Ricerc@Sapienza

Land monitoring is widely recognized as an invaluable tool to ensure situation awareness
for prevention, prediction and protection in almost any field when threats due to exogenous agents
may harm human safety and productivity. One of the applications in which monitoring systems can unleash
their full potential is agriculture, which accounts for 40% of the land usage on the entire planet.
The scarcity of infrastructure available on fields and the typical harshness and vastness of terrains make
it difficult to enable complete and continuous monitoring by skilled human personnel and calls for agile,
autonomous systems for accurate and precise inspection.
This project envisions the realization of a cooperative network of aerial sensing devices, capable of autonomous
and adaptive deployment over a field of interest, with the purpose of providing collaborative
heterogeneous sensing. The aerial monitoring network will enable threat and disease recognition, by autonomously
interacting with a complex and rich machine learning system which operates in a continuum
spectrum of fidelity, adaptively determined on the basis of the findings and of the uncertainty of detection
and related risk. On the basis of the output of the machine learning system, the aerial mobile sensors may
be called to more refined and complex missions on the field, to enhance network coverage of the field.
The combination of the two above autonomous systems, the autonomous aerial network and the machine
learning system, will enable monitoring at adaptive levels of accuracy for fast detection of diseased
crops, with identification of changes in the monitored features across the space and time domain, without
human guidance. The use of this system will enhance human performance by providing insight levels from
Umanned Aerial Vehicles (UAVs) that would be impossible for expert humans to see or record directly.