Two UWF teams place in top 5 in national artificial intelligence competition
Two University of West Florida student teams showcased their Artificial Intelligence skills on a national stage. The UWF teams each placed in the top 5 of the prestigious AI Tracks at Sea Challenge.
Hosted by the Naval Information Warfare Center Pacific and the Naval Science, Technology, Engineering, and Mathematics Coordination Office, 32 teams from public and private institutions across the country participated in the competition. Various research institutions including those from the Ivy League, Historically Black Colleges and Universities, and Hispanic Serving Institutes competed in the $200,000-prize challenge.
UWF graduate computer science students Tobias Jacob, Raffaele Galliera and Muddasar Ali placed third, winning $35,000 for the UWF Department of Computer Science. The students participated in the competition as members of UWF’s AI and Data Analytics (AIDA) Research Group. Dr. Thomas Reichherzer, chair of computer science, served as the sponsor and Dr. Sikha Bagui, professor in computer science, served as the faculty advisor for the AIDA Research Group. UWF computer science major Zach Mueller, a machine learning intern at Novetta Solutions LLC mentored the students.
“When we found out we finished third we couldn’t believe it,” said Galliera, who, like Ali, is pursuing dual master’s degrees from UWF and Ferrara University in Italy. “That whole morning, we talked about the things we went through during the challenge. We were so happy and so proud of what we accomplished.”
The second UWF team, ArgoTracks, finished fifth and secured $20,000 for the Department of Intelligent Systems & Robotics. The team consisted of Bhavyansh Mishra, a doctoral student in intelligent systems & robotics, and mechanical engineering majors Brendon Ortolano and Luke Fina. Dr. Hakki Erhan Sevil, intelligent systems & robotics assistant professor, served as their sponsor and faculty advisor. The students are members of the Sevil Research Group at UWF. UWF alumnus Carson Wilber, a research associate at Florida Institute for Human & Machine Cognition, mentored the students.
ArgoTracks formed its team a month after the challenge started. Mishra learned about the challenge from Wilber and then contacted Sevil who assisted him in finding teammates. The students put in long hours to catch up and submitted their entry just a few hours before the deadline.
“Computer vision is my bread and butter, so I just hopped onto it as soon as Carson told me,” Mishra said. “Considering the fact we only had one month compared to most teams that had two months, we felt good about where we placed.”
Each team was tasked with developing a computer vision system capable of plotting the tracks of shipping traffic exclusively using the passive sensing capability of a single onboard camera. Current traffic avoidance software relies on an automatic identification system and radar for tracking other craft and avoiding collisions. In a contested environment, emitting radar energy presents a vulnerability to detection by adversaries.
Organizers provided each team a dataset consisting of recorded camera imagery of vessel traffic along with the recorded GPS track of a vessel of interest that was seen in the imagery. Submitted solutions were evaluated against additional camera data correlated to recorded vessel tracks. The same vessel and the same instrumentation were used in both the competition dataset and the judging dataset. Judging criteria was based on track accuracy and overall processing time.
The two UWF teams were among only five that submitted solutions that worked. Ten of the teams submitted solutions that only partially worked or failed to work, and the remaining teams failed to submit solutions.
“It was really difficult because the data they gave us wasn’t preprocessed, the camera wasn’t calibrated and we didn’t have a lot of data,” said Jacob, who is pursuing a master’s degree at UWF after earning his bachelor’s degree from RWTH Aachen University in Germany. “We had many moments where we thought, OK, this isn’t going to work, and then we always found a way to make it work.”
For more information on the AI Tracks at Sea Challenge, visit challenge.gov/challenge/AI-tracks-at-sea/. For more information on UWF’s Department of Computer Science, visit uwf.edu/computerscience. For more information on UWF’s Department of Intelligent Systems and Robotics, visit uwf.edu/isr.