A group of Stanford researchers have harnessed the use of autonomous drone technology and machine learning approach with an aim of learning and gathering crucial data to study Antarctica's ice sheets to increase our understanding of the processes that drive sea-level rise.

Since the problem is modern, solutions too need to be the same - which means we need a more efficient and intelligent data collection system. For the same, the researchers have approached the problem with a two-pronged solution - create a new data-collecting platform with the help of autonomous drones, aka unmanned aerial vehicle (UAV), equipped with ice-penetrating radar to acquire more accurate readings; and identifying the key areas where valuable data can be extracted to maximise the impact of their research.

“We want to equip policymakers with information to decide how to adapt, but given the difficulty of gathering data from Antarctica, we can’t survey everything,” says Dustin Schroeder, one of the researchers working on the project. “We need to focus on collecting the most impactful data. The question of where that data is — or how we would know in a formal way — is a hard, technical, AI-rich problem,” he explains further. Schroeder has been working on the project along with PhD candidate in electrical engineering Thomas Teisberg, and their collaborator Mykel Kochenderfer, an associate professor of aeronautics and astronautics, both from Stanford. 

The research will gather data on the depth of ice beds, the actual temperature of ice, how the melting rate is affected by tides, seasons, and the passage of time from an area stretching over 5.5 million square miles!

Researchers are focusing on data-driven strategy with an aim to develop a strategy that will enable them to simultaneously assess the observations of the ice bed, the 3D flow of the ice, and the principles that govern its movement. The models process each fresh batch of data in real-time to determine a constantly changing flight plan.

Primary work has already begun as the team has started exploring ice shelves on the Machine Learning side of the project, which have exciting dynamics on a smaller scale than the polar ice sheet. If the models can learn and forecast how quick changes have significant effects on ice shelves, it will be a substantial step toward addressing the broader problem of ice sheets.

 While the current goal aims to make ice sheet research more intelligent and efficient, in the future, this approach can potentially change the way all glaciologists collect and interpret data in the long run.