V International Conference on Forest Fire Research

D. X. Viegas (Ed.), 2006

Unmanned Aerial Vehicles as tools for forest-fire fighting

A. Ollero 1, J. R. Martínez-de-Dios 1, L Merino 2
1
University of Seville, Seville, Spain, {aollero, jdedios}@cartuja.us.es
2
Pablo de Olavide University, Spain, lmercab@upo.es

Abstract: This paper reviews the possibilities of UAV applications to forest-fire fighting in the
typical before-during-after scheme. Before-fire they can be used for monitoring of the vegetation
and the estimation of hydric stress and risk index. UAVs can also be applied for forest-fire
detection, confirmation, localisation and monitoring. Finally, the UAVs are also useful for the
evaluation of the fire effects and particularly for the estimation of the burnt area. The paper also
shows how UAV with different characteristics can cooperate in forest-fire fighting. Some of the
concepts and applications presented in the paper are illustrated by means of recent experimental
results carried out in the COMETS project. In this project a fleet of heterogenous UAVs was
applied for forest-fires detection, confirmation, localization and monitoring.

Keywords: Fire fighting tools, unmanned aerial vehicles

1. Introduction

Airborne means play an increasing role in operational forest fire activities. In

addition to forest fire extinction and transportation of fire brigades, they are also used for
fire monitoring and, in some countries, for fire detection. In these applications the crew of
the aircraft has to include people experienced in fire monitoring or fire detection. Human
piloted aircrafts are also used by the forest fire services for the measurement of the burnt
area.
Unmanned Aerial Vehicles (UAVs) have advantages over these aerial systems.
They avoid the human risk inherent to manned aerial vehicles, and also the cost can be
lower. The UAVs can be used to send images and data to a ground field station or to a
control centre. Moreover, UAVs can carry on board enough computing power for automatic
detection and measurement of fire properties.
In the last decade UAVs have attracted a significant interest. They have been widely
used for military applications but, recently, the evolution of UAV technologies, the
miniaturisation of the sensors and cameras and the new advances in communication and
control systems, (Ollero and Merino, 2004), have extended UAV technology to a wide
range of civilian applications such as natural disasters, inspection, search and rescue, traffic
surveillance and law enforcement.
Forest fire detection and monitoring have been recognized as a potential UAV
application proposed by several authors in the last years. The manoeuvrability of UAVs
makes them suitable to obtain better points of view while reducing the human risk of using
traditional means.
This paper studies the potentialities of UAV technology for several forest-fire
related activities. First the currently existing UAV platforms are briefly reviewed in section
2. Then, the possible applications of UAVs in the before-during-after scheme are analysed
in sections 3, 4 and 5. Conclusions, acknowledgements and references are the last Sections.
In order to illustrate the potentialities of UAVs some results obtained in the
framework of project “COMETS: Real-time coordination and control of multiple
heterogeneous unmanned aerial vehicles” (IST-2001-34304) of the IST Programme of the
 V International Conference on Forest Fire Research
D. X. Viegas (Ed.), 2006

European Commission, are presented. The main objective of COMETS project was to
develop an open architecture for multi-UAV coordination and control, (Ollero et al., 2005).
The COMETS system was demonstrated for automatic fire detection and monitoring.

2. Unmanned Aerial Vehicles

UAVs are self-propelled air vehicles that are either remotely controlled or are
capable of conducting autonomous operations after being launched. UAVs differentiate
from purely remotely guided vehicles in that at least they provide functionalities for data
gathering and transmission. Some UAVs are equipped with enough instrumentation to
allow off-sight tele-operation. Others are fully autonomous. The autonomous operations
that some UAVs can perform range from waypoint following to autonomous take-off and
landing, autonomous event detection and tracking and others.
UAVs have been typically classified depending of the flight altitude and endurance.
Stratospheric UAVs operate at an altitude higher than 10 km and with autonomy of up to
several months. They are based on fixed-wing or blimp platforms and they will be able to
carry earth-observation sensors and communication means. UAVs such as the USA/NASA
Stratospheric AeroVironment “Global Observer” UAV, “StratSat” Stratospheric airship of
Advanced Technology Group and the Lockheed-Martin Stratospheric airship are planned to
be available in the next five years.
Global Hawk from Northrop Grumman is a High Altitude and Long Endurance
(HALE) UAV. HALE-UAVs have significant payload and can carry on board heavy
sensors. They can be used for earth observation and as a communication backhaul, very
useful in crisis situations in which traditional communication systems can be damaged. In
earth observation, they can offer data with higher spatial and temporal resolutions than
satellites. QinetiQ has also recently developed Zephyr, a HALE UAV that is being used in
some projects as a remote sensing platform (Fransaer, Vanderhaeghen and Everaerts,
2004).
The Predator from General Atomics (Figure 1a) is an example of Medium Altitude
Long Endurance (MALE) UAV.
The so-called tactical UAVs, such as the Pioneer from Pioneer UAV Inc. have
higher maneuverability but lower payload. They can carry light sensors and cameras, which
can be of interest for local observations.
Furthermore, in the last years, man-portable or hand-launched UAVs, called
“Organics UAV”, such as Pointer and Dragon Eye from AeroVironment, have been
presented. Moreover, many mini UAVs have gained recently a lot of attention. Organic and
mini-UAVs can carry only very light cameras and sensors. These include, among others,
the Black Widow manufactured by AeroVironment and the MicroStar from BAE. These
UAVs can be used as local flying sensors providing images and data at very short distances.
Table 1, shows the comparison of several UAV platforms. Details of these and other
military UAV platforms as well as the existing related technologies can be found in the
Unmanned Aerial Vehicles Roadmaps of the USA Office of the Secretary of Defense
(www.uavforum.com and www.fas.org/irp/program/collect/uav_roadmap2005.pdf).
Besides the fixed wing aircraft, which is the main platform in existing UAVs, there
are many different Vertical Take-Off and Landing (VTOL) UAVs including helicopters
such as the Fire Scout, the Yamaha R50 and Rmax (see Figure 1c) and several new designs
such as the Guardian from Bombardier, and the Sikorksy´s Cypher or Dragon Warrior (see
 V International Conference on Forest Fire Research
D. X. Viegas (Ed.), 2006

Figure 1d). The main advantage of helicopter platforms is their manoeuvrability and ability
to maintain the aerial vehicle in hovering, which is needed for many applications. However,
they are more difficult to control than fixed-wing UAVs, require experienced safety pilots
for their development and application in emergency conditions, are more difficult to control
and usually have lower endurance and flight range.

a) b)

c) d)
Figure 1: a) The Predator from General Atomics. b) The EADS UAV c) Yamaha Rmax
platform used in the WITAS project d) The Sikorsky´s Dragon Warrior (Cypher2).

Table 1: Characteristics of some UAVs.

Weight Length Wingspan Payload Ceiling Radius Endurance
(kg.) (m.) (m.) (kg.) (m.) (km.)
Global Hawk 12134 13.5 35.4 890 19800 10000 32 hr.
Predator 10206 8.75 14.8 204 7625 750 >24 hr
Predator B 4763 11 20 1360 15259 3700 > 30 hr
RQ8 FireScout 1202 7 8.4 136 5800 280 5 hr.
helicopter (rotor span)
Pioneer 205 4.3 5.2 34 4600 185 5 hr.
Yamaha Rmax 85 3.3 3.11 10 2000 16 1.5 hr
(rotor span)
Dragon Eye 2 0.75 1.16 0.5 300 4.5 45-60 min.

Low altitude autonomous airships have been also developed. Airships can be used
to provide global views or to act as communications relay. However, the main problem
with low altitude airships for forest-fire fighting is the maximum wind velocity they can
withstand, which could be a problem with the wind velocities that may be encountered in
forest-fire operational scenarios.
The cooperation of several UAVs has also many potentialities in forest-fire
activities. For instance, different UAVs of a fleet can survey different regions at the same
time, allowing the establishment of priorities in the surveillance according to the interest or
risk of the region. Different types of UAVs can play a different role. Thus, for example,
tactical UAVs or mini-UAVs can be used to confirm and precisely localize alarms detected
by the High-Altitude and Long Endurance (HALE) UAVs. Moreover, the complementary
 V International Conference on Forest Fire Research
D. X. Viegas (Ed.), 2006

of data gathered by different UAVs can be used for confirmation of the fire alarms and thus
increase the robustness of the system. For instance, automatic forest-fire detection is
affected by many sources of false alarms such as solar reflections, heated objects, human
activities and others, (Ollero et al., 1999). A very high rate of false alarms can be discarded
by using data from different sources, (Arrue et al., 2000).
In (Casbeer et al., 2005) the feasibility of the application of a team of small (low-
altitude short endurance) UAVs to cooperatively monitor and track the propagation of large
forest fires is explored. The paper provides simulations using a six degree of freedom
dynamic model for the UAV and a numerical propagation model for the forest fire. The
cooperation and control of a fleet of heterogeneous UAVs has been researched in COMETS
Project.
The approach adopted in COMETS is to use several low-cost heterogeneous UAVs
cooperating to perform coordinated activities, (Ollero et al., 2005). The heterogeneity of
UAVs and sensors onboard them increases the complexity of the problem, but also provides
clear advantages such us the possibility of exploiting the complementarities of different
UAV platforms with different mobility attributes and also different sensors with different
perception functionalities. Moreover, many applications require several sensors that can not
be carried by only one UAV due to payload limitations. The multi-UAV coordination and
control architecture developed in the COMETS Project was demonstrated for the
autonomous detection and monitoring of fires. The whole fire surveillance mission
demonstrated in COMETS can be decomposed in the following stages: fire search, fire
confirmation, fire observation and post-fire stage. The mission is very similar (at a different
scale) to fire detection operations currently performed by fire extinguishing services using
aerial manned aircrafts.
Traditionally, three main platforms are used for fire fighting activities: ground
systems, systems on aerial means, and satellite based systems. These platforms still have
different technological and practical problems for their use in operational conditions, such
as lack of reliability, costs and others. The integration of these systems with UAVs provides
many potentialities for forest-fire fighting allowing the implementation of multi-spatial and
multi-time resolution approaches. Figure 2 illustrates this concept.

Figure 2: Multi-time and multi-spatial resolution in forest-fire fighting.

 V International Conference on Forest Fire Research
D. X. Viegas (Ed.), 2006

In the following sections, we present a brief review of the already tested and
potential applications of UAVs for forest-fire fighting in the before-fire, during-fire and
after-fire scheme.

3. Potential Applications of UAVs before fire

Some of the most relevant before-fire activities are the forest surveillance, building
of fire risk maps, monitoring of the vegetation and even generating 3D vegetation maps
with scanning sensors such as Lidars, see for instance (Wehr and Lohr, 1999) and (Lefsky
et al., 1999). These tasks are widely carried out by satellites, (San Miguel Ayanz et al.,
2005). Aerial manned aircrafts have been also used for detailing mapping with significant
cost. Although satellite-based surveillance systems have been proved to be valuable for
surveillance in large and homogeneous regions, their main drawback is that the spatial and
temporal resolutions of the data they produce is still low for some activities.
Some current HALE UAVs and future stratospheric UAVs are suitable for this task
due to their high endurance and high payload capable of carrying relatively heavy
observation sensors. These UAVs are suitable to generate fire risks maps and hydric stress
maps. Moreover, small and low-cost UAVs can be also used for local high resolution
mapping, which can be useful in special protected areas and also to support forest fire
research and development activities.
Several techniques have been already developed to generate accurate digital
elevation maps from data and images gathered by UAVs. See for instance (Lacroix et al.,
2003). Moreover, multi-resolution data fusion techniques can be used to extend the
resolutions of satellite data by merging them with data of higher resolutions gathered with
HALE UAVs. Additionally, the images and data provided by UAVs can be used to survey
the forest and detect potentially dangerous activities.
Several UAV technologies have been tested for forest surveillance. For instance, the
use of small and simple fixed-wing UAVs has been analyzed in the UAVNet European
Network. The use of ultra-light UAVs for forest surveillance is considered in (Hucaljuk,
2004) and in (Restas, 2006). In these works the UAVs were used mainly to relay images to
a control centre. Figure 3 shows two photographs taken in the UAVNet Regional Meeting
in Szendro (Hungary) in September 2005. In these projects the applications of UAVs for
fire detection and monitoring, as considered in the next section, is also studied.

Figure 3: Small planes can be used to relay images.

The approach adopted in the COMETS Project involves several low-cost

heterogeneous UAVs cooperating to perform coordinated activities. Fire experiments were
carried out at the airfield of Lousã (Portugal) in May 2003, 2004 and 2005. In these
 V International Conference on Forest Fire Research
D. X. Viegas (Ed.), 2006

experiments the following UAVs were deployed: the helicopter Marvin (Remus et al.,
2002), the airship Karma (Lacroix et al., 2003) and the helicopter Heliv. In forest
surveillance, the COMETS control centre divides the surveillance area among the UAVs
attending to the capabilities of each UAVs (i.e. endurance, and radius) and their sensors on-
board. While patrolling, each UAV applies automatic image processing techniques.

4. Potential Applications of UAVs during fire

The most extended traditional automatic systems for fire detection is based on
ground systems. There are some automatic forest-fire detection systems such as BOSQUE,
which relies on infrared cameras, or ARTIS-FIRE, which performs smoke-plume detection
with visual cameras. These systems have drawbacks such as the coverage and the lack of
reliability of the automatic detection in changing environmental conditions. The use of
UAVs can clearly overcome these drawbacks with potentialities ranging from the
automatic detection, confirmation and precise localisation of fires up to fire monitoring and
measurement.
Although satellite systems have been applied for automatic fire detection and
monitoring (Chuvieco and Martin, 1994), the detection delays and low resolution of
satellite data are still significant drawbacks. Stratospheric, HALE and MALE UAVs can
provide data with higher frequency and resolution. Furthermore, tactical UAVs and small
low cost UAVs can be used to confirm and localize the alarms, as well as to monitor the
fire, when launched nearby.
UAVs suitable for fire detection should have high endurance capabilities. Thus,
fixed-wing UAVs are more adequate for detection applications. Besides, for monitoring
applications, VTOL UAVs can hover at a desired position and can be used to obtain
detailed images of a fire. Furthermore, the UAVs can also obtain automatically the geo-
referenced position of the fire front and send it to a control centre for fire fight
management.
The use of complex single UAVs in forest-fire fighting scenarios has been analyzed
in the FiRE project in the United States and in the COMETS project in Europe.
The ALTUS UAV, an evolution of the Predator UAV, has been demonstrated in
Fire experiments in the FiRE project (Ambrosia et al., 2003). The data received at the
ground station are geo-referenced. The processing time from data acquisition onboard the
UAV, through satellite uplink/downlink, processing into a geo-referenced image data file,
to a fire manager in a remote control centre, is less than an hour. During a recent
experiment using satellite uplinks, scientists were able to transmit thermal data to the fire
manager in just 10 minutes. A 24-hour mission, using a UAV to collect data about select
target fires supplying real-time data by telemetry is planned for 2006 in the framework of
the USA WRAP (Wildfire Research and Applications Partnership).
In the COMETS system, the fire detection mission is divided in tasks for each UAV
with detection capabilities. Heliv was equipped with a low-cost micro-camera in the far
infrared band 7-14 microns and visual video camera. Marvin carries a fire sensor, whose
main component is a photodiode set-up to limit its sensibility to the band of [185, 260] nm,
normally associated to fires. Marvin and Heliv have motorized pan and tilt units that allow
orientating the cameras and the fire sensor independently from the body of the vehicle.
Figure 4 shows the images and data received from Heliv during a fire experiment.
COMETS implements a wide variety of automatic processing techniques ranging
from fire segmentation techniques in images, automatic stabilisation of images, automatic
 V International Conference on Forest Fire Research
D. X. Viegas (Ed.), 2006

geo-location of objects, and data fusion techniques for cooperative detection and
confirmation, (see Merino et al., 2006).
While patrolling, each UAV applies fire detection automatic processing. Heliv
performs fire segmentation in infrared images, and Marvin fire detection with its fire
sensor. If a fire alarm is detected, the control centre rebuilds the plans, and the UAV that
detected the fire is commanded to hover at a safety distance from the fire alarm. The
remaining UAVs are sent to confirm the alarm by using their sensors. Then, cooperative
fire detection techniques described in (Merino et al., 2006) are applied to merge results
from Heliv and from Marvin. If the alarm is found to be false, forest detection is resumed.
If the alarm is confirmed, the fire monitoring stage stars. Then, the UAVs are commanded
to hover on to the fire alarm and to synchronously obtain stabilized images of the fire from
different points of view. Figure 5 left shows an aerial image taken by Marvin at the fire
observation stage. The images are stabilized in real-time, using the procedure described in
(Merino et al., 2006). With these stabilized images, and using the fire segmentation
algorithms, it is possible to determine parameters of the fire evolution. Figure 5 right shows
images with the fire front marked. Cooperative perception techniques are applied to
improve the estimation of the fire front position. Figure 6 illustrates the approach.

Figure 4: Heliv teleoperation station, showing the data gathered during controlled forest-
fires experiments.

Figure 5: COMETS fire monitoring using Marvin and Heliv. The evolution of the fire front
is automatically obtained from consecutive images.
 V International Conference on Forest Fire Research
D. X. Viegas (Ed.), 2006

Another important function that can be carried out by UAVs is communications

relay. A fire can take place where no communications infrastructure is available. An UAV
could be placed hovering or near-hovering at high altitude acting as a repeater for different
communicating systems (Djuknic, Freidenfelds, and Okunev, 1997), (Thornton et al.,
2001). Several projects around the world deal with HALE UAVs for communications, as
the SkyNet initiative in Japan, the finished HeliNet project founded by the European
Commission, the Solitair platform by DLR in Germany and others.

Figure 6: Cooperative perception in fire monitoring.

5. Potential Applications of UAVs after fire

UAVs can be used also for post-fire analysis. MALE and tactical UAVs can be
applied to detect the presence of active fire embers. Other tasks carried out during this stage
are burnt-area mapping. Different techniques have been developed for high resolution aerial
mapping from UAVs. For instance, (Lacroix et al., 2003) proposes a method to obtain
digital elevation maps with accuracy of one centimetre from images gathered by UAVs.
Figure 7 shows a map obtained by on-line mosaicking more than 500 aerial images from
Karma UAV in COMETS experiments.

6. Conclusions

This paper reviews the possibilities of UAV applications in forest-fire fighting in the
typical before-during-after scheme. Before-fire they can be used for forest surveillance and
monitoring of the vegetation and the estimation of hydric stress and risk index. UAVs can
also be used for forest-fire detection, confirmation, precise localisation and monitoring.
Finally, the UAVs are also useful for the evaluation of the fire effects and particularly for
the estimation of the burnt area.
The paper also shows how UAVs can cooperate in forest-fire fighting operations.
The paper presents some results from recent experiments of cooperative fire detection with
 V International Conference on Forest Fire Research
D. X. Viegas (Ed.), 2006

several heterogeneous UAVs (helicopters and blimps), with heterogeneous sensors

(infrared and visual cameras, fire sensors and others).

Figure 7: Mosaic on-line built from Karma images in COMETS experiments.

The main current barriers for the applications of UAVs are related to the costs,
reliability and robustness for the forest-fire operation conditions. It is expected that the
evolution of UAV technology together with new particular R&D efforts addressing the
forest-fire peculiarities will lead to systems that could be operated in the near future. In this
context it is very important to consider the operation by forest-fire management services
and the integration in the operational procedures of these services. Some of these problems
are common with other applications and require new regulations allowing the use of UAVs.
The integration with manned flights in the forest fire scenario is another important issue
requiring particular efforts.

7. Acknowledgements

The work described in this paper has been developed in the project “COMETS:
Real-time coordination and control of multiple heterogeneous unmanned aerial vehicles”
(IST-2001-34304) funded by the European Commission(5thFP) and AEROSENS
(DPI2005-02293) funded by the Spanish Government. The authors especially thank the
COMETS teams of the LAAS at Toulouse, and the Technische Universität Berlin for their
images and data and their participation in the general COMETS experiments, as well as to
Prof. D.X. Viegas and the ADAI team from the University of Coimbra (Portugal) for the
support in the fire experiments, and all the other members of the COMETS team.

8. References

Ambrosia, V.G., 2002, Remotely Piloted Vehicles as Fire Imaging Platforms: The Future Is
Here! Wildfire Magazine, May-June.
 V International Conference on Forest Fire Research
D. X. Viegas (Ed.), 2006

Arrue, B. C., A. Ollero, J.R. Martínez-de Dios, 2000. An Intelligent System for False
Alarm Reduction in Infrared Forest-Fire Detection. IEEE Intelligent Systems, vol. 15,
no. 3, pp. 64-73.
Casbeer D.W., D.B. Kingston, R.W. Bear , T.W. McLain, and S.M. Li, 2005. Cooperative
Forest Fire Surveillance Using a Team of Small Unmanned Air Vehicles. Intl. Journal
of System Science, January 2005, pp 1-18.
Chuvieco, E., Martin, P.: 1994, A simple method for fire growth mapping using AVHRR
channel 3 data, International Journal of Remote Sensing, 15, 3141-3146.
Djuknic, G. M., L.J. Freidenfelds, Y. Okunev, 1997. Establishing wireless communications
services via high-altitude platforms: a concept whose time has come?, IEEE
Communications Magazine, pp. 128-135, Sept. 1997
Fransaer, D., Vanderhaeghen, F., Everaerts, J., 2004. PEGASUS: Business Plan for a
Stratospheric Long Endurance UAV System for Remote Sensing. The International
Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences,
Istanbul, Turkey, GITC, Lemmer, The Netherlands.
Hucaljuk, M., 2004. Remote sensing of wild fires by an ultra-light Unmanned Aerial
Vehicle. 24th EARSeL Symposium New Strategies for European Remote Sensing
Dubrovnik, Croatia, 25-27 May 2004
Lacroix, S., I-K. Jung, P. Soueres, E. Hygounenc J-P. Berry, 2003. The autonomous blimp
project of LAAS/CNRS - Current status and research challenges. Experimental
Robotics VIII, Springer Tracts in Advanced Robotics, 5, pp. 487-496.
Lefsky, M.A., D. Harding, W. B. Cohen, G. Parker, and H. H. Shugart, 1999. Surface Lidar
Remote Sensing of Basal Area and Biomass in Deciduous Forests of Eastern
Maryland, USA, Remote Sensing of the Environment Environ, 67, pp. 83-98.
Merino, L., F. Caballero, J.R. Martínez-de Dios, J. Ferruz, A.Ollero, 2006. A Cooperative
Perception System for Multiple UAVs: Application to Automatic Detection of Forest
Fires. Journal of Field Robotics. vol. 23, no. 3, pp. 165-184
Ollero, A., L. Merino, 2004. Control and perception techniques for aerial robotics. Annual
Reviews in Control, 28, Issue 2, pp. 167-178.
Ollero, A., B.C. Arrue, J.R. Martínez-de Dios, J.J. Murillo, 1999. Techniques for Reducing
False Alarms in Infrared Forest-Fire Automatic Detection Systems”, Control
Engineering Practice, vol. 7, no. 1, pp. 123-131.
Ollero, A., S. Lacroix, L. Merino, J. Gancet, J. Wiklund, V. Remuss, L.G. Gutierrez, D.X.
Viegas, M.A. Gonzalez, A. Mallet, R. Alami, R. Chatila, G. Hommel, F.J. Colmenero,
I. Veiga, B. Arrue, J. Ferruz, J. R. Martínez-de Dios, F. Caballero, 2005. Multiple
Eyes in the Skies. IEEE Robotics & Automation Magazine, vol. 12, no. 2, pp. 46-57.
San-Miguel-Ayanz, J., Ravail, N., Kelha, V., A. Ollero, 2005. Active Fire Detection for
Fire Emergency Management: Potential and Limitations for the Operational Use of
Remote Sensing. Natural Hazards vol. 35, no. 3, pp. 361-376.
Remus, V., M. Musial, G. Hommel, 2002. MARVIN -An Autonomous Flying Robot-Based
on Mass Market. Int. Conf. Intell. Robots Systems, IROS, Workshop Aerial Robotics,
pp. 23-28.
Restas, A., 2006. Forest Fire Management Supported by UAV Based Air Reconnaissance
Results of Szendro Fire Department, IEEE Intl. Symposium on Environment
Identities and Mediterranean Area. In press.
Thornton, J., Grace, D., Spillard, C., Konefal, T., Tozer, T.C., 2001. Broadband
communications from a high-altitude platform: the European HeliNet programme,
Electronics & Communication Engineering Journal, vol.13, no.3, pp.138-144.
 V International Conference on Forest Fire Research
D. X. Viegas (Ed.), 2006

Wehr, A, and U Lohr. 1999, Airborne laser scanning—an introduction and overview:
ISPRS Journal of Photogrammetry and Remote Sensing, 54, pp. 68–82.