Please view my video presentation on YouTube: https://youtu.be/TrFFL2BBoP0
A PDF version of this Final Report is available here – A4 ICT571 JJ
Abstract
With the frequency and ferocity of bushfires forecasted to increase due to a changing climate, there is a pressing need to adopt innovative technologies such as drones and artificial intelligence in preventing these potential disasters. Research shows that drones are primarily being used to detect and monitor bushfires and in order to prevent these occurrences, there are several challenges which should be addressed. Academic research into the prevention of bushfires is limited; however, commercial drone, sensor and artificial intelligence manufactures have the individual technologies and capabilities to create a converged system of bushfire prevention. It is apparent that further research and investment in this field is required before wide scale adoption of these technologies can be deployed.
Introduction
This research project set out to document the effectiveness and maturity of drone technologies in the prevention of bushfires. The findings of this report were obtained using a research methodology which used the following sources: commercial websites, grey literature and peer reviewed journals. This report provides background information regarding bushfires including: how they occur, their negative consequences and future predictions based on climate change. General information about drones, sensors and artificial intelligence is also discussed. This report presents the key findings of the research including: how drones are currently used to detect and monitor bushfires, the opportunities for drones to prevent bushfires, as well as describing several challenges which should be addressed. The report concludes by answering the main research question, can drones and artificial intelligence be used to prevent bushfires.
Research Methodology and Techniques
The research methodology used throughout this project included a detailed analysis of information from several sources including: internet-based, grey literature and peer reviewed academic literature. The diagrams below provide further details regarding the iterative methodology and techniques used to conduct the research.
The research techniques used to gather information included a combination of specific key words as shown in the table below:
Results and Findings
This section of the report provides key results and findings of the research project.
Background Information
Background information has been compiled to support the key findings relating to the use of drones and artificial intelligence to prevent bushfires.
Bushfires
The 2019-20 Australian bushfires burned 5.5 million hectares of land, claimed 26 lives, caused billions of dollars in damage and resulted in severe ecological damage including the loss of more than 1 billion animals.
As documented by the CSIRO (2020), bushfire behaviour is influenced by three factors: weather, terrain and vegetation. The risk of bushfire is highest when a combination of strong winds, high temperatures, low humidity and low rainfall affect an area. These are elements which can be detected through the use of drones and sensors.
With the Australian Government Bureau of Meteorology (2018) predicting fire conditions to worsen in the future due to climate change, short-term and long-term strategies should be adopted to harness innovative technologies (including drones, sensors and artificial intelligence) to combat and potentially prevent bushfires.
Drones
Drones can be used for many consumer, commercial or military purposes and generally fall into three categories: rotor, fixed wing, or hybrid. Some basic features of drones are shown in the table below.
This report considers that these features form the baseline of any drone which could be used in the prevention of bushfires.
Sensors and Cameras
Sensory technologies are essential in the combat against bushfires as these devices collect the data that feed into decision support systems. As shown in the table below, several types of sensors can be mounted on drones to capture data regarding bushfires.
Artificial Intelligence
With drones and sensors capturing large amounts of data, technologies such as big data analytics, machine learning and artificial intelligence could be leveraged to support bushfire prevention. The CSIRO’s digital research arm, Data61 (2020), developed ‘Spark’ which is a “tool for fire prediction and analysis” (Chelvan, 2020). As shown in the figures below, ‘Spark’ can use vectors such as temperature, humidity and elevation to determine potential bushfire spread.
How are Drones used to combat Bushfires?
Research shows that there are five potential ways in which drones can be used to combat bushfires, which are shown in the figure below.
Evidence suggests that the primary applications of drones in bushfire mitigation focus on the detection and monitoring of bushfires. During the 2019-20 Australian bushfires, drones were deployed to identify hot spots using thermal sensing and geolocation technology with the data sent to control centres to aid in operational decision making. Several academic articles have focused on early detection capabilities as well the development of algorithms to actively monitor fire fronts.
Researchers have started to look at drones with fire extinguishing capabilities, however these technologies are largely prototypes and have not been used in bushfire scenarios. Examples include equipping drones with fire retardant balls which can be dropped on to flames, as well as drones fitted with water canons. Drones have also been used to asses damage post-bushfire, locate stranded animals and assist with land re-vegetation.
The Bushfire Earth Observation Taskforce was formed in a wake of the 2019-20 bushfires to consider the role of space-based observation in the mitigation of bushfires. This report considers drone technologies to be better suited to serving “operational support of bushfires in progress” (Australian Space Agency, 2020). Figure 9 highlights some of the detection and monitoring capabilities and limitations of drone technologies.
The general methodology of drone based forest fire monitoring is to use a single or swarm of cooperative drones to fly over high-risk areas of a forest and capture data using on-board sensors and cameras. The captured data would be processed on-board the drone or by a centralised repository to identify a fire or risk of fire. If certain criteria is met, the system would alert relevant parties to begin the remediation process. As shown in Figure 10, Yuan et al. (2017) provides a general concept of using drones (UAV) to detect forest fires.
The NSW Bushfire Inquiry report found that while remote sensing technologies were used before and during the devastating 2019-20 bushfires, they were “not used as extensively as it could have been given technological advances” (NSW Bushfire Inquiry, 2020). Furthermore, the Inquiry recommends harnessing technology advancements in remote sensing, data science, AI and unmanned aircraft in the combat against bushfires.
Can Drones be used to Prevent Bushfires?
Research suggests that it is possible for drones to prevent bushfires; however, the application of the concept is yet to be applied in operation. Several researchers have proposed methodologies and theories about how drones can be used to predict and prevent bushfires. For example, researchers at the University of NSW Digital Grid Futures Institute have developed sensors to monitor changes in the atmosphere which can be attached to drones. As published by Spires (2020), the data collected can be sent to an AI system to create a heat map to predict the potential location of bushfire outbreaks.
In order for bushfire prevention to be possible, research shows that there are three core technology elements required: the drone, specific sensors and an AI platform capable of processing large data sets in a short amount of time. The table below details the key requirements each technology should have in order to prevent bushfires.
Using data collected from sensors, Harkiran et al. (2019) propose a model for predicting forest fires. As shown in Figure 13, the model uses temperature, relative humidity, precipitation and wind speed data to produce a ‘Forest Fire Vulnerability Index’ score.
Using cameras, drones can also be used to assess the state of fire trails, which are essential access lines for firefighters as well as checking power lines for damage which could cause a fire to ignite.
Benefits of Drones over Existing Technologies
The table below provides a summary of the benefits of using drones over existing bushfire mitigation technologies.
While several benefits exist, the research and evidence shows that there are also many challenges which need to be considered and addressed before wide scale adoption can be applied.
What are the key Challenges?
The application of drones in the prevention of bushfires faces several challenges. The table below outlines the top 10 challenges identified through this research project.
Communications and Connectivity Challenges
Ensuring reliable, persistent communications remain available for drones is a key challenge in the pursuit of using drones to prevent bushfires. In NSW, a Public Safety Network (PSN) is used to support radio-based voice communications for emergency services during a crisis; and while investment is being made to expand land coverage from 35% to 85%, this network would not be suitable for drone technologies which largely depend on cellular technology.
Telstra, Australia’s largest mobile provider provides coverage to populated areas; however, coverage is limited in non-populated areas – in particular areas prone to bushfires. Additionally, due to potential fire damage and network congestion during disasters, drone communications over commercial cellular services would be “insufficient to support mission critical operations” (Australian Government Productivity Commission, 2015).
To overcome these communication challenges, the NSW Government, with funding support from the federal government, is building a Public Safety Mobile Broadband (PSMB) communication service which will provide “a data-rich platform for emerging technologies” (Council of Australian Governments, 2018). Importantly, this network would provide drones with the ability to communicate effectively in remote areas and during a crisis without reliance on commercial cellular services.
Conclusion
Most academic and independent research into the use of drones for combating bushfires has focused on their detection and monitoring capabilities. While detection and monitoring are important, the prevention of bushfires using drones, sensors, data analytics and artificial intelligence has the ability to save billions of dollars in damage, save lives and protect against ecological damage.
The findings of this research project show that while the technology components required to prevent bushfires exist, the integration of drones, sensors and AI requires further research and funding. A key recommendation made in the 2019-20 bushfire inquiry report states that a ‘Bushfire Technology Fund’ be created to assist with the rapid development of technologies to sense, fight, mop up after and protect from bushfires. Government funding will help researchers and innovators overcome many of the presented challenges.
As our climate changes and the likelihood and impact of bushfires increases, using drones and artificial intelligence to prevent bushfires is a novel approach to mitigating potential disasters.
References
