In recent times, the devastating impact of wildfires, exacerbated by the effects of climate change, has been felt across various regions, from the idyllic shores of Maui to the picturesque landscapes of the Mediterranean.
These catastrophic events have not only claimed numerous lives but have also pushed firefighters to their limits, necessitating the search for innovative solutions.
This is where artificial intelligence (AI) comes into play.
By harnessing the power of AI, firefighters and emerging startups have been able to employ AI-enabled cameras to scan the horizon, tirelessly searching for any indications of smoke.
Furthermore, a German company is in the process of constructing a network of satellites, equipped with AI technology, to detect fires from space.
Even tech giant Microsoft has joined the fray, utilizing AI models to predict the potential locations of future blazes.
As wildfires continue to grow in size and intensity due to the global rise in temperatures, the urgency to stay one step ahead of these flames has compelled firefighters, utility companies, and governments to tap into the latest AI advancements.
The potential of AI to revolutionize our lives has evoked a mixture of fear and excitement, but for the increasingly overwhelmed first responders, it offers a glimmer of hope.
Nevertheless, it is important to note that human intervention is still indispensable to verify the accuracy of these technological advancements.
California’s primary firefighting agency has recently embarked on a groundbreaking initiative by implementing an artificial intelligence (AI) system to detect smoke from over 1,000 mountaintop camera feeds.
This innovative technology has proven to be highly effective in identifying potential fire outbreaks and enabling rapid response measures.
Encouraged by its initial success, the agency has decided to expand the implementation of this AI system statewide, marking a significant step towards enhancing California’s firefighting capabilities.
By leveraging the power of AI, the state aims to bolster its early warning systems, minimize response times, and ultimately mitigate the devastating impact of wildfires on both human lives and the environment.
This pioneering approach exemplifies California’s commitment to harnessing cutting-edge technology to protect its citizens and preserve its natural resources.
The system in question has been intricately designed to identify any potential “abnormalities” and promptly notify emergency command centers.
Subsequently, the center’s staff will then verify whether the detected anomaly is indeed smoke or something else present in the air.
Phillip SeLegue, the staff chief of intelligence for the California Department of Forestry and Fire Protection, emphasized the convenience of this system, explaining that the alert immediately appears on the screen, enabling dispatchers or call takers to analyze the situation and make an informed decision regarding dispatching a crew.
This innovative technology incorporates a network of cameras that previously necessitated constant monitoring by personnel.
The cameras provide an enormous amount of data for the AI system to process. Although human confirmation is still required for any smoke sightings, the system effectively mitigates fatigue among staff members who typically monitor multiple screens and cameras.
It alerts them only when there is a possibility of fire or smoke, thereby optimizing their efficiency. The effectiveness of this system has already been demonstrated, as a battalion chief received a smoke alert during the night.
He confirmed the alert on his cellphone and promptly contacted a command center in San Diego, resulting in the immediate mobilization of first responders to the remote area.
According to SeLegue, the dispatchers emphasized that had they not been alerted, the fire would have grown significantly larger as it would have gone unnoticed until the following morning.
This highlights the crucial role of early detection in preventing disasters. Pano AI, a San Francisco startup, adopts a similar approach by installing cameras on cell towers.
These cameras employ computer vision machine learning, a form of artificial intelligence, to scan for smoke and promptly notify relevant stakeholders, including fire departments, utility companies, and ski resorts.
CEO Sonia Kastner explains that the cameras are specifically trained to distinguish between smoke and non-smoke through exposure to images of both.
Additionally, the cameras utilize data from government weather satellites that identify hotspots, as well as information from various sources such as social media posts.
This comprehensive approach aims to enhance the effectiveness of fire prevention and emergency response.
The emergence of new technology has addressed a significant challenge in the conventional method of detecting wildfires, which heavily relies on 911 calls from individuals seeking confirmation from personnel before dispatching firefighting crews and water-dropping planes.
According to Kastner, CEO of Pano AI, only one out of every twenty 911 calls actually turns out to be a wildfire, as many instances involve clouds, fog, or even barbecues.
However, Pano AI’s systems still require final confirmation, with managers reviewing a time-lapse of the camera feed to ensure that the observed activity is indeed smoke rising.
Larry Bekkedahl, Senior Vice President of Energy Delivery at Portland General Electric, Oregon’s largest utility and a Pano AI customer, emphasizes that technology is becoming increasingly vital in combating forest fires.
This sentiment is particularly relevant as utility companies sometimes inadvertently contribute to wildfires when their power lines are toppled by strong winds or struck by falling trees.
In fact, Hawaii’s electric utility recently admitted that their power lines sparked a devastating blaze in Maui during the summer, likely due to high winds causing them to collapse.
Pacific Gas and Electric (PGE), the provider of electricity to 51 cities in Oregon, has recently implemented a state-of-the-art solution to enhance their emergency response capabilities.
They have deployed a total of 26 Pano AI cameras, which have proven to be instrumental in expediting response times and improving coordination with emergency services.
Prior to the implementation of these cameras, fire departments would often find themselves frantically searching for the source of fires without a clear understanding of their exact location.
However, the introduction of the AI cameras has revolutionized this process by enabling the detection of fires at an accelerated pace, allowing teams to be dispatched promptly.
As a result, response times have been reduced by up to two hours, a significant improvement considering the rapid spread and growth of fires.
Juan Lavista Ferres, the chief data scientist at Microsoft, emphasized the importance of having a sufficient number of cameras to cover vast and remote areas, such as those in northern Canada that have experienced devastating wildfires this summer.
While the utilization of AI to detect smoke from fires is relatively straightforward, the challenge lies in ensuring comprehensive coverage across all regions.
Ferres’ team at Microsoft has been diligently working on the development of AI models aimed at predicting the locations where fires are most likely to occur.
Through the utilization of historical maps of previously burned areas, as well as climate and geospatial data, the model has been trained to identify patterns and trends.
It is important to note, however, that the system does have its limitations and cannot predict random events such as lightning strikes.
Nevertheless, by analyzing historical weather and climate data, the model can identify areas that tend to be drier, and even consider factors such as the presence of roads, which indicates human activity and poses an additional risk.
While it may not be infallible, Ferres emphasizes that the model can generate a probability map based on past occurrences.
Microsoft plans to make this technology available as an open-source tool, which can greatly assist first responders in determining where to allocate their limited resources.
On a similar note, German startup OroraTech has taken a different approach by harnessing the power of satellite images and artificial intelligence.
Leveraging the advancements in camera, satellite, and AI technologies, OroraTech has successfully launched two mini-satellites into low orbit, approximately 550 kilometers above the Earth’s surface.
With plans to launch eight more satellites next year and eventually deploy a total of 100, the Munich-based company aims to provide comprehensive coverage.
During the recent wildfires in central Chile, OroraTech was able to offer thermal images during the night when the use of aerial drones is less frequent.
Shortly after the launch of its second satellite, the company successfully detected a fire near the community of Keg River in northern Alberta, where the flames had repeatedly ravaged remote stretches of boreal forest throughout the summer.
CEO Thomas Gruebler explains that the satellite is equipped with highly efficient algorithms that enable the rapid detection of fires.
Additionally, the AI incorporated in the system takes into consideration factors such as vegetation and humidity levels to identify potential flare-ups that could escalate into devastating megafires.
By providing real-time information on the location and potential propagation of fires, this technology can greatly assist firefighting agencies in allocating their resources more effectively and preventing further damage.
Gruebler emphasizes that their system allows them to accurately predict which fires will grow into significant threats and which will eventually extinguish on their own.