Scientists are aware of the fact that icebergs breaking from the Greenland ice sheet are melting fast in response to climate change, but the specifics of how quickly they are melting are still unclear. Hanumant Singh is a professor of Electrical and Computer Engineering at Northeastern University. He believes that the problem is figuring out how to measure large pieces of ice which are in constant motion.
Singh factored into a kayak-sized robot’s algorithms to render sharp 3D maps of icebergs in Sermilik Fjord, a 50-mile inlet into the shoreline of Eastern Greenland. Accurate measurements of the shape and surface of an iceberg need to be compared over time in order to get a concrete idea of the rate of melting. However, mapping an object in constant motion is a robot’s Achilles heel. Engineers who work on autonomous drones often come up against this challenge, since robots need to scan and map their environments to move effectively and autonomously. It is usually calculated with the assumption that things won’t move. A robot’s measurements of a moving target can be easily distorted and difficult to correct. Singh has made it his goal to get around that challenge with calculations that account for the movement of an iceberg simultaneously as the robot navigates around it and maps it.
In a recent paper, Singh and a team of researchers describe the technology and algorithms they used to correct for the movement of a dozen icebergs floating in Sermilik Fjord. The algorithm’s 3D reconstructions of icebergs show high-resolution details of the geometry and relief of the ice, which is sometimes impossible to capture even with raw images snapped by an ocean drone. Compared to real life ice, the accuracy of those models is remarkably close. This robotic system is just a camera and a sonar sensor mounted on a commercially available, gas-powered kayak. Singh calls it a break from tradition since ocean robots are really expensive. It takes raw images of the part of the icebergs exposed above the water and then uses those pictures to navigate strategically around the iceberg, helping the sonar sensor take high-quality measurements of the submerged ice, which accounts for more than 90 percent of the ice structure.
For decades, researchers have struggled to get up close and personal with icebergs. These ice formations can be several times bigger than a large parking garage and have protruding ice above and under the water that can be dangerous if the iceberg capsizes or breaks.
Shahjadi Jemim Rahman