Using mobile phone sensors and drones for a zero-carbon future
Amelia Holcomb wanted more than a great job at a large tech company. Now the grad student is researching how technology can measure forest carbon
Amelia Holcomb wanted more than a great job at a large tech company. Now the grad student is researching how technology can measure forest carbonBy Adriann Kennedy University Relations
Amelia Holcomb was working at one of the biggest tech firms in the United States after graduating from Yale University when she started to grow restless. “I found myself asking why I was doing what I was doing. If I succeed … how does that change the world?”
Holcomb realized she wanted to use her background in math and computer science to tackle climate change. Today, she’s a graduate student in the University of Waterloo’s David R. Cheriton School of Computer Science. “I got really excited about Waterloo, specifically the thesis program and the ability to direct my own studies,” Holcomb says. “Not a lot of schools offer a research thesis option. It's a bit of a non-traditional path within computer science.”
The question led Holcomb to the forest inventory process, and the much larger challenge of calculating how much carbon a forest sequesters. The traditional inventory process is expensive, time consuming and labour intensive with people using tape measures, calipers and clipboards to measure trees.
To Holcomb, it begged the question: How could technology make this process more efficient and more accurate?
“I wanted to see if we could collect the data automatically and also get more data,” she explains. “Through the manual process you're getting a couple of data points per tree, and then a human still has to enter that data and you're not going to get a three-dimensional picture of that tree.”
Her idea was to use the depth sensors on mobile phones to create a collection of small data points with X, Y and Z coordinates, to a create a three-dimensional structure called a point cloud. It’s the same sensor technology that some phones use to take photos in portrait mode, adding that dreamy blur to the background of your selfie. The sensors send infrared waves that bounce off of an object — like a tree — and then measure the time it takes for the light to return to produce a depth image.
Unlike traditional computer vision methods, which can require many pictures to reconstruct a three-dimensional object, Holcomb has found that the single depth image produced by these sensors can be paired with a regular RGB image to assess the diameter of a tree.
Collecting data from one tree is just the start. The next challenge is a full forest inventory, which requires understanding everything from how many trees are in a plot to how far apart they are. She sees a role for technology there too.
“We are wondering if we can use this technology to do the work aerially — maybe by flying a drone over a forest — because a phone is still limited to someone walking through the forest,” Holcomb says. “We're still thinking about new, creative ways to collect this data more quickly, more efficiently and more accurately.”
Holcomb’s research is being supervised remotely by Srinivasan Keshav, a former Waterloo professor who joined the University of Cambridge in the U.K. in 2019. He connected Holcomb to the members of Cambridge Zero, a group of researchers and policy experts working to build a zero-carbon future.
For Holcomb, this kind of collaboration is the key to successfully combating climate change. She’s working on ideas to get more computer scientists involved in climate work. And she’s thinking about how to bridge the gap that exists between technology and policy to maximize impact.
“Sometimes there is not enough knowledge about what technology can do. There's a really important role that we as computer scientists need to play in proselytizing technology to policy makers and saying, ‘What do you need, and do you know what we could do for you?’ It’s one of the things that I think about a lot.”