How AI, Drones and Virtual Reality Could Help Tackle the Amazon Fires

Originally published in The Telegraph

This month, images of the burning Amazon rainforest have reverberated around the world.

These fires, believed to have been set deliberately by cattle ranchers and loggers, have now spun out of control, leading to unprecedented destruction and dire warnings from environmentalists that the crisis will lead to the loss of a precious ecosystem and an acceleration of climate change.

Brazil has rejected aid and the crisis has been blamed on the country’s president Jair Bolsonaro, who critics say has encouraged farmers and loggers to burn far more of the forest than they typically do to clear land in order to graze cattle.

But elsewhere, technology is helping some communities prevent and tackle wildfires more efficiently, offering hope to communities who are threatened every year by the prospect of their homes and livelihoods being destroyed by an out-of-control fire.

AI and mapping the forest

Prevention is better than cure, and technological advances in computing, which allows companies to buy up cheaper equipment and use processing power through cloud services provided by Amazon Web Services, Microsoft and Google, to name a few, has vastly improved forestry mapping.

“We now have access to data that has never been economically viable to collect. It has changed a lot since ten years ago, when we were sending in firefighters with pens and pencils,” says Max Nova, co-founder of Silviaterra, a satellite imagery company that works with the forestry industry.

San Francisco-based Silviaterra built the first ever high resolution digital inventory of vegetation in the United States which can be used by landowners like timber producers to decide where to harvest wood, conservationists to assess wildlife and governments to regulate deforestation.

Now, topography, vegetation and even weather data can be combined to detect areas that are at most risk. Of course, an entire forest is flammable, but there are certain vectors that could cause a particularly bad outbreak like the type of tree, the coverage density and the climate. Plus, data on species of animals can also be considered, so that foresters and governments can work out where to start controlled fires to clear land if needed.

Big data, computational and satellite tools humans now have at their disposal may help inform policies when it comes to decisions made when deliberately clearing land and or growing trees for materials.

Mr Nova says: “Everything is either mined or grown. All of the items we rely on come from forests - and all of it is part of an economic system.

“So this is partly a story about the environment but largely about economics. Technology can help us look at what our values are, and the most efficient way to realise these values."

For example, policy makers and industries can suggest places for farmers to clear land in areas where endangered species are no longer at risk and look to develop forests in areas that have a high demand for homes, rather than somewhere cramped that is at risk of becoming an uncontrollable fire.

“Having this data helps us put the numbers on what we value,” he adds.

It is only in recent years that pooling this data together and applying complex algorithms that can point out helpful patterns has become possible. “What is exacerbating wildfires is the extreme weather that we are experiencing around the world,” says Bonnie Lei, Microsoft’s AI for Earth project manager.

She works with conservationist groups to which Microsoft donates compute power and resources for environmental good. “It is incredibly important to monitor what is happening over time so we can try and prevent things like fires before they become impossible to stop”.

Speed is of the essence. With cheaper cloud computing first responders and scientists are edging closer to retrieving the data from sensors, satellites and drones quickly enough to analyse it and react in time.

Aid from space

The Silicon Valley company Planet has produced devastating images of the destruction of the Amazon, and says its technology could help evacuation and damage assessments.

A Planet spokesman said: "The combination of medium- and high-resolution satellite imagery provides greater situational awareness and enables many actors to move faster and more effectively in the event of a devastating wildfire, such as those currently burning in the Amazon.”

Another problem when tackling natural disasters in remote areas is the lack of internet access. Firefighters and other emergency services struggle to communicate with each other and confirm that colleagues are safe, and it’s also tough to get data back from sensors in remote locations monitoring things like air quality.

One company trying to bring greater connectivity to these places is Swarm Technologies, a Silicon Valley firm which specialises in tiny satellites. While it doesn’t yet work with any wildfire groups, chief executive Sara Spangelo sees this as a growth area for the business.

The company currently works with development company SweetSense which uses solar powered sensors to monitor water supply in east Africa and is moving to work on drought prevention in California. Similar sensors could potentially be placed in wildfire-prone areas to detect a decline in air quality

“If someone deploys a sensor, we're able to bring back that data from wherever it might be deployed. We can also track, monitor and return messages from people or assets that might be out fighting those fires, offer some of the support network needed in those situations,” she said.

Fighting wildfires is one of the most dangerous jobs in the world. According to the US Fire Administration, 485 firefighters died between 1990 and 2017 while tackling brush, grash or woodland fires.

Drones have been deployed to find firefighters lost in the brush while tackling a conflagration, and they can also be used to find victims trapped by flames and drop lifesaving items to them if immediate rescue is impossible. Heat sensors can identify hotspots, the most active parts of a fire, and keep an eye on the changing fire conditions.

Drones could also prevent fires, by identifying damaged power lines, one of the biggest causes of California wildfires. San Diego Gas and Electric, a local utility company, uses drones to inspect power and gas supply lines, particularly in remote areas which are difficult to access on foot.

Virtual reality fire-fighters

It may sound far-fetched, but firefighters may benefit from using virtual reality headsets to train for a face off with a wildfire. Leicestershire Fire and Rescue Service (LFRS) was the first in the UK to trial the technology, using the HTC Vive VR headset and using realistic, game-like software to teach recruits in a cheaper, safer way.

During training, firefighters wear the helmet and protective clothing and operate a hose in response to an immersive visual that is displayed in the virtual reality headset, lending the sensation of being in front of a fire.

LFRS wants to integrate touch, surround sound and even replicate the smells of burning to create a truly immersive experience. It might not replace the skills learned from facing real flames, but it can help with fear and give supervisors a chance to intervene during training to point out areas for improvement.

The technology is also being used to train smokejumpers in the United States, who leap from a plane to face wildfires head on, by putting them through a virtual parachute jump.

Protecting animals

Immediate victims of the Amazon fires, aside from our fragile ecosystem, are the animals that call the forest home. Conservationists are using technology like cameras and microphones to monitor wildlife that may be affected by wildfires in a non-invasive way.

“This is the sort of technology that when deployed at scale helps scientists make decisions about extreme weather events and gaining a better understanding of animals and areas that are most at risk,” Lei says.

The Elephant Listening Network can detect species using infrasound, eavesdropping on noise inaudible to humans from large mammals to bats.

It can decipher chainsaws or gunshots, indicating a poacher is nearby. It is not too far of a stretch to see how this could be applied in areas where the sounds associated with burning and falling trees could be used as a warning system, although it is not being used at the moment.

The aftermath

In November 2018, northern California was reeling from the aftermath of the most destructive wildfire in the state’s history.

The Camp Fire had displaced thousands from their homes, almost 100 people were killed or missing, and residents were desperate to know when they could return and what was left of their communities and possessions.

Local authorities called in Dr Gregory Crutsinger, a drone data consultant who usually works in agriculture mapping, to help them with the aftermath.

“You program the drone to take lots of photos, and then you stitch those photos into maps, and then get those map layers to the agencies,” he says.

Dr Gregory Crutsinger working at the site of the Camp Fire last year Credit: Casey Tholborn/Contra Costa Sheriffs Department

16 drone teams, led by the local sheriff's office, operated aircraft to take pictures of the damaged areas.

A runner drove back to San Francisco with the a hard drive containing the images and took it to the software company DroneDeploy, where a team worked through the night to stitch together the images to create a functional map.

“We mapped half a terabyte of information and we had it turned around in 48 hours, and publicly available the day after,” he says.

The maps let residents see what had happened to their homes, as well as helping insurance companies assess the damage using super high-resolution images, without having to send people into wildfire-stricken areas which were still smoke-filled and treacherous.

California is a state prone to wildfires, but Silicon Valley’s proximity to these northern California fires has helped it deploy cutting-edge technology. Now Crutsinger gets calls from all over the world from people looking for help, most recently from South America.

This drone technology, says Crutsinger, has developed alongside smartphones and has been “available and really easy off the shelf for the past four to five years”.

In the US, local sheriff’s offices tend to use drones made by DJI, a Chinese manufacturer that dominates 80pc of the market.

Drones aren't always the best option - if the area to be mapped is very large, satellites are more suitable.

"I think [it's] the right tool for the right job," he says. "Drones are are easy. They're easy to buy, compared to bringing in helicopters, and unmanned aircraft and things."