Frank Frisby of CECOM is on the AI Avengers team. Not pictured are CECOM's Maluki Montgomery and DEVCOM’s Roy Trieu. (Photo Credit: U.S. Army) VIEW ORIGINAL

The U.S. Army Communications-Electronics Command has its own Avengers, but our local team is focused on national security in a different way with far fewer aliens. The CECOM Software Engineering Center Artificial Intelligence Avengers are the 1st place champions in the U.S. Army’s 2023 Deep Green Data Science and Artificial Intelligence Challenge Phase II. Our AI Avenger team consists of CECOM SEC data scientists Frank Frisby and Maluki Montgomery, and DEVCOM’s Roy Trieu.

The challenge was separated into two phases with the overall focus being the perception capabilities of ground autonomous navigation systems that utilize Computer Vision, Light Detection and Ranging, and AI algorithms in off-road environments. The Phase II portion of the challenge focused on “building a Semantic Segmentation model using raw images, point cloud data, or a fusion of both datasets,” according to the competition description. Further, “the goal of Semantic Segmentation is to classify every point/pixel on an image or LiDAR scan as a specific class.”

The AI Avengers focused on image pixels over LiDAR in order to tackle this phase of the challenge. The competition took place over a period of 6 months and the team worked on their entry in addition to their regular duties.

"We were swamped with competing priorities at the time and nearly gave up," Frisby said. "But in the last month before the competition ended, we realized we could actually pull this off. The game-changer was knowing that our work could actually make a real impact, helping warfighters in the field. That turned our time crunch into a deep desire to push even harder."

The team’s model will be used as a baseline performance prototype for Army autonomous vehicle development.

"I participated in the challenge to learn more about AI while helping to address a real-world use case for the Army," Montgomery said. "AI is a rapidly emerging technology that will transform how software is used to support our warfighters, and I wanted to do my small part to help accelerate the adoption."

The Deep Green event began in 2021 and is designed to provide two main benefits. The competition aims to solve some of the most difficult challenges the Army’s analytic and research communities face and promotes skill development in applying AI and machine learning in their technology development efforts across the Army.

Deep Green also broadened its scope for Phase II by opening to students from service academies and select military colleges. This supports strengthening the connection with academia while harnessing the intellectual diversity and power of a wider range of eligible participants.

The Army develops Deep Green and other similar competitions and utilizes them as opportunities to look ahead in support of Army readiness efforts and adapt to the rapidly changing operational environment.

Congratulations to our local Avengers for excelling in the important fight to keep our Nation safe from emerging threats!