Reducing Wildfires with IoT, ML, and Drones

1 month ago
Eric Vanderburg

The wildfire that ripped through California in early October caused tremendous damage and the loss of over 30 lives.  So far this year 8.5 million acres have been burned, and last year 4.8 million acres were destroyed due to wildfires.  The loss of life, property, and our valuable forests is staggering.  However, at the recent Dell IQT Day, solutions were discussed for a variety of modern challenges, and I believe IoT can aid in reducing wildfires.

According to the National Park Service, a bureau of the Department of the Interior, only ten percent of wildfires are due to natural causes such as lightning or lave.  The remaining ninety percent of US wildfires are caused by humans through unattended campfires, cigarettes, trash burning, and arson.  These fires start small but can spread quickly, especially in dry or windy conditions.  The key to stopping fires is to stop them while they are still small.  This requires intelligence on the area and conditions and effective response.

Data collection with IoT

The first step in addressing such a problem is in gathering data.  IoT sensors can be deployed across national forests and parks.  Biodegradable sensors would be best so that their expiration would not litter our national parks.  These sensors would gather information on heat, air flows, air composition, smoke particles, or other fumes.

Forest systems could be divided into manageable zones containing IoT sensors that were scattered through the area.  Each zone would be managed by an IoT gateway.  Posts could be spaced periodically throughout regions, each containing a wind turbine or solar collectors to generate power for the IoT gateway and a small drone.  The gateways could be connected through a wireless mesh extending throughout the region.  Each gateway would collect data from the sensors in its area, isolate indicators of a fire and report these along with other relevant data back to distributed cores at fire operations centers.

Fire operation centers would be more distributed than the geographic area coordination centers.  However, data from systems could be provided to coordination centers as well as the National Interagency Coordination Center (NICC) for forecasting, deep learning, and coordination.

Analysis with Machine Learning (ML)

A large number of sensors deployed through the forests would allow systems to identify anomalies.  False positives can be identified based on readings from other sensors in the zone.  This vast network of sensors also makes anomalies stand out.  Machine learning models could be created to better identify real fires, and these would continue to improve as more sensors are deployed, more systems integrated, and more data collected.

Immediate response through drones

A local drone could be immediately dispatched to retrieve footage of the conditions.  If a fire was confirmed, the drone could apply extinguishing agents to the fire.  This would occur within minutes of detection since the drone is located within the IoT zone.  Larger drones could be dispatched from nearby locations if additional extinguishing agents were required.

Drones have typically not been used in fire operations because wildland air traffic is not aware of them nor their flight plans.  However, official drones could all be tracked so that firefighting traffic control would know their positions and avoid collisions with other aerial firefighting craft.

Cleanup

The last stage would be to report the incident to firefighters, police, or park services staff so that the fire suppression chemicals could be cleaned up and evidence collected.  Drone footage could be helpful in identifying the people who cause fires so that corrective action can be taken.  This could further help in reducing wildfires.

This use case of IoT, ML, and drones shows the potential these technologies have in solving major problems like wildfires.  These technologies can also be leveraged to solve unique business problems.  IoT can improve decision making by making new data available to organizations.  When a large amount of IoT data is analyzed by machine learning, new insights and operational capabilities can be discovered.  IoT systems can also improve problem detection and resolution through enhanced monitoring of real-world items by digital systems.

This post was sponsored by Dell, but the opinions are my own and don’t necessarily represent Dell Technologies’ positions or strategies.