Robotic soil sensing and mapping
Field sampling, nutrient analysis and adaptive mapping for precision agriculture.
Soil properties vary across a farm, but collecting and analysing samples takes time. My work in Integrated AgriSense addressed both the sensing hardware and the choice of where to sample. The University of Sydney project combines robotics and IoT sensing for nutrient and water mapping.
I led sensing, spatial mapping and robotic integration as a Vonwiller Postdoctoral Research Associate from February 2024 to February 2026. I worked directly with soil scientists and NSW farmers during system development and field evaluation.
Sampling and analysis
The T-FR paper describes a mobile sampler and a separate stationary analysis unit. A three-day pasture trial collected 30 samples. Mean analysis time was 544.6 seconds. Sample transfer, calibration and cleaning remained manual.
The earlier ICRA workshop paper describes the sample acquisition and analysis subsystems. It is listed among the workshop’s presented papers.
Spatial mapping and sampling decisions
Our Computers and Electronics in Agriculture paper (December 2025) presents a heterogeneous sensing system for soil moisture mapping. My earlier RA-L paper (July 2025) develops multivariate active learning with multi-kernel Gaussian processes. This work studies how measurements of several environmental variables can inform adaptive sampling.
Full citations are available in the publication list. Talks and field demonstrations are collected on the media page.