Using robots to detect forest fires

Rex Sham, co-founder and chief science officer of Insight Robotics, has already co-authored two patents and won awards for his research. Photos by Calvin Sit.

Rex Sham and his WALL-E-inspired robots are improving fire detection to slash carbon emissions and make the world better.

By Stephen Corby

If he were in a movie, Rex Sham would be the bad guy, or at the very least the well-meaning scientist who unwittingly wipes out the human race. In reality, Sham, the co-founder and chief science officer of Insight Robotics, is using his ingenious, WALL-E-like fire-detecting robots to save the planet.

What Sham has developed is an automated early warning system that combines a high-precision, pan-tilt robot with thermal imaging sensors and advanced artificial intelligence (AI) vision technology. In its tests of the Computer Vision Wildfire Detection System, the Guangdong Academy of Forestry (Insight Robotics’ research partner since 2010) has recorded a 100 per cent detection rate in multiple field trials and deployments.

At 32, Sham has already co-authored two patents, won awards for his research, and seen his fire-protection systems being used by 41 forestry departments all over China, as well as in his native Hong Kong, private plantations in Indonesia, and in Malaysia and Brazil. His systems now monitor 1.5 million hectares of forest.

“Forest fires are a very big issue all over the world. According to NASA [National Aeronautics and Space Administration], the smoke from those fires contributes about 30 per cent of the global carbon emissions every year, making it one of the largest causes of global warming,” Sham says.

As he explains, detecting the very first spark of a fire, and putting it out quickly, would prevent those carbon emissions being released and make an enormous difference.

“Quite simply, if you want to reduce carbon emissions, this is one of the best ways to do it.”

Professional Development: CPA Q&A. Access a handpicked selection of resources each month and complete a short monthly assessment to earn CPD hours. Exclusively available to CPA Australia members.

In the past, fire detection was carried out by humans standing on platforms and scanning vast vistas with their puny human eyes. Another method used CCTV, which saw different humans staring at walls of screens all day. 

“There was a lot of late detection with that system, because of mental fatigue from the job being so boring,” says Sham.

Satellites, while useful for tracking the movement of big fires, are no use for early detection; you need a fire to be at least 900sq m in size before a space-based camera can pick it up. That’s too late, because you need to detect a new fire within 15 minutes of the first spark to have a chance of containing it, Sham explains.

“Our robots can detect fires at a distance of 13 kilometres, when the fire is as little as six metres wide, which is basically just one tree on fire,” he says. 

“We’ve created the world’s first automated detection system that can detect a single tree on fire in 200 square kilometres of forest, and it’s fully automatic, requiring no human intervention, and runs 24 hours a day.  

“It’s also not affected by weather, because its thermal cameras can see through things like mist that a human can’t.

“I feel really good about it, because we’ve found something that can help people. Over the past two years alone, we’ve detected more than 2000 fires in the Chinese region.”

A business to help the world

Gore at he premiere of An Incovenient Sequel at the 13th Zurich Film Festival on 8 October 2017.

Growing up as an only child in Hong Kong, Sham spent a lot of time playing with toys – robotic ones in particular – and became obsessed with a cartoon called Doraemon, about a robotic cat.

“When I was like five or six, I thought that when I grew up, there would be a Doraemon that could be bought off the shelf. I’m still waiting,” Sham says. 

“Finally, when I got access to the internet, I discovered that real robots can’t even walk upstairs, and fall over a lot. I made myself a promise that I’d go to university and study robotics so I could make my own Doraemon. That’s how it started.”

After entering several robotics competitions as a representative of the Chinese University of Hong Kong, where he studied (“I spent a lot of time making things that didn’t perform as I’d designed them to”), he graduated and immediately set up his own company, EduTechnic, in 2008, making educational kits to teach other students how to make their own robots.

“Learning how to build a robot through online research, or passive learning, is very difficult. What you really need to do is participate, use your hands and DIY. I started a company to help people do this,” he says. 

“It went very well for two years, and then we hit a financial tsunami, where schools no longer had the money for that kind of education. That closed down the company very quickly, in 2010.”

Fortunately, Sham had also been working on developing robots to work in disaster zones, and for earthquake rescue in particular. It was all charitable work, but he did receive some donations from Kevin Chan, the man who is now CEO of Insight Robotics.

Unfortunately, funding research through charitable works was unsustainable.

“We needed to find a way to earn our living and sustain the research, so we pulled out the eye of the robot and used the software algorithm for detecting body temperature, in earthquake rescue, and fine-tuned it to be able to detect fires that are far away,” Sham explains.

“In early 2010, we collaborated with the Guangdong Academy of Forestry in having some fire trials and collected raw data about early stage wildfires, then we wrote a new software algorithm for detecting wildfires. 

“After setting up Insight Robotics in 2011, we made a better robot prototype. After three years of development, we launched our FD1 robot in late 2013.”

Insight Robotics gained seed funding of US$2 million in 2011 from investors including Bright Success Capital, Caldera Pacific Capital Partners and Radiant Venture Capital. In 2014, it raised another US$2 million in Series A funding to invest in research and development, talent acquisition and international growth.

"Most jobs that are repeatable, that are mechanical, will be replaced by robots."

The business is profitable, but making money is not the core strategy of Insight Robotics. As Sham puts it, if he had just wanted to be rich he would never have entered the robotics business, he’d be writing software programs for stock speculation instead.

“The set-up of our company is to use whatever technology we can develop, and integrate it into something that can do good for the world, while generating revenue and profit at the same time. I expect myself and the other engineers of my company to believe that we can save the world, ultimately.”

However, that’s not the message he uses to sell his systems.

“We start from a business angle,” says Sham. “We don’t say to companies, ‘please pay us $1 million because it will help reduce the world’s carbon emissions’, because people won’t pay for that. But if we talk about how much in assets they’re losing to forest fires, and how soon they won’t be able to buy insurance, and tell them that this system will help bring those premiums down, then they’re interested.

“That’s actually what happened in Indonesia and the Asian region; people were no longer able to get insurance, because the insurance companies were not willing to insure any forest land … but our robots are helping with that now, too.

“We help the government agencies and the forest managers to solve their problems of losing forestry assets, but at the same time we are reducing the carbon emissions for the world in a big way.”

Inspired by WALL-E

Insight Robotics’ machines, which have one hugely powerful zoom-lens eye and another with a thermal camera on their “faces”, look a lot like kid-favourite WALL-E, and that’s no coincidence. Sham is a big fan of the movie, “so we mimicked the look of that quite a bit”.

While it would have been theoretically possible to have his robots zooming over the treetops in a sci-fi fashion, Sham says it wasn’t practical because most countries have strict laws about drones and other flying objects being operated without line of sight.

“Effectively, we would have had to get flying licences for the robots, and they also would have struggled in high winds, which is actually when you need them the most in forest fires,” he explains.

“Humans have been working on top of watch towers for thousands of years – it’s a very traditional business. But it’s getting harder to get young people to do it, because it’s cold and remote, and there’s no internet access for them to look at Facebook.

“We wanted to build something to replace humans having to work in this harsh environment, so the easiest way was to mimic what a human can do, but to do it with super-human capabilities. We designed our robots to be installed on these stationary platforms, because people have already built hundreds of thousands of them, so we don’t need any new infrastructure.

“One human can now sit in an air-conditioned room and watch over more than 100 robots that are scanning an area, all from an app on their cell phone.”

Winning over robot-phobes

Sham is very aware that people are suspicious of robots and fear them taking human jobs. Movies such as the Terminator series paint the technology he works with as sinister.

Not surprisingly, he prefers to point to friendlier but no less clever robots, such as Star Wars’ R2-D2 and, one of his personal favourites, the Iron Man suit.

“I think Iron Man is trying to change the way that people think about robots,” he enthuses. “Robots should have an assisting role, like in Iron Man, rather than being replacements for humans, like a Terminator.

“Robots should always be seen as an assistant, to help us to work more efficiently and to take the roles in dangerous areas, doing dangerous jobs that humans shouldn’t be doing.”

In strategy terms, Sham says it’s easy to identify areas where robots should be, and will be, replacing humans. Essentially, they should be able to liberate us from repetitive, non-value-added jobs.

"I will not be scared that my articifically intelligent microwave will one day kill me."

As he explains, if you do the same thing 1000 times and the problem is still there – like picking up rubbish, or sweeping a street – you’d be better off coming up with a new kind of machine that would use less energy to do the same role.

While he doesn’t deny that robots are going to take a lot of human jobs, he predicts that largely they will be the ones no one wanted to do anyway.

“There are areas where robots can’t help. A machine can’t be a nurse – I wouldn’t want a robot to give me drugs when I’m sick, and nor would I want to be looked after by one when I have given 60 years of my life to society, because I would lose dignity. It would be like using a machine to feed chickens,” he says, with passion. 

“But a robot can detect a forest fire. It can pick up garbage.

“Most jobs that are repeatable, that are mechanical, will be replaced by robots. Take putting fruit into a plastic bag: we don’t need to waste the so-precious human brain on how to put fruit into a bag – we can use a machine to do that. Instead, you can think about how to make food that tastes better, or you can make a cake that is more beautiful. 

"I think robots can help us to have more time to enjoy the more human-like aspects of life.”

Don’t fear the microwave

A large part of Sham’s research, and what makes his robots so impressive, involves the use of AI. While commentators such as Stephen Hawking and Elon Musk (who has compared AI to “summoning the demon”) fear machines may become self-aware and decide to wipe out weak, mentally simplistic humans, Sham has no such concerns.

“We do have some kinds of artificial intelligence in our robots, but it’s a very layman type of AI,” he says. 

“You look at chat bots, which can now talk to you on the internet and seem human, but it’s not really ‘intelligent’; it’s just mimicking the way a human would respond. It’s using logic and statistical recordings of the way real humans would respond; it’s just a playback.

“I‘m not worried about it, but I will worry about it if AI can reach a level where it can feel what humans feel, when it can understand feelings of happiness or sadness or hate. But I really don’t think it will reach that status within the next 100 years.

“I will not be scared that my artificially intelligent microwave oven will one day kill me, because it thinks it’s more intelligent than me, because it can cook better than me. I will embrace the technology if it can help me to cook anything I want, and cook it better than me.

“I’m not concerned about robots being better than me; not until they can understand what a soul is. That would be scary.”

Sham, however, is not someone to be afraid of. He’s more the kind of scientist we should all be grateful is out there working to make our world a better place.

Rex Sham was a keynote speaker at CPA Congress 2017, in Hong Kong.

Read next: Here's why robots won't take over the economy anytime soon

Like what you're reading? Enter your email to receive the fortnightly INTHEBLACK e-newsletter.

Like what you're reading? Enter your email to receive the fortnightly INTHEBLACK e-newsletter.

December 2017
December 2017

Read the December issue

Each month we select the must-reads from the current issue of INTHEBLACK. Read more now.