Panos Pardalos, Ph.D., Receives Funding for Research in Navigation Approaches for Autonomous Vehicles

Panos Pardalos, Ph.D., Distinguished Professor of Industrial & Systems Engineering, in the Department of Industrial & Systems Engineering, received funding from the U.S. Air Force Research Laboratory Munitions Directorate. The funding supports Dr. Pardalos’ research to develop deep learning navigation applications with synthetic aperture radar (SAR) image data used by unmanned aerial vehicles.  

An unmanned aerial vehicle (UAV) is an aircraft that is guided either autonomously or by remote control. The military uses UAVs to carry sensors and other electronic transmitters that can detect enemy targets. These UAVs rely on an embedded global positioning system (GPS) that provides position, navigation, and timing information. However, when in a GPS-denied environment, UAVs need to rely on other mechanisms.  

“Conventionally, UAVs or airplanes rely on information from GPS dominantly when acquiring information on geographic coordinates for navigation. Our research focuses on developing a deep learning-based system for image recognition that enables navigation in a region where GPS is not available. We also utilize the SAR images that can be acquired, regardless of the weather conditions, as data for training the deep learning system,” said Dr. Pardalos. “To this end, we need to build the deep learning system that enables real-time calculation as well as accurate SAR image recognition.” 

This research will impact the future use of autonomous agents in GPS-denied environments. Not only will networks be capable of real-time inference and navigation using SAR image analysis but will also be more accurate than the technology that is currently available. This is possible by applying super-resolution concepts that have been shown to improve the quality of SAR images, which can then be processed to improve navigation methods. 

“It is critical to fly when GPS is not available, such as when jamming devices interfere with GPS signals for navigation. Image-based navigation powered by a deep learning system is crucial to cope with these situations. We hope our efforts can contribute to systems that make UAVs or airplanes functional in hostile environments.” said Dr. Pardalos 

Dr. Pardalos, and Maciej Rysz, Ph.D., Assistant Professor at Miami University, will utilize the UF Research Engineering Education Facility’s high-performance computing lab to conduct their research.