Logo: University of Southern California

Diving In

USC Viterbi Ph.D. student Stephanie Kemna conducts research on autonomous underwater vehicles that help biologists
By: Natalia Velez
August 31, 2015 —

Kemna's research has the potential to significantly increase the level of autonomy and coordination in underwater vehicles.   (Illustration by Michelle Henry)
Mapping the ocean’s environment is a top priority when it comes to addressing environmental hazards. However, characterizing the physical and chemical gradients in these ecosystems is an ongoing challenge for marine biologists.

In hopes of providing a new and better approach, USC Viterbi Ph.D student Stephanie Kemna conducts research on underwater vehicles that can help biologists gather data.

“I am very interested in working in something that can help improve the world from a computer-science perspective,”  said Kemna, who landed an internship at Amazon Robotics.

Currently, Kemna is a part of the Robotic Embedded Systems Lab, whose robots can sample for indicators such as presence of algae in the water column, temperature, dissolved oxygen and pH. However, the question is not about what can be sampled, but about finding algorithms that will allow robots to organize themselves and make autonomous decisions.

“Stephanie's research has the potential to significantly increase the level of autonomy and coordination in underwater vehicles, making it easier for non-roboticists to deploy and use them,” said Gaurav Sukhatme, Kemna’s advisor and Chair of the Computer Science Department.

Stephanie Kemna, recent Ph.D. graduated from  USC Viterbi.   (Photo: Luke Fisher Photography)

Kemna is particularly interested in coordination between autonomous robots. She would like to build a system in which they can work together to optimize mapping efforts. This requires more complex planning than single robots, as it involves high levels of specialization and division of labor.

“I want to be able to coordinate a team of robots,” Kemna said. “It would be ideal if they could just be released into the ocean and know what to sample for, and where to go.”

In the future, Kemna hopes to incorporate ideas from information theory into the brains of the robots, so that they can work in teams towards accomplishing a common goal, and reorganize themselves when facing an obstacle.

“I’ve been thinking about the future of my work, and I want to help biologists gather the best data possible. I want my research to be useful in both biology and computer science,” Kemna said. “My plan is to incorporate some sort of knowledge into these machines to accomplish this.”

Kemna’s interest in aquatic robotics came about serendipitously. Originally from the Netherlands, she obtained her undergraduate and master’s degrees in artificial intelligence from the University of Groningen, and was initially drawn towards rehabilitation robotics.

“When I finished my master’s thesis I wasn’t interested in doing a Ph.D. so I started looking for jobs,” Kemna said. “After attending a trainee program with the Dutch Navy and TNO, I ended up working at NATO, Centre for Maritime Experimentation.”

The job at NATO’s Center for Maritime Research and Experimentation involved a lot of experiments with underwater vehicles. After three years there, Kemna was hooked, and decided to pursue a Ph.D. to conduct her own research.

“I wanted to focus my research on the coordination between multiple vehicles, and USC has one of the few labs in the U.S that does this kind of research,” Kemna said. “When you’re looking to choose a Ph.D., it’s all about who you will be working with and what collaborations the department has, and for me, the decision of coming here was a combination of those.”