development (Towler
et al.
2012). Siminski (2014) reported
regarding the monitoring of the Fukushima, Japan plant’s
radiation with an
UAV
; this post-disaster intervention with
persistent observation minimizes the threat and hazard to the
inhabitants and response personnel.
Oil Spill Detection
Oil slicks and perhaps other pollutants on the sea surface are
becoming more prevalent creating large contaminated areas.
The field-of-view of the camera-based systems onboard
UAVs
rarely covers the entire area. To address this problem with
a single
UAV
equipped with a camera, Lanillos
et al.
(2009)
proposed a strategy for boundary detection of a surface that
is partially imaged with the camera. Each image contains a
segment of the oil spill extent, and the
UAV
is used to search
for joining consecutive segments to close the full boundary.
A simulated scenario is used for testing and an optimization
strategy is selected for closing the complete contour.
The oil slick detection problem was addressed in Muttin
(2011), where the
UAV
guides a ship for subsequent oil conten-
tion and recovery, based on a non-linear dynamic model for
the aerial umbilical including variable length domain and
material elasticity using different numerical examples.
Floods and Avalanches
Some years ago, nanotechnologies were seen as useful for
weather observation (Manobianco
et al.
, 2008). Now these
technologies are proposed for monitoring flash floods from
UAVs
. During the first phase, a number of transmitter sensors
are deployed inside the potential area where the flood could
occur, and then
UAVs
, equipped with receiving antennas,
identify each sensor position and build a distribution map.
Abdelkader
et al.
(2013) used lagrangian (mobile) micro-
sensors emitting a unique identifier (
ID
), similar to the radio-
frequency identifiers (
RFID
). When transmitters are dropped,
UAVs
track their movements using passive receiving antennas.
This application is very appropriate for multiple
UAVs
in col-
laboration to build a map of the transmitters.
In post emergency situations, Weng
et al.
(2011) used
UAVs
for monitoring debris flow in Zhouqu County area in China; a
collection of high-resolution images was acquired to evaluate
the disaster.
SAR
-based systems onboard
UAVs
were considered to be
used for detecting and studying temporal evolution of wet
snow in avalanche prone areas (Malnes
et al.
, 2015).
A new research topic emerged to study the feasibility of
utilizing 4G-LTE signals in combination with
UAVs
for search
and rescue of avalanche victims (Wolfe, 2014).
Epidemiology
Fornace
et al.
(2014) have used a commercially available
UAV
for mapping landscapes with the aim of detecting potential
areas for future human infections with the zoonotic malarial
parasite Plasmodium knowlesi. The flights were conducted in
two study sites in Sabah, Malaysia and one site in Palawan,
The Philippines. Spatial information to integrate movements
of human and macaque was recovered and analyzed to local-
ize the focus of epidemiology infections.
Jones (2014) provides a review about trends in plant virus
epidemiology, focusing on new or improved technologies ap-
plied to research in this topic.
UAVs
, equipped with different
sensors, are considered useful tools for the future to acquire
sufficient knowledge of different types of plant epidemics,
their development, and how they could be controlled.
Barasona
et al.
(2014) have used
UAVs
to capture informa-
tion about the spatial epidemiology distribution of tuberculo-
sis in the ungulate community in the Doñana National Park in
Southwestern Spain).
Humanitarian Localization and Rescue
When a disaster has occurred, people risk their own lives to
rescue others trapped due to the consequences of the event.
As mentioned before, disasters involve people requiring
urgent rescue. Tsunamis, earthquakes, shipwrecks, fires, or
nuclear leaks are typical examples of these events. Simulated
and real scenarios are used for experimentation with
UAVs
.
Rapid intervention is crucial in the early hours of the event
for effective search and rescue. With such purpose, Naidoo
et
al.
(2011) designed and simulated a stable
UAV
quad-copter
platform equipped with three main modules, each with
different sensors onboard: (a) communication (wireless and
radio controller); (b) vision system (cameras and sonar range
finders); and (c) attitude and heading reference system (
GPS
,
magnetometer, accelerometer, and gyroscope). This system
was successfully simulated.
Molina
et al.
(2012) used an unmanned helicopter
equipped with thermal and optical cameras for search and res-
cue, considering that operations could be conducted at night.
In search and rescue tasks, the time is a critical parameter
for the survival of the trapped or missing persons. Search
strategies in
UAVs
equipped with camera-based systems must
be optimized, so that the search is done focusing on areas
of maximum probability in minimum time. Cameras with
sufficient field-of-view for recognition, combined with ef-
ficient path planning strategies are suitable for such purpose,
such as the methods proposed in Lin and Goodrich (2009),
where different simulated scenarios were used for testing. In
this way, Lanillos (2013) proposed a minimum time-based
search strategy for moving targets in simulated scenarios with
uncertainty. Several agents (
UAVs
) collect information from
the sensors onboard for target identification and location. A
flight path planning process based on a probabilistic Bayesian
framework was established for early actuation.
Multiple simulated observations, based on different alti-
tudes and sensor configurations were studied in Waharte and
Symington (2010) to assess the robustness of the target detec-
tion for the purpose of covering the full area to be examined.
A camera with birds-eye view, onboard a quad-copter, was
used for search and rescue in Symington
et al.
(2010), where a
tracking approach is proposed for a static target. Image video
sequences were used for training, where key-points invari-
ant to translations, rotations, and scale changes are used for
recognizing the target (persons) on the image. This goal was
achieved by applying image-based similarity measurements
and different parameters from the observation model to up-
date a recursive Bayesian estimator.
Rudol and Doherty (2008) combined thermal and color
imagery to locate areas with high probability of presence of
humans to be rescued. The thermal camera, mounted on a
pan-tilt-unit, discriminates temperatures and locates human
temperature ranges; the
CCD
color camera at a later stage veri-
fies the presence of persons. The
UAV
platform was a helicop-
ter with total length of 3.6 m, including the main motor and a
maximum
TOW
of 95 kg.
Coyle (2014) described some cases where
UAVs
have been
used for early rescue in snow avalanches. Fixed-wing aircraft
or rotary-wing copters, equipped with video cameras and
infrared imagery sensors, are mentioned.
Mardell
et al.
(2014) have applied simulation techniques to
compare the performance of image inspection modes for visual
search and rescue tasks in wilderness areas based on an
UAV
re-
mote sensing system.. Live video and serial visual image analy-
sis (a static image remains in view until replaced by a new image
rate) captured from a downward facing camera were analyzed.
Surveillance: Target Detection and Tracking
Surveillance is an important issue in
UAVs
applications; most
tasks described above involve this application. For example,
300
April 2015
PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING
