Alumnus researches use of artificial intelligence for humanitarian efforts

For many years, unmanned aerial vehicles (UAVs), aka drones, have been widely used for military operations. 

But a recent computer science graduate took a deep dive into researching another use of drones: search and rescue efforts for humanitarian purposes. 

For his thesis project, computer science alumnus Juan Gonzalo Cárcamo Zuluaga developed a prototype to enable an autonomous drone to learn how to successfully fly, navigate and search for objects without human intervention. He said he was motivated by the devastation caused by Hurricanes Irma and Maria in Puerto Rico.

"The initial search effort after a natural disaster or crisis is critically important because people need to be rescued and cared for immediately," said Zuluaga, who received a master's degree in computer science from Grand Valley in 2018. "A drone provides a bird-eye's view of locations, can map out terrain and delivers supplies." 

Zuluaga said that while UAVs have been used to assist search and rescue teams, his research adds another layer. 

He sought to enhance the performance of drones by incorporating a newly developed machine learning method that would allow a drone to operate without a human operator. Machine learning is a specialized form of artificial intelligence.

"Many drones are manned by a human operator," he said. "I sought to create a program that would allow a drone to operate on its own by using a technique called deep reinforcement learning, which combines an artificial neural network with the way humans learn through reward and punishment." 

Zuluaga developed the prototype using Microsoft AirSim, a virtual drone simulator. He said that while his research is preliminary, he hopes it serves as a first step in the process to create a drone that uses artificial intelligence to operate. 

"It's leading-edge work and we have enough technology to start exploring these kinds of solutions," he said. 

Greg Wolffe, professor of computing and information systems, and Zuluaga presented the research at the Institute of Electrical and Electronics Engineers' National Aerospace and Electronics Conference in July in Dayton, Ohio.

Zuluaga is currently an adjunct professor in the School of Computing and Information Systems and plans to pursue a doctorate in computer science. He came to Grand Valley from Colombia in 2016 as a graduate student in the School of Computing and Information Systems.