Abstract:
Google the phrase "drones transform agriculture" and you'll find roughly
500,000 hits proclaiming an upcoming revolution in agriculture, the
agricultural sciences, and much of field biology. However, one issue is
often left unaddressed in these optimistic pronouncements. Exactly how
is this revolution going to occur? While flying platforms for gathering
data have many immediate applications, the vast majority of these
systems focus on gathering huge quantities of aerial imagery that can
easily overwhelm researchers and require intensive manual analysis.
This talk examines the design and use of autonomous hexacopters for
high-throughput phenotyping. I present new methods for creating
extremely detailed geographical models that identify and locate all
plants within a field and can track their individual growth over a
season with mm2 resolution. Results from preliminary studies
conducted at the West Madison Agricultural Research Station are used to
illustrate our research goals, including predicting overall crop yields
early in the growing season. This work reveals new challenges that
require intense collaboration across a broad range of disciplines,
making this area intensely exciting and demonstrating one view of how
combining computer science, statistics, and agriculture is likely to
have major scientific and economic promise for Wisconsin and beyond.