieving. The whole visualization and navigation is made in Google Earth by using smart usage of the KML format.
Presenters:
-
Corentin Chéron
Corentin Chéron is a last year graduate student in ESIEA engineering school. Is currently working in long term intership in the ATIS laboratory. He is passionate with aerospatial and embedded electronics. He has his flying model licence and is working on his private pilot licence ! The last two year he worked with Sebastien Monat on the Faucon Noir micro-UAV project.
-
Antoine Gademer
Antoine Gademer is a last year PhD student in the ATIS (Data and Signals Acquisition and Processing) laboratory at ESIEA Paris. He is obtained is ESIEA engineer's degree in 2005 and a Paris-Est University Geographic Information System's degree in 2006. He is specialized in aerial and satellite image processing and more recently in sensor integration.
Links:
Similar Presentations: