Drones and AI predicted crop stress early and adjusted irrigation

A study found that the system lowered modeled yield vulnerability and used water more efficiently than rule-based irrigation.

Editorial illustration — not from the study.

Researchers developed a system that uses drone images, soil sensors and meteorological data to track crop conditions over space and time. Rather than directly predicting yield, it forecasts how stress-related yield vulnerability may change, allowing an artificial intelligence agent to choose irrigation for different areas of a field.

In the reported experiments, the system forecast stress with 96.3% accuracy. Its adaptive irrigation strategy performed better than conventional machine-learning, image-analysis and transformer-based comparison systems, while also reducing modeled yield vulnerability and improving water-use efficiency relative to rule-based irrigation.

What the system found

The system forecast crop-stress vulnerability with 96.3% accuracy. Compared with rule-based irrigation, it reduced predicted yield vulnerability by 46.6 and improved water-use efficiency by 41.1, according to the abstract. It used drone-based visible-light, multispectral and thermal images together with soil and weather measurements to set irrigation actions for different field zones.

Why the results matter

Irrigation based on early warnings could let growers respond to developing crop stress before it is reflected in yield forecasts. The reported results suggest that combining drone monitoring with automated, zone-specific irrigation may reduce modeled yield risk while using water more efficiently than fixed rules. The abstract presents the approach as intended to support scalable precision agriculture, but does not establish how it would perform across different farms or crops.

Evidence and caveats

This is a journal research article reporting experimental results from a framework evaluated with the Crop Health and Environmental Stress Dataset. The abstract does not specify the crops, experimental setting, study duration, or whether the system was tested in live farm operations. It also reports predicted yield vulnerability rather than measured yield loss, and the meanings or units of the reported 46.6 and 41.1 changes are not given in the abstract. These details limit what can be concluded about real-world performance.

A microwave-and-plasma process kept Turkish mantı in good condition for 30 days

A combined microwave-drying and cold-plasma treatment preserved the color and sensory quality of Turkish stuffed pasta, known as mantı, during 30 days of storage. The researchers also found that the treatment affected the retention of different minerals in different ways.

56 minutes ago

Biology

An in-ear sensor recorded brain signals like nearby scalp electrodes

Researchers tested a flexible dry in-ear EEG sensor in 15 healthy adults and compared its recordings with standard scalp electrodes. The ear sensor recorded brain responses that closely resembled signals from a nearby temporal scalp site, particularly during auditory tasks.

1 day ago·46 views

Biology

Pseudomonas biofilms spread more pyocyanin in lung tissue than on lab material

A study used chemical imaging to compare Pseudomonas aeruginosa biofilms grown on a laboratory surface and on mouse lung tissue. Biofilms on lung explants showed greater production and deeper tissue penetration of pyocyanin, a phenazine involved in biofilm activity and host interactions.

2 days ago·11 views