IEEE Robotics & Automation Magazine - December 2023 - 48

15
10
5
-15
-10
-5
-5
-10
-15
(a)
(b)
5
15
consistently within 10 cm of the target.
A video of the deployment process is
available as supplementary material.
To test the repeatability of retrieval,
the sampler was placed on the target
and then retrieved with the UAV, also
piloted manually. The percentage of
successes (11 of 12) is shown in Table 2,
along with mean wind speed during
retrievals (
WS )R . Similar to the deployt
= 0 s
(c)
t = 10.08 s
ment, the UAV takes off, flies to 10-m
altitude, flies to the sampler, descends
and collects it, then returns to the
takeoff location to deposit the sampler
(Figure 6). A retrieval is considered a
success if the sampler is collected from
the target and placed on the designated
area next to the takeoff pad with the
retrieval mechanism properly engaged.
There were three practice runs before
the experiment. The single failure case
during the experiment occurred when
the sampler was laterally misaligned,
such that it hooked onto one of the magnets,
which caused the other magnet not
to disengage. This was counted as a failure,
even though the sampler remained
attached and returned to the takeoff
pad. A video of the retrieval process is
available as supplementary material.
t = 5.72 s
(d)
(e)
FIGURE 5. Deployment accuracy and outdoor test. (a) Deployment: downward-facing UAV
camera view of the sampler descending on the target, the red dot is the top of the sampler
hook assembly. (b) Scatter plot of sampler locations after being deployed with the UAV (n =
10), in centimeters. (c) Deployment process: (left) sampler suspended beneath the UAV,
(right) sampler successfully deployed (red wire highlighted for clarity). (d) Upward-facing
sampler view after deployment. (e) Downward UAV view after deployment.
TABLE 1. Deployment accuracy.
xr
(CM)
1.4 ± 6.04
yr (CM)
2.9 ± 5.59
t = 5.52 s
OUTDOOR VOLATILE COLLECTION
FROM DAMAGED MAIZE PLANTS
We tested whether our samplers were
capable of recovering volatiles emitted
from maize plants exposed to simulated
insect herbivory damage outdoors.
For this experiment, 15-day-old maize
plants (Zea mays var. Delprim) placed
outdoors were damaged, and volatiles
collected with our lightweight samplers
at different time periods. The maize
plants were grown in individual pots at
24 °C and 50% relative humidity in the
WSD (M/S)
1.71
TABLE 2. Retrieval repeatability.
RETRIEVAL (%) WSR
(M/S)
91.6
0.35 ± 0.26
RETRIEVAL TIME (S)
61.67 ± 16.91
48 IEEE ROBOTICS & AUTOMATION MAGAZINE DECEMBER 2023
greenhouse facilities at The University of Zurich Irchel campus.
To induce the plants, we imitated insect herbivore damage
with mechanical damage and application of insect
regurgitant to wounds [26]. The mechanical damage on the
leaves was performed by scratching the leaves with fine sandpaper
(100 grit) and fine tweezers (Supplementary Figure 3).
The regurgitant of the Egyptian cotton leafworm Spodoptera
littoralis from L4 and L5 larvae stages was used (10 µL per
plant; see [27] for details). The experiment consisted of placing
10 plants in a circle around each of the four samplers at a
distance of approximately 30 cm from the sampler (Supplementary
Figure 3). This occurred at 08:30 a.m., and the plants
IEEE Robotics & Automation Magazine - December 2023

Table of Contents for the Digital Edition of IEEE Robotics & Automation Magazine - December 2023
