**1. Introduction**

Forest fires are one of the most common and, at the same time, serious emergencies facing humanity. They threaten not only natural areas, where they cause important losses of plant and animal diversity, but also urban areas, where they can cause dramatic human and material losses. Furthermore, forest fires cause significant emissions of greenhouse gases and consequently are contributing to global warming. For these and other reasons, states develop policies for fire prevention, early detection, and rapid intervention.

Quantifying the fires and their consequences along the world is not a trivial task. According to the World Fire Statistics [1], a report published by the International Association of Fire and Rescue Services that collects data from multiple governments, there were 4.5 million fires and 30,800 deaths in countries with 2700 million inhabitants in 2018, which means 1.7 fires per 1000 and 1.1 deaths per 100,000 inhabitants that year. Although these figures do not take into account the whole world, they allow us to quantify the magnitude of the problem. The information provided by several space agencies supports this thesis: the European Space Agency (ESA) publishes the World Fire Atlas with the information collected by ATSR-2 [2] and Sentinel-3 [3]), whereas the National Aeronautics and Space Administration (NASA) does the same with the Global Fire Atlas [4].

Current forest firefighting missions consider prevention, surveillance, and extinguishing tasks. The first ones seek to prevent the occurrence of fires and limit their consequences, the second ones look for detecting fires early, and the third ones search to put them out quickly and safely. Firefighters reveal the lack of human and material means and the degraded information of the scenario as the main problems in these tasks. They routinely use

**Citation:** Roldán-Gómez, J.J.; González-Gironda, E.; Barrientos, A. A Survey on Robotic Technologies for Forest Firefighting: Applying Drone Swarms to Improve Firefighters' Efficiency and Safety. *Appl. Sci.* **2021**, *11*, 363. https://doi.org/10.3390/ app11010363

Received: 29 November 2020 Accepted: 25 December 2020 Published: 1 January 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional clai-ms in published maps and institutio-nal affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

multiple types of vehicles and machinery to improve the performance and safety of these operations. However, the use of robots and especially drones is not common, although these autonomous systems could solve some of the current challenges.

This paper aims at analyzing the current problems in forest firefighting missions and the potential of robotic technologies to solve them. Therefore, we pose the following two research questions:


For this purpose, the paper analyzes the data provided by governments, the results of two original surveys on firefighters, and the literature on robotics applied to forest firefighting.

Finally, the paper proposes a concept of operation for the application of drone swarms to fire prevention, surveillance, and extinguishing tasks.

The remainder of the paper is organized as follows: Section 2 addresses the current situation of firefighting, analyzing the public statistics provided by multiple countries and presenting the results of our surveys and interviews to professionals. Section 3 collects various works developed in the context of academy and industry that apply robots to firefighting tasks. The concept of operations using drone swarms to support firefighters in all these tasks is presented in Section 4. Finally, the main conclusions of the work are summarized in Section 5.

#### **2. Firefighting State**

This section analyzes the current state of firefighting. For this purpose, it describes current operations of fire prevention, surveillance, and extinguishing, collects relevant statistics to identify main problems, and presents the opinions of professionals through two surveys. Note that most of the information presented in this section is from Spain, but can be generalized to at least European and Mediterranean countries.

In 2019, 10,883 fires burned 83,963 ha in Spain: 7290 of these fires affected less than 1 ha, whereas 3593 affected more than 1 ha [5]. As reported, these figures were similar to the previous years, having an average of 12,182 fires and 99,082 ha per year between 2009 and 2018. In other words, every year, 0.356% of the forest surface of Spain suffers the consequences of fires.

Most of these fires occur in spring and summer, especially in March, July, August and September, whereas the worse consequences occur in July and August when more surface burns than the rest of the year [6]. In the case of summer, this behavior can be explained by the high temperatures, which favor the appearance of fires and their spread throughout the territory. In the case of March, most of these fires occur in the north-west of the country and are caused by an accidental, negligent, or intentional use of fire. However, firefighting is performed throughout the year, since it involves not only extinguishing fires but also preventing them.

Fire prevention involves a set of activities that seek to reduce the probability of fire occurrence, as well as to limit their effects if they occur [7]. More than half of forest fires in Spain between 2006 and 2015 were caused intentionally, whereas 28% were caused by accidents or negligent behaviors, 7% had natural origins, and 12% still have unknown causes. Therefore, there are two main groups of activities: prevention on causes and prevention on combustibles. The first category groups those activities that seek to reduce the risks and usually present a social character, such as the awareness campaigns to avoid the use of fire in the primary sector and negligent behaviors in natural environments. The second one covers the actions performed on land uses and vegetation distribution, which seek to generate discontinuities to prevent the expansion of potential fires.

Fire surveillance involves the activities performed to detect fires as early as possible. The damages caused by forest fires highly depend on detection and response times. The information of Spain in 2019 is clear: the average burned surface when response time was shorter than 1 h was 7.10 ha, whereas the one when response time was longer than 1 h was 30.66 ha [5]. For this reason, minimizing detection and response times is key for firefighting. Currently, detection times are addressed with a watchmen network distributed throughout the land and, to a lesser extent, ground and aerial mobile surveillance. In Spain, 60% of fires are detected thanks to citizen collaboration, 27% by static watchmen, 1.6% by mobile watchmen, and 0.5% by aerial means [6]. Meanwhile, response times are addressed by the effective coordination of the teams and the use of helicopters to deploy firefighters in the affected area.

Fire extinguishing involves not only the actions performed to put out the flames but also some activities that support these actions, such as creating firewalls, routes for entry and exit of vehicles, runways, heliports, etc. In Spain between 2006 and 2015, these activities involved the participation of humans (100% of fires), ground vehicles and machinery (94.8%), and aerial means (23.5%) [6]. Extinguishing operations are dangerous because any accident can cause injuries or even deaths among the professionals. The government of Spain reports 24 accidents between 2006 and 2015 with 37 deaths, including only firefighting professionals [6]. According to this study, the causes of these deaths were air accidents (43%), entrapments (30%), medical problems (8%), falls (8%), accidents with vehicles (5%), and accidents with machinery (5%). Therefore, it would be good if technological solutions could reduce both accident rate and mortality in the cases of entrapments and falls, which can be caused by the lack of information about fire evolution and terrain features.

We performed a set of surveys and interviews with firefighting professionals to check and broaden this information. The surveys allowed us to involve a high number of professionals and distinguish collective consensus from individual opinions. Meanwhile, the interviews were done before and after the surveys: the first ones allowed us to prepare the questions, whereas the second ones provided us with more details about the answers. These activities aimed to collect information about current problems of firefighting and opinions about potential technological solutions.

Two surveys were carried out with firefighting professionals: one focused on their problems at work (see Section 2.1),and another on their opinion about multiple technologies (see Section 2.2). Both surveys had between 10 and 20 questions and required fewer than 5 min to maximize the answer ratio. The separation of problem and technology surveys prevented the influence of the questions of one on the responses of the other.

The dissemination of the surveys sought to reach professionals who perform all the firefighting roles in most of the regions of Spain. For this purpose, we sent the surveys by email to fire stations and firefighter unions, as well as share them in firefighting groups on various social networks. In this way, we avoided getting a sample biased towards a specific firefighting role or geographic area.

#### *2.1. Problem Survey*

Our first survey was focused on the problems on current forest firefighting missions. We performed this survey to obtain more information about the first research question. Although the official data previously analyzed are useful to answer this question, the opinions of the professionals involved in these activities are also relevant.

In this survey, we pose the following questions:


• Problems in extinguishing tasks: This question seeks to find the most relevant problems in current extinguishing activities, according to the opinions of firefighters.

We had the support of several firefighting professionals in the writing of the questions and their possible answers. In this way, we could check that our surveys were sound and easy to understand by our target public. Furthermore, we sent the questionnaires to a sample of twenty professionals before their dissemination to check if they could understand them adequately. Finally, we allowed open answers to some questions and shared our contact data to receive doubts.

This survey was sent via email to fire stations, unions, and associations, as well as shared with firefighters' communities on several social networks. A total of 140 professionals from different regions of Spain took part in that survey in three weeks (note that this survey is still open to new responses (Forest fires in Spain: Problem survey (https://forms.gle/e4327HBxqqWVMUbY7) [in Spanish])). A summary of the results is shown in Figure 1.

**Figure 1.** Results of the problem survey about forest fires in Spain.

The survey sample is representative of the professionals involved in forest firefighting in Spain, given that it includes not only the main roles (forest firefighters, technicians, firefighters, forest agents, and volunteers) but also other less common roles (pilots, army forces, researchers, support staff, meteorologists...), in comparison to the official reports [5]. The vast majority of them take part in extinguishing tasks (94%), whereas three quarters have experience in prevention and a half in surveillance tasks. Besides these fundamental tasks, around one-third of the respondents have carried out the coordination of operations (31%) and management of human and material resources (38%). Regarding the means used to perform these tasks, three quarters used ground vehicles and fire trucks, sixty percent aerial vehicles (this includes pilots and airborne firefighters), and fewer tractors (17%) and drones (11%).

Spanish firefighters often say that "summer fires should be extinguished in winter", remarking on the importance of prevention activities in firefighting. In this sense, the professionals surveyed assign very close scores to all the prevention tasks. Mainly, they give a slightly higher score to the preparation of vegetation (4.4 in a scale from 1 to 5), and prioritize awareness campaigns against malpractices (e.g., barbecue, smoke or throwing glass bottles in the bush) over campaigns against the use of fire in agricultural and livestock activities (4.3 vs. 4.1, respectively). There are more differences between the main problems faced in prevention tasks. They highlight the lack of means for preparing the vegetation (4.2), together with the difficulty to find perpetrators (3.8) and convict them (3.9).

Regarding surveillance and detection tasks, the participants highlight static watchmen (4.5), mobile watchmen (4.1), and citizen collaboration (4). Note that this evaluation does not coincide with the actual situation, given that sixty percent of fires are detected by citizen collaboration, whereas only twenty-six percent are detected by static watchmen and less than two percent by mobile watchmen. Aerial means are considered less important for fire surveillance and detection: planes and helicopters receive 3.1 points and drones 2.9 points. According to their opinions, the main problem in these tasks is the lack of human resources (3.9), followed by the lack of material means (3.7), and the lack of risk information, which would allow reinforcing surveillance in areas with a higher risk of fire.

Finally, the professionals surveyed consider health risks and the need for real-time information as the main problems in extinguishing tasks with 4.5 and 4.4 points, respectively. Both problems are closely related, considering that most accidents are caused by the lack of information about the fire evolution, such as entrapments and falls. Other relevant problems are the lack of human and material resources (3.9 and 3.8, respectively), and the difficulties to coordinate the teams on the ground (4).

#### *2.2. Technology Survey*

Our second survey was focused on some technologies that can contribute to solving the reported problems. In this case, we performed this survey to obtain more information about the second research question. The objective was to collect opinions from professionals to estimate their predisposition to use these technologies.

For this purpose, the survey included the target technologies of this study (drone swarms and immersive interfaces), together with some control technologies. These technologies were chosen after a review of research and the commercial literature and served as a reference in the evaluation of target technologies. Incentive systems


of artificial intelligence to predict the risk of fire, which allows performing this task over specific areas.

• Extinguishing: The survey asks about the application of drones to monitor the evolution of fires. In addition, it considers three alternatives to receive the information during field operations: an immersive interface, a mobile device, and a voice assistant. In this way, the target technology can be compared to two common methods to receive information.

We took the same measures as in the previous survey to ensure that questions and possible answers were understandable.

The participants of the first survey who gave their emails were invited to fill the second survey. In this case, a total of 70 professionals submitted their responses in the first three weeks (again, the survey is still open to new responses (Forest fires in Spain: Technology survey (https://forms.gle/XV4ScxL9jyCgr4fJ7) [in Spanish])). A summary of the results is shown in Figure 2.
