Project dates and role
Micro Atlas ran from June 2015 to May 2018 at the University of Manchester. Dr. Costilla Reyes supported development of the drone-based system, which was designed to detect crop disease, drought, pests, and nutrient shortages. His records also list an Agri-Sensor Research Associate position from August 2016 to August 2017, held alongside his doctoral studies in the School of Electrical and Electronic Engineering. He was also a research assistant supporting undergraduate courses in C programming, robotics, sensors and instrumentation, and data networking.
Sensing for agricultural monitoring
Drone-based crop monitoring collects imagery over fields so that early indicators of disease, drought stress, pest damage, or nutrient shortage can be flagged for closer inspection before they become visible from the ground. Hyperspectral and multispectral sensors extend this idea by recording reflectance across many narrow wavelength bands, which allows differences in plant or surface chemistry to be analyzed with pattern-recognition methods. The project belongs to the sensing and machine-learning theme of Dr. Costilla Reyes’s Manchester research.
The agri-sensor research appointment
The agri-sensor appointment investigated whether hyperspectral imaging could detect rodent activity through suitable chromophores, light-absorbing or fluorescing compounds, at a sensitivity sufficient for rapid, real-time intervention, using unsupervised learning, which finds groups in data without labeled examples. That work is described in the 2017 IEEE SENSORS paper on unsupervised learning for spectral data analysis, coauthored with Zachary Coldrick and Bruce Grieve, which clustered the fluorescence spectra of rodent urine against reference compounds. The paper’s laboratory measurements characterize the spectra that a field sensor, such as one carried by a drone, would read.