Douskos Valsamis

Advanced Learning and Perception for Infrastructure Security (ALPIS)

The ALPIS project aims at developing novel methods in industrial-grade algorithms for efficiently fusing and analyzing multimodal visual data for monitoring critical infrastructures and for providing situation awareness in disaster management scenarios. On one hand, advancements of UAV/ ROV technology and optical sensors have increased their reliability and operativity for the acquisition of the data required for monitoring applications.
On the other hand, the development of advanced embedded computing solutions allows the use of cutting-edge recognition and reconstruction algorithms. The visual data will be derived from RGB, multispectral, thermal and short-wave infrared cameras, which will be integrated on UAVs (fixed-wing, or drones), ground vehicles and ROVs for maritime installations. The main contributions of the proposed project are:

• the in-time acquisition of short-range data for monitoring infrastructure elements and for now-casting using unmanned aerial vehicles (UAVs);
• the development of suitable algorithms for on-board processing for understanding and reacting in real-time to the environment using high-end embedded computing solutions;
• the introduction of novel approaches based on cutting-edge deep learning and 3D reconstruction techniques for off-line processing of the acquired data for nowcasting and situation awareness regarding critical infrastructures.