IEEE Robotics & Automation Magazine - December 2016 - 51

Instead, let us now consider the sequential deployment of
drifters. Assuming that we drop one drifter at the HTP outfall once an hour and that a drifter travels with an average
speed of about 0.3 m/s and reaches the boundary of the
monitoring area after one day on average, then we require
around 30-60 drifters for continuous operation. The exact
number of drifters depends on various factors; e.g., it
depends on the number of ships in service for timely drifter
recovery, once the drifters reach the boundary or get stuck at
a certain location. Also, the required number of drifters
would multiply if more than one drifter, potentially with different drogue depths, was deployed at a time. In practice,
with regard to the concrete HTP monitoring task, a reasonable monitoring plan could make use of about 10 drifters,
where a single drifter with drogue at depth of 3 m (i.e., the
depth of the outfall pipe) is dropped four times a day.
The above numbers motivate our overall drifter design,
which aims at an ocean-monitoring network of multiple
low-cost drifters that are easy to recover. For now, we tested
the monitoring task under the multidrifter scenario in field
experiments with up to three RESL drifters deployed in the
ocean [Figure 10(d)]. In the experiment, a HOBO Pendant
temperature sensor was attached to the center of the drogue
of each drifter. The sensors are 5.8 cm × 3.3 cm × 2.3 cm
in size and allowed to measure temperature at varying
depth. The experiment confirms that it is feasible to add
small sensors to a drogue without sensibly affecting the
drogue's mechanism.

Google

Figure 10(a) shows the trajectories of two drifters with
drogue depths of 1 m and 5 m. Once deployed, they started
separating from each other
over time. The radio communication was enabled
Ocean currents can change
for the experiment. The
data received from the
significantly over time and
SPOT tracker only included 10 GPS coordinate
have different directions
points for the experiment
duration of 2.5 h. The
at different depth layers.
combined location data
from both SPOT tracker
and radio modem consists of 86 coordinate points, resulting in more accurate tracking of the drifters' trajectories. Figure 10(b) visualizes these
differences in accuracy.
Figure 10(c) shows the trajectories of all three drifters,
with their drogues adjusted to 1 m, 3 m, and 5 m in depth
for a deployment over 3 h. The drifter with the drogue
depth of 1 m (yellow trajectory) deviates slightly from the
initial deployment location. This, together with the measured final separation between the three drifters, confirms
that the resulting movement vectors (and thus the ocean
currents) at different depth layers can be significantly different. The multidrifter experiments were conducted six
days apart, at the same location and time of day. The movement of the drifters, however, pointed into opposite

Google

(a)

(b)

Figure 9. Drifter trajectories based on ROMS data. Trajectories resulting from ROMS for a simulated drifter with (a) drogue at 1-m
depth and (b) drogue at 10-m depth. The trajectories generated by ROMS are clearly different from the data obtained by a deployed
real drifter (as shown in Figure 8).

DECEMBER 2016

IEEE ROBOTICS & AUTOMATION MAGAZINE

Table of Contents for the Digital Edition of IEEE Robotics & Automation Magazine - December 2016