Robot Engineer Hanu Singh Talks Glaciers and Robots
Hanumant Singh builds robots that ride kayaks, examine the undersides of polar ice floes, and bump up against Greenland glaciers where they dump into icy fjords. His robots go where humans can't. An engineer at Woods Hole Oceanographic Institution, on Cape Cod, Singh is a tenured associate scientist with a Ph.D. from MIT, and he can tell you about the time that his nine-year-old daughter knew the international news about his robot, when he was on a very limited phone connection from Antarctica. He designs and builds robots for use in marine biology, marine chemistry, and marine archaeology. Among many other projects, he is working on how to make a robot more curious, so that it can curate the information it finds.
1. When you get past the whiz-bang electronic fun of it all, why do marine expeditions use robots? If you look at access in the ocean, it's hard to come by. It's expensive--you need a big oceanographic ship. The farther you go, the more it costs. Even if National Science Foundation pays for the expedition, there are only so many days of work available on the ship, and the time is spread over all the biologists and workers, etc. It's impossible to see very deeply into the ocean without being right there: No WiFi or satellites can penetrate it. If you want to see 10 meters below the water, you have to be there.
In another alternative to using a ship, people can take stationary-point measurements with, say, a buoy. But you want spatial measurements too. You want to be able to see change and different areas. In terms of robots, we can move one into, say, Antarctica using a helicopter. Or we can put it in some other location and leave it, then collect it later.
2. Tell me about one of your robot expeditions: In 2013 or so we designed a thing called a JetYak, a robot on a motorized kayak. It was for a study conducted by Fiamma Straneo, a physical oceanographer, and Sarah Das, a glaciologist. The robot's purpose was to go to the very edge of a huge active glacier as it entered a Greenland fjord. Because this is an area where massive chunks fall off the glacier into the water once or twice a day, it's much better for a robot to go there than to risk human life and limb. In such a case, we make a calculated risk to construct the robot as inexpensively as possible -- the robot itself costs less than the sensors that gather the information -- and we are happy if we get the robot back safely. In this case, the robot was collecting information to help the two scientists learn more about why glaciers are melting as fast as they are. The robot gathered some really great information.
3. What would you say about your work in making a robot more discerning? I'm working as a postdoc on that. Basically, all robots do stuff that is boring and time-consuming. It's like mowing the lawn: going back and forth, back and forth, etc. Scientists are usually looking for a beautiful little flower in the middle while mowing a lawn. We don't know if we'll find it. The problem is that if you are a robot you'll possibly miss that flower. Generally we're trying to study a dynamic--whether it's fish schooling or whatever that is interesting. The hope is that if you see something, you can quickly react and map it while the mapping is good, rather than go back later after mowing a large part of the ocean and find that thing gone. We're working on increasing the quality of how well a robot can match up the images in a succession of images: discerning what has been seen so far, and are there anomalies.
4. What are other plans marine scientists have for robots? Discoveries have solely been made in the summers of the polar regions. We want to know what takes place in the winter in Antarctica and other areas. The winter is so harsh and unaccomodating--but we'd like to build a robot that can take measurements in the winter. Right now we can leave single instruments that can make measurements all winter. We want to be able to develop large area maps over time, over winter.
To see the story of the Jetyak robot, which was used to go near Greenland glaciers, click here for Oceanus magazine from Woods Hole Institute.
For related Nature World News updates, please see:
Follow Catherine on Twitter at @treeswhales