How we’re putting robots in forests to stop wildfires

The InsightFD1 fire detection robot scans an area for fire

The InsightFD1 fire detection robot scans an area for fire

This is the story of how the company I work for, Insight Robotics, is trying to change the way forest fires are fought, and some of the surprising twists and turns that mission has taken us on along the way. Fair warning, the story involves sensors, robots, drug cartels and a whole lot more.

To a non-specialist, the scale of wildfires can come as a shock. Globally, they burn over 3% of the world’s vegetated landmass every year. Some level of wildfire is an important element in biodiversity, but human activity and climate change have caused rapid and dangerous increases in their frequency and intensity worldwide. In densely populated areas such as Europe, North America and Asia for example, humans are responsible for up to 95% of wildfires. And fire season has gotten almost 20% longer since 1980 due to climate change, which by 2050 is also expected to cause a 50-100% increase in the area burnt by wildfires in countries like the United States.

Biomass fires already emit more carbon than every car on every road in the world, and the smoke they emit kills an estimated 340,000 people every year. An increase in the incidence and severity of wildfires will only push these numbers higher. Fire, in other words, is a big problem that keeps getting bigger.

But what’s just as surprising as the scale of the problem is the means we currently employ to address it. Some elements of firefighting have changed a lot in the past hundred years, with high-tech ‘heavy-metal’ solutions like fire planes and helicopters, along with smaller scale innovations like flame-retardant suits and breathing masks for firefighters. But the way we spot fires in the first place is essentially the same as it’s always been – have someone stand in a tower or on a hill and hope they spot a fire’s smoke plume before it gets too big.

As you can imagine, that doesn’t always work very well. This is both a problem and an opportunity, because catching fires early means fewer emissions, less damage from destroyed property and fewer lives put in danger. It also means less money spent on expensive suppression efforts that put firefighters’ lives on the line. All of that is why the U.S. Congressional Budget Office has estimated that every dollar spent on wildfire mitigation will save five dollars in prevented future losses.

Our goal was to come up with a fire detection system that would empower fire services and make a meaningful difference to the speed and accuracy with which a fire could be detected. Other solutions did exist, like automated smoke detection, but their performance was poor. The conclusion of a particularly damning study by CSIRO, the Australian scientific agency, was that “detection by the camera systems was slower and less reliable than by a trained human observer.”

At first, we tried playing around with a distributed network of thermal sensors scattered throughout a forest and all linked back to a central command center. There was a time when such a solution was thought to be potentially revolutionary, because of the incredibly low price at which you could blanket a huge area of forest. It turned out, however, that deploying 10,000 sensors also meant maintaining 10,000 sensors and making sure that 10,000 sensors were performing without malfunction. As far as wildfire detection went, sensors were out.

Which turned out to be fortunate for us, because we ended up designing a robot instead, and sleek, shiny robots have distinctly more cachet than sensors. The InsightFD1 robot, for Fire Detection version 1, was our attempt to produce the absolute best product that modern technology makes possible. We combined the latest sensors with computer intelligence, meaning that the raw optical power of the system was enhanced by analyzing the data it gathered. In other words, unlike a camera, the robot knew and understood what it was looking at.

The result was the first system in the world capable of detecting a single tree on fire up to 5 kilometers away and pinpoint its location, a vast improvement on technology such as satellites which have only a 50% chance of detecting a fire a 100 square meters in size. The system was also able to detect a fire much sooner than other solutions, because its thermal sensor did not rely on a fire growing large enough to generate a visible smoke plume. In all the 500,000 hectares the system has been deployed to protect so far, it has not failed to detect a single fire that has broken out.

Of course, the problem of wildfire is far from solved. I did promise you that the story would involve drug cartels, and this is where they come in. Fires may seem to be the main obstacle in firefighting, but especially in a global context, there are a lot of other issues to consider. In Bosque Primavera, for example, Mexico’s largest big-city forest reserve and a prime area for wildfire, drug traffickers have set up illegal labs within the reserve, taking advantage of its protected status. The cartels are also involved in illegal logging in the area. Firefighters have been driven off by armed gangs and Environmental Department employees killed while patrolling the reserve.

A robotic wildfire detection system can safeguard the forest from fire without endangering the lives of park personnel. But how do you safeguard the robot itself from being destroyed or stolen? Drug cartels aside, in areas of rural China where the robot is currently deployed, its parts represent a big enough value for local residents to risk trying to dismantle it for resale.

There’s also the question of deploying our technology. We’re confident that the data shows our robot has the potential to save lives, trees and property all over the world, but that world happens to contain over 4 billion hectares of forest. That’s a lot of trees in a lot of countries, and a lot of financial, government and land-owning partners who need to be persuaded, convinced and leveraged.

All of these problems can be overcome, and in many cases we’ve already created solutions to them. But in the complex, surprising, and vital mission to protect our world’s forests, the work continues.

Further reading:
Lloyd’s – Wildfire: A burning issue for insurers?
Fire Management – Global Assessment 2006
Climate-induced variations in global wildfire danger from 1979 to 2013
Flammable Planet: Wildfires and the Social Cost of Carbon
Global Emissions by Source
Effects of biomass burning on climate, accounting for heat and moisture fluxes, black and brown carbon, and cloud absorption effects
Environmental Health Perspectives – Estimated Global Mortality Attributable to Smoke from Landscape Fires
Potential cost savings from the pre-disaster mitigation program
Providing satellite-based early warnings of fires to reduce fire flashovers on South Africa’s transmission lines

Blogpost and picture submitted by Surya Balakrishnan (Insight Robotics Limited, Hong Kong) – surya.balakrishnan(at)
Picture courtesy Bun Mok

The content, structure and grammar is at the discretion of the author only.

This post is published as entry #37 of our #Forests2015 blog competition. It is submitted in the “Open” category.

The first selection of the winners will be based on the number of comments, likes and views each entry gets.

As a reader, you can support this speaker’s entry:

  • Leave a comment on this project in the field at the bottom of this page
  • Support the post by clicking the “Like” button below
  • Spread this post via your social media channels, using the hashtag: #Forests2015

Have a look at the other blog competition entries too!

16 thoughts on “How we’re putting robots in forests to stop wildfires

  1. I believe Insight Robotics role in increasing response speed and reducing control time will render forest fires to manageable levels. Way to go Insight Robotics !!!


    • Thank you, Lizzy.

      Those are indeed the guiding principles of our company and the reason we choose wildfires as our first target area. We wanted to focus cutting edge technology on solving an age old problem.


    • Thanks, Erica.

      Yes, what’s particularly exciting about investing in fire detection is not just that its good for the environment and neighboring communities, but that it actually pays for itself in saved trees and property and lower suppression costs. I don’t think there are too many other environmental technologies that can say the same.


    • Hi Natalie,

      Great question. There is undeniably some truth to the fact that fires are part of many of the world’s ecosystems, but unfortunately climate change and human activity have made them far worse than they would be otherwise.

      In places like Europe and North America people are responsible for upwards of 90% of fires, and climate change could increase their intensity by 50-100% by 2050.

      What’s more, the emissions they create kill over 300,000 people globally every year. Those are consequences that we have to deal with, rather than just live with.


  2. I am one of the newbie in Insight Robotics. Yet, I would like to honestly share my view from my medium level field engineering experience background.

    I will speak from a field experience point of view. The business potentials and technical theory are much better explained by many of my excellent colleagues.

    This product is truly user-friendly and operator-friendly. Firstly, some of the major components are commercially-common technologies, ensuring that the majority of technicians are familiar with some common issues with that technology already even they have not heard of the robot before, thus they can easily gain mastery of the overall maintenance skills with some sense of confidence. The relatively much less no of maintenance sites, as explained in the above passage also, greatly reduces running cost of the system.

    The operate-by-click-and-drag without text command input and the avoidance of complicated maths calculations on the operator’s side allow less senior operators to handle the product easily too.

    Data sources wise, the robot is designed in such a way that important data sources are not clogged together with each other in a special, incomprehensible single formula – meaning that if a source of a particular data is not good enough, new source can be used with partial reprogramming only. This allows flexible adaption to different situations.

    There are always room for further improvement, such as larger detection radius etc. What I appreciate the most is, Insight Robotics engineers adhere to very strict standards including considerations of actual operation concerns even they are moving ahead of a tiny step only. The open discussion atmosphere makes one enjoy the work here a lot!


    • Thanks, Winnie!

      I’d agree with the sentiment that the robot is something to be proud of but, as with any technology, there are always lots of new directions it can be taken in in the future.


  3. This is a huge step forward in helping the environment while also incorporating new and innovative technology.


Leave a comment

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s