Call for Code runner-up: Project Lali predicts and detects wildfires with sensor networks
IoT, cloud services and low-cost sensors provide real-time ground data to prevent wildfires.
Growing up in Ecuador, developer Kevin Cando says wildfires were so common that he got used to walking to his college classes as ash rained down.
This experience, and the desire to stop wildfires before they begin, planted the seed for Lali Wildfire Detection System, which came in 3rd in the 2018 Call for Code Global Challenge.
“In Ecuador, most of the time, we do have wildfires and, because of price constraints, they prefer to let the wildfires burn,” said Cando, who has lived in the United States for 24 years. “We believe our solution can help prevent millions of dollars of loss in the United States, but also save millions of lives in South America and Ecuador.”
Watch the replay of the Call for Code Global Award Celebration
Wildfires often sweep through brush and woodland in the most rural areas, presenting a real challenge to firefighters looking to understand and anticipate wildfire movements. At the Silicon Valley Call for Code hackathon, team Lali, whose members include Cando, Nassim Bettach, and Jay Nagdeo, realized that Internet of Things (IoT) sensor technology could be used to gather information from sparsely populated areas, and help firefighters accurately monitor the spread of a fire and plan how to contain it.
How it works
The technology itself has progressed to a level where small devices can record temperature, light, and other environmental factors that are triggered during a fire. Then they transmit this information using low-power networks like SigFox. All data is collected using the IBM Watson IoT Platform, on which visualizations and data prediction models can be built, showing first responders which communities are most in danger.
“What we do is basically measure temperature data, send it to the IBM Cloud, and take it in the IBM Watson platform for further processing — like machine learning and AI algorithms to predict fire intensity, the fire shape, and where it’s going to spread,” Nagdeo said.
Other data, such as wind direction from Weather APIs and land topography maps, can help refine the models. Team Lali was able to build a working demo of the IoT system in less than a day, giving them confidence in the potential of soon testing a live prototype with a fire department.
“What we saw with the IBM Cloud platform is we can take lots of data and mix it together to make something really more efficient,” Bettach said.
The sensors are currently less than $5, have a range of 10-15 km and a battery that can last up to 10 years, which makes this solution viable for developing nations that are often the hardest hit by wildfires.
The data pipeline from the low-cost sensors feeds into IBM IoT, providing live updates to Node-RED-powered maps to visualize all sensor locations with the relevant data. These temperature maps can help measure the pattern a fire will follow, allowing response teams to create a more effective containment plan.
“We could potentially save millions of lives with a five dollar device,” Cando said. “Our overall goal is to have this deployed in as many countries we can.”
Project Lali was awarded USD $25,000 and will also receive long-term open source support from The Linux Foundation.
Daryl Pereira, Kevin Allen, and Liz Klipp contributed reporting to this article.