High-Throughput Field Phenotyping of Plants

1
High-Throughput Field Phenotyping of Plants
Sri Harsha Atluri Sriharsha.atluri_at_ttu.e
du
2
Background

• World population is likely to exceed 9 billion by
  2050
• Will we be able to meet the food requirements

3
Background
The DNA and the environment (soil type, weather,
nutrition, pest, diseases, etc.) influence how a
plant will develop and grow. This is the reason
why two plants having exactly the same DNA
(genotype) do not always look alike (phenotype).
4
Background
DNA sequencing have greatly improved genotyping
efficiency and reduced genotyping costs. Methods
for characterizing plant traits (phenotypes),
however, have progressed much more slowly.
5
Background
Let us assume a mapping population

• 25 crosses each represented by 200 lines 5,000
  lines.
• 2 field replicates 10,000 plots per treatment
• 2 treatments (dry land and irrigated for example)
• Using a single row, 1-m wide by 4-m long plots
  and ignoring the need for walkways or borders the
  net row-length would be 10,000 24 80,000
  meters (about 50 miles).

6
Background

• A person walking 3km/h would need about 27 hours
  to visually score traits assuming no stopping.
• Halting at each plot for 30 seconds would require
  an additional 167 hours (about 7days).

High throughput phenotyping is needed
7
Project goal
High throughput phenotyping of individual plants
or lines in field environment for use by breeders
and biotechnologists.
8
State of the art (CSA News)
9
Greenhouse scale
Phytomorph (University of Wisconsin) Lemnatec
(Germany)

• Individual plants positive
• Greenhouse negative (too different from real
  world)

10
Field scale
The Maricopa Agricultural Centers high-clearance
tractor in operation over young cotton plants at
Maricopa, AZ. Replicated sets of sensors allow
simultaneous measurement of plant height, foliage
temperature, and foliage color (spectral
reflectance). GPS provides positional accuracy
under 2 cm. Photo by Michael Gore
11
Field scale
Researchers at CSIRO use a remote-controlled
gas-powered model helicopter called the
phenocopter to measure plant height, canopy
cover, and temperature throughout a day. Pictured
here are Scott Chapman (left), a principal
research scientist at CSIRO, and Torsten Merz,
developer of the phenocopter.
12
Our tool
13
Corobot explorer
14
Problems to solve

• Navigation

Position Accuracy less than 2 cm is
required Detect and recognize a Plant
15

• Imaging (RGB, hyper spectral, infrared)

16

• Data handling

Store data so that the data can be efficiently
interpreted
17
Navigation tasks

• Plant detection
• Plant mapping

18
RTK GPS
19
RTK GPS
20
LiDAR
Source Weiss, U., et al. Plant detection and
mapping for agricultural robots using a 3D LIDAR
sensor. Robotics and Autonomous Systems, 59(2011)
265-273
21
Test field
22

• References
• Wikipedia
• Dr. Eric Hequet
• Dr. Hamed Sari-Sarraf
• RTK Library www.rtklib.com
• Weiss, U., et al. Plant detection and mapping
  for agricultural robots using a 3D LIDAR sensor.
  Robotics and Autonomous Systems, 59(2011) 265-273.