UAVs Applications.Civil and Research Applications
|
Infrastructure Inspection
AMPL researchers collaborated in infrastructure inspection drone developed in the Department of Systems engineering and Automatics of UC3M with the collaboration of COPISA; for the development of a UAV capable to perform an analysis of the defects and the structures in buildings, based on a 3D camera mounted on a quadcopter. This system is presented as European patent.
The presented system is divided into two main stages; Depth-Color image correlation, Plane segmentation and distance estimation, and Visual servoing. The system has been validated with real flight tests, and the obtained results show the accuracy of the system in both inspection measurements and the UAV maneuver controlling. |
|
Aerial Visual Surveillance
This project is related to the detection and counting of people during the day and night, thanks to the use of the cameras (Color and Thermal) mounted on the UAV.
n this way, the UAV is equipped with a sophisticated security system with a great range of work through the use of a monocular perception system. The implemented system is based on the Histogram of Oriented Gradients (HOG) algorithm combined with machine learning algorithm (Support Vector Machine (SVM)). The algorithm combines the detection of head and body to improve the capacity of the system of people counting to high altitudes and complex environments. |
|
Search and Rescue
A vision-based system implemented on a UAV; in order to detect the human bodies and the objects; for rescue as well as transportation purposes. In this system, two algorithms as a framework to cope with cutting-edge UAVs technology. Real-time object detection and classification and maneuver control are studied as a complex and essential task for intelligent aerial vehicles in rescue and transportation systems.
The proposed algorithms take the advantages of onboard low-cost depth sensor to extract the objects from the ground plane and calculate its distance to the UAV. Moreover, by using an SVM human detector based on HOG features, the system classifies the human bodies of the victims from all the extracted objects. In addition, a semi-autonomous reactive control is presented; to control the velocity and the position; in order to perform safe and smooth approaching maneuvers to the detected objects. |
|
Traffic Monitoring
Unmanned Aerial Vehicles (UAVs) are being widely used for traffic flow monitoring and for surveillance labours. This platform is used due to its high manoeuvrability, being able to cover unreachable areas by fixed cameras. Our cutting edge technology can automate this task by detecting and tracking objects even in such a complex scenario where objects present high variation in size and scale as a result of the camera angle.
|